Re: Long load time of managed object context
Hi Matteo, How do you load your model? Do you pass [NSBundle allBundles] to NSManagedObjectModel's mergedModelFromBundles: ? Maybe your application bundle contains more than one model (e.g. from linked frameworks) which can confuse CoreData (i.e. slow down or generate errors). - Diederik On 16 jan 2009, at 13:03, Matteo Manferdini wrote: Hi everyone, I'm testing performance in my application, to see how it behaves with a big load of data inside of it. My application is core data based and uses array controllers in entity mode to fetch data. What I discovered is that my app fetched all the managed objects present in core data, so I set the fetch predicate of array controllers at startup. This reduced dramatically the number of objects fetched. The problem though is that my app still takes a lot of time to start when is full of data. I ran the app within instruments, and the core data fetches instrument confirms that this is not the problem (my app fetches only 20-30 objects when starts). But running it under Shark I can see that for 4-5 seconds after startup the only call on the stack is to the app managed object context. So I think it has something to do with interaction between managed object context and data stored, even if I don't fetch all the data. If I empty the app or leave little data inside of it, it starts very quickly. Does somebody know why this happens? Am I missing something obvious? Why does it take so much to the managed object context to load if I only fetch few objects? I always read that core data scales well and the programming guide states that 10.000 objects are not much for core data, but in my app it makes a big difference, so I'm wondering where I am wrong. Thank you very much. Cheers. Matteo Manferdini Pawn Software www.pawn-soft.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/mailings%40obviousmatter.com This email sent to maili...@obviousmatter.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: NSNumberFormatter setFormat: in 10.3.9
Yes, I saw that. Maybe I misunderstood it. I thought that if the app was running under 10.3.9 the formatter would always be created as 10.0 format since the 10.4 format stuff didn't exist in the 10.3 frameworks. Isn't that right? Jon I believe that in 10.5, the default formatter behavior is 10.4 instead of 10.0 (in 10.4 the default behavior was still 10.0). Check out: http://developer.apple.com/DOCUMENTATION/Cocoa/Conceptual/DataFormatting/DataFormatting.html Neil On Jan 23, 2009, at 12:30 PM, Jonathan Fewtrell wrote: My app, developed on 10.5 but targeted at 10.3.9 and above, contains the following code: NSNumberFormatter *formatter = [[NSNumberFormatter alloc] init]; [formatter setFormat:@#0;#0;#0]; It compiles without warnings. When running on 10.3.9 it crashes with EXC_BAD_ACCESS in - [NSNumberFormatter setFormat:]. It works fine on 10.4 and 10.5. For such a simple bit of code, I struggle to see what could be wrong. A previous version, developed on 10.4, did not show this problem, so I'm wondering if it's a build setting problem. If it is, I can't see which one. I can see from the archives that NSNumberFormatter has various issues, but they do not seem to me to be relevant here. Any insight appreciated. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Knowing when a NSArrayController is ready
On 24 Jan 2009, at 02:43, David LeBer wrote: This may be a stupid question, but I've had reason to wonder a number of times. If I have an NSArrayController in a nib, bound to a MOC, configured with an Entity name and set to prepare content: Is there a way to be notified when it has finished loading and arranging it's objects? Hi David Adding an observer on the controller's arrangedObjects property should do the trick. [arrayController addObserver:self forKeyPath:@arrangedObjects options:0 context:(void *)ControllerContentContext]; In a persistent document based app, for example, the controller's MO content is not yet available in - (void)windowControllerDidLoadNib: (NSWindowController *)windowController. Adding the observer means you can set up any dependent code when the binding/core data machinery has finished doing its thing. ;david -- David LeBer Codeferous Software 'co-def-er-ous' adj. Literally 'code-bearing' site: http://codeferous.com blog: http://davidleber.net profile:http://www.linkedin.com/in/davidleber twitter:http://twitter.com/rebeld -- Toronto Area Cocoa / WebObjects developers group: http://tacow.org ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/jonathan%40mugginsoft.com This email sent to jonat...@mugginsoft.com Jonathan Mitchell Central Conscious Unit http://www.mugginsoft.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Poof! Recently-saved Core Data document suddenly gets dirty
On Jan 23, 2009, at 22:44, Jerry Krinock wrote: I realize I could probably paper over this problem by disabling undo registration during the import The stack trace clearly shows that an undo action is being recorded, which means that some property has been changed. (Changed in the sense that KVO-notifications were produced, which is what the undo mechanism notices, AFAIK. The actual values of the properties are likely irrelevant.) Core Data is fanatically correct about recording changes as undo actions, and it's really, really difficult to find out what it thinks has changed, sometimes. It's possible that the delayed-action effect on dirtiness is a deferred processPendingChanges happening. It's possible that coalescing of changes by processPendingChanges might be suppressing change notifications. So, disabling undo at times when you don't want any undo isn't really papering over a problem. It's more of a defense against Core-Data- induced madness. Or, given your heroic investigations, you could almost justify submitting a bug report and hoping Apple will debug the rest of the problem for you. Or, you could turn Core Data undo off completely, and handle undo yourself. Getting undo for free in Core Data is a benefit only if it causes less work for you, not more. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: NSNumberFormatter setFormat: in 10.3.9
On Sat, Jan 24, 2009 at 3:49 AM, Jonathan Fewtrell jonathanfewtr...@mac.com wrote: Yes, I saw that. Maybe I misunderstood it. I thought that if the app was running under 10.3.9 the formatter would always be created as 10.0 format since the 10.4 format stuff didn't exist in the 10.3 frameworks. Isn't that right? Yes, that's correct. Your format string seems to be correct as well. Are you sure you haven't called a 10.4-only method against the number formatter before calling -setFormat: ? ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Targetting Tiger
Hi, I'm in a sticky situation. I personally have no machine capable of running tiger, but my customer needs me to provide a tiger version of my app. I attempted to create a build that targeted tiger, but they are reporting that it is failing. What's going wrong: The application crashes almost immediately (on tiger, but not leopard) while loading its nib, here's the back trace: Thread: 0 Exception: EXC_BAD_ACCESS (0x0001) Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x Thread 0 Crashed: 0 libobjc.A.dylib 0x90a594c7 objc_msgSend + 23 1 com.apple.AppKit0x932639d8 loadNib + 303 2 com.apple.AppKit 0x932633d9 +[NSBundle(NSNibLoading) _loadNibFile:nameTable:withZone:ownerBundle:] + 918 3 com.apple.AppKit 0x9326303a +[NSBundle(NSNibLoading) loadNibFile:externalNameTable:withZone:] + 164 4 com.apple.AppKit 0x93262f7c +[NSBundle(NSNibLoading) loadNibNamed:owner:] + 392 5 com.apple.AppKit0x93262cc3 NSApplicationMain + 433 6 com.davie.klondike 0x237a _start + 216 7 com.davie.klondike 0x22a1 start + 41 Thread 0 crashed with X86 Thread State (32-bit): eax: 0x ebx: 0x927db6ae ecx: 0x90ac0010 edx: 0x0041 edi: 0x0008 esi: 0xbaf8 ebp: 0xbb88 esp: 0xbac4 ss: 0x001f efl: 0x00010202 eip: 0x90a594c7 cs: 0x0017 ds: 0x001f es: 0x001f fs: 0x gs: 0x0037 What I've done to try and get it to run on tiger: Set the following build settings in my target: ARCHS = $(ARCHS_STANDARD_32_BIT) SDKROOT = macosx10.4 MACOSX_DEPLOYMENT_TARGET = 10.4 GCC_ENABLE_OBJC_GC = unsupported In the xib file, set the target to 10.4 (I've also tried saving a version 2.0 nib instead of a xib). I'm building using Xcode 3.1.2. If anyone could point me to what setting I've missed, or some documentation I can read on this, I'd be really grateful. Thanks Bob ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: NSNumberFormatter setFormat: in 10.3.9
On 24 Jan 2009, at 09:27, Kyle Sluder wrote:
On Sat, Jan 24, 2009 at 3:49 AM, Jonathan Fewtrell
jonathanfewtr...@mac.com wrote:
Yes, I saw that. Maybe I misunderstood it. I thought that if the
app was
running under 10.3.9 the formatter would always be created as 10.0
format
since the 10.4 format stuff didn't exist in the 10.3 frameworks.
Isn't that
right?
Yes, that's correct. Your format string seems to be correct as well.
Are you sure you haven't called a 10.4-only method against the number
formatter before calling -setFormat: ?
Yes. The only possibly relevant code I have is in the class's
+initialize method:
+ (void)initialize
{
if ( [NSNumberFormatter
instancesRespondToSelector:@selector(setFormatterBehavior:)] ) {
NSLog(@They do respond);
[NSNumberFormatter setDefaultFormatterBehavior:1000];
}
}
I put this in to force 10.0 behaviour when running on 10.5. The log
message is not called on 10.3.9, confirming that the
+setDefaultFormatterBehavior method is not called under 10.3.9.
Anyway, the problem still exists even if I remove the +initialize
method.
I've also tried different format strings. And retyping the whole thing
to make sure no invisible characters crept in. Makes no difference.
Bafflement reigns.
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Forcing allocation of a subclass
Hi,
I need to be able to force the requested allocation of a cocoa class
to always return an instance of my subclass. I have looked into ways
of doing that.
I have a solution that seems to work, by using a category on the class
to replace the regular invocation, but from the archives I understand
that is a Bad Idea that can lead to all manner of unpredictable and
hard to trace errors. The current implementation looks like this:
@implementation PDFDocument (PDFDocument_Alloc)
+ (id)alloc
{
if ([[self class] isEqual:[PDFDocument class]]) {
return [ANPDFDocument alloc];
} else {
return [super alloc];
}
}
@end
As I wrote, this appears to work fine. But if indeed this technique is
better avoided, what would be the best alternative? I have looked at
class_replaceMethod, but am having difficulties implementing it. What
would be the best way to implement a safe alternative to the method
listed above?
-António
---
Some things have to be believed to be seen.
--Ralph Hodgson
---
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
I would recommend against the category approach. As your search of the
archives no doubt explained, there's no guarantee that PDFDocument may
at some point internally use its own category that would override yours.
I don't know exactly how you'd do what you want (+poseAsClass: is
deprecated) with the runtime functions, but may I advise that whatever
your eventually solution you replace +allocWithZone: rather than +alloc
Mike.
On 24 Jan 2009, at 09:19, Antonio Nunes wrote:
Hi,
I need to be able to force the requested allocation of a cocoa class
to always return an instance of my subclass. I have looked into ways
of doing that.
I have a solution that seems to work, by using a category on the
class to replace the regular invocation, but from the archives I
understand that is a Bad Idea that can lead to all manner of
unpredictable and hard to trace errors. The current implementation
looks like this:
@implementation PDFDocument (PDFDocument_Alloc)
+ (id)alloc
{
if ([[self class] isEqual:[PDFDocument class]]) {
return [ANPDFDocument alloc];
} else {
return [super alloc];
}
}
@end
As I wrote, this appears to work fine. But if indeed this technique
is better avoided, what would be the best alternative? I have looked
at class_replaceMethod, but am having difficulties implementing it.
What would be the best way to implement a safe alternative to the
method listed above?
-António
---
Some things have to be believed to be seen.
--Ralph Hodgson
---
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/cocoadev%40mikeabdullah.net
This email sent to cocoa...@mikeabdullah.net
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Using the security framework
On Fri, Jan 23, 2009 at 10:59 PM, Joe Turner joetur...@me.com wrote: Okay, so, it seems everyone was right :) I went to cocoabuilder to find some of the responses to this, that I never got. Anyways, it seems I can just call AEWP() once, and it should stay suid. So, my last question to everyone is, how do I know if the tool is suid. I mean, I could implement some sort of lock-unlock thing, and then I would be pretty sure, but is there anyway to know for sure? And, is there a way to take SUID away from it? So, if I want to lock it, it would work. man 2 stat to check the bit, man 2 chmod to set it. Mike ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Turn autosaving on forever - is it OK?
On Fri, Jan 23, 2009 at 9:41 PM, Vitaly Ovchinnikov vitaly.ovchinni...@gmail.com wrote: Michael, cocoa's embedded autosaving saves to separate file and leave the main document dirty. It helps only in case of power lost or software hang. But as it runs in background, some conflicts are possible. That's what I wanted to know. Ah, I didn't realize that. Well, one of my three potential reasons goes away then, but the other two remain. I don't know how important they are, but they're at least plausible Mike ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Opening Invisible Files
I want my NSOpenPanel to show all files - same as the Finder does, if one adds AppleShowAllFiles = YES to: ~/Library/Preferences/ com.apple.finder.plist. Is there a way to accomplish this? Kind regards, Gerriet. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Opening Symbolic Links
I have a text file, called Text. I did in Terminal: ln -s Text SymlinkToText When I try to open SymlinkToText in TextEdit SymlinkToText is shown in the Open Panel, but when I select it, the Open-button is disabled. Is this a bug or a feature (in 10.5.6) ? I believe this did work in Tiger, but have no machine to test this. Kind regards, Gerriet. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Using the security framework
Cool, thanks! Then I now have another question: Why not just run chmod on my utility when it's 'unlocked', and change the userID to 0. Then when it's locked, change it back to 501? Or, is this exactly what AEWP() will do? Thanks! Cheers, Joe Turner On Jan 24, 2009, at 6:47 AM, Michael Ash wrote: On Fri, Jan 23, 2009 at 10:59 PM, Joe Turner joetur...@me.com wrote: Okay, so, it seems everyone was right :) I went to cocoabuilder to find some of the responses to this, that I never got. Anyways, it seems I can just call AEWP() once, and it should stay suid. So, my last question to everyone is, how do I know if the tool is suid. I mean, I could implement some sort of lock-unlock thing, and then I would be pretty sure, but is there anyway to know for sure? And, is there a way to take SUID away from it? So, if I want to lock it, it would work. man 2 stat to check the bit, man 2 chmod to set it. Mike ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/joeturner%40me.com This email sent to joetur...@me.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: NSNumberFormatter setFormat: in 10.3.9
On 23 Jan 2009, at 18:30, Jonathan Fewtrell wrote: My app, developed on 10.5 but targeted at 10.3.9 and above, contains the following code: NSNumberFormatter *formatter = [[NSNumberFormatter alloc] init]; [formatter setFormat:@#0;#0;#0]; It compiles without warnings. When running on 10.3.9 it crashes with EXC_BAD_ACCESS in - [NSNumberFormatter setFormat:]. It works fine on 10.4 and 10.5. For such a simple bit of code, I struggle to see what could be wrong. A previous version, developed on 10.4, did not show this problem, so I'm wondering if it's a build setting problem. If it is, I can't see which one. I can see from the archives that NSNumberFormatter has various issues, but they do not seem to me to be relevant here. I have now established that the problem only occurs if the above code is in a framework bundle rather than an app bundle. And it only occurs if that framework was built on 10.5 (specifically XCode 3.1.2 on OS X 10.5.6). If it is built from the exact same source files on 10.4 (XCode 2.4.1; OS X 10.4.11) the problem does not occur. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Not Resolving Aliases
When I do: [ sharedWorkspace openFile: @/path/to/symlink withApplication: @SomeApp ] then SomeApp does NOT get /path/to/symlink but the content of the symlink. Same for Aliases. Is there a way to open a path without resolving aliases? Kind regards, Gerriet. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
iPhone Compositing Filters
CoreImage is not available on iPhone so the CALayer property compositingFilter is undefined. The docs say that by default a layer is composited using source- over. Is this true for iPhone as well? David Blanton ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
Your build settings below look reasonable, but given where the crash is happening it looks like the problem is with your main nib file. Is it possible that it contains a new Leopard-only view class, for example? Does IB show any incompatibilities with your xib or nib with a 10.4 target? And I'd suggest getting your hands on a 10.4 system of your own, particularly if current or future customers have that as a requirement. My customer does so I'm doing all my development with Xcode 2.5 on a 10.4.11 system. I make sure everything works on 10.4 first, then also make sure that nothing weird happens on 10.5. steve On Jan 24, 2009, at 1:44 AM, Thomas Davie wrote: I'm in a sticky situation. I personally have no machine capable of running tiger, but my customer needs me to provide a tiger version of my app. I attempted to create a build that targeted tiger, but they are reporting that it is failing. What's going wrong: The application crashes almost immediately (on tiger, but not leopard) while loading its nib, here's the back trace: Thread: 0 Exception: EXC_BAD_ACCESS (0x0001) Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x Thread 0 Crashed: 0 libobjc.A.dylib 0x90a594c7 objc_msgSend + 23 1 com.apple.AppKit0x932639d8 loadNib + 303 2 com.apple.AppKit 0x932633d9 +[NSBundle(NSNibLoading) _loadNibFile:nameTable:withZone:ownerBundle:] + 918 3 com.apple.AppKit 0x9326303a +[NSBundle(NSNibLoading) loadNibFile:externalNameTable:withZone:] + 164 4 com.apple.AppKit 0x93262f7c +[NSBundle(NSNibLoading) loadNibNamed:owner:] + 392 5 com.apple.AppKit0x93262cc3 NSApplicationMain + 433 6 com.davie.klondike 0x237a _start + 216 7 com.davie.klondike 0x22a1 start + 41 Thread 0 crashed with X86 Thread State (32-bit): eax: 0x ebx: 0x927db6ae ecx: 0x90ac0010 edx: 0x0041 edi: 0x0008 esi: 0xbaf8 ebp: 0xbb88 esp: 0xbac4 ss: 0x001f efl: 0x00010202 eip: 0x90a594c7 cs: 0x0017 ds: 0x001f es: 0x001f fs: 0x gs: 0x0037 What I've done to try and get it to run on tiger: Set the following build settings in my target: ARCHS = $(ARCHS_STANDARD_32_BIT) SDKROOT = macosx10.4 MACOSX_DEPLOYMENT_TARGET = 10.4 GCC_ENABLE_OBJC_GC = unsupported In the xib file, set the target to 10.4 (I've also tried saving a version 2.0 nib instead of a xib). I'm building using Xcode 3.1.2. If anyone could point me to what setting I've missed, or some documentation I can read on this, I'd be really grateful. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
I agree with Steve as this is exactly how I develop to be sure 10.4 and 10.5 work. On Jan 24, 2009, at 8:56 AM, Steve Christensen wrote: And I'd suggest getting your hands on a 10.4 system of your own, particularly if current or future customers have that as a requirement. My customer does so I'm doing all my development with Xcode 2.5 on a 10.4.11 system. I make sure everything works on 10.4 first, then also make sure that nothing weird happens on 10.5. David Blanton ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: NSCaledarDate's deprecation
Please excuse this possibly dumb question but here it is: Why spend so much time and effort to remove some code that is going to keep running fine for years? The class isn't even deprecated yet. It's in this weird going to be deprecated limbo. How many years does it take between when something is deprecated and when it actually breaks? (not a rhetorical question) ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
Not sure what you mean by capable of running tiger. If you have a machine capable of running Leopard, it should be able to run tiger. Someone correct me if I'm wrong as I haven't actually done this, but can' t you simply get an external drive you connect through USB and install tiger onto it while leaving your current system in tact on the internal drive? It's going to be a nightmare to debug the environment if you don't have a dev environment to match. I've (hate to admit it) tried this before. It did not work out well and the customer was pretty frustrated as I was trying to get them to try my debug builds and report back to me. Not one of my brighter moments. ;-) Seems like getting a tiger system set up shouldn't be too tough. Or am I way off here? Best regards, -Matt On Jan 24, 2009, at 8:56 AM, Steve Christensen wrote: Your build settings below look reasonable, but given where the crash is happening it looks like the problem is with your main nib file. Is it possible that it contains a new Leopard-only view class, for example? Does IB show any incompatibilities with your xib or nib with a 10.4 target? And I'd suggest getting your hands on a 10.4 system of your own, particularly if current or future customers have that as a requirement. My customer does so I'm doing all my development with Xcode 2.5 on a 10.4.11 system. I make sure everything works on 10.4 first, then also make sure that nothing weird happens on 10.5. steve On Jan 24, 2009, at 1:44 AM, Thomas Davie wrote: I'm in a sticky situation. I personally have no machine capable of running tiger, but my customer needs me to provide a tiger version of my app. I attempted to create a build that targeted tiger, but they are reporting that it is failing. What's going wrong: The application crashes almost immediately (on tiger, but not leopard) while loading its nib, here's the back trace: Thread: 0 Exception: EXC_BAD_ACCESS (0x0001) Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x Thread 0 Crashed: 0 libobjc.A.dylib 0x90a594c7 objc_msgSend + 23 1 com.apple.AppKit0x932639d8 loadNib + 303 2 com.apple.AppKit 0x932633d9 +[NSBundle(NSNibLoading) _loadNibFile:nameTable:withZone:ownerBundle:] + 918 3 com.apple.AppKit 0x9326303a +[NSBundle(NSNibLoading) loadNibFile:externalNameTable:withZone:] + 164 4 com.apple.AppKit 0x93262f7c +[NSBundle(NSNibLoading) loadNibNamed:owner:] + 392 5 com.apple.AppKit0x93262cc3 NSApplicationMain + 433 6 com.davie.klondike 0x237a _start + 216 7 com.davie.klondike 0x22a1 start + 41 Thread 0 crashed with X86 Thread State (32-bit): eax: 0x ebx: 0x927db6ae ecx: 0x90ac0010 edx: 0x0041 edi: 0x0008 esi: 0xbaf8 ebp: 0xbb88 esp: 0xbac4 ss: 0x001f efl: 0x00010202 eip: 0x90a594c7 cs: 0x0017 ds: 0x001f es: 0x001f fs: 0x gs: 0x0037 What I've done to try and get it to run on tiger: Set the following build settings in my target: ARCHS = $(ARCHS_STANDARD_32_BIT) SDKROOT = macosx10.4 MACOSX_DEPLOYMENT_TARGET = 10.4 GCC_ENABLE_OBJC_GC = unsupported In the xib file, set the target to 10.4 (I've also tried saving a version 2.0 nib instead of a xib). I'm building using Xcode 3.1.2. If anyone could point me to what setting I've missed, or some documentation I can read on this, I'd be really grateful. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/matt.long%40matthew-long.com This email sent to matt.l...@matthew-long.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Log4Cocoa
On Jan 23, 2009, at 3:33 PM, Joel Norvell wrote: This doesn't answer the original question, but I believe it is pertinent to this thread. It is also possible to log from within Xcode, something I hadn't realized until I saw the video of an excellent talk Joar Wingfors gave at a Silicon Valley Cocoaheads. Joel http://video.google.com/videoplay?docid=3598467380866353869 http://theocacao.com/downloads/DebuggingWithXcode.pdf Thanks for the links. Logging for development debugging is only one small use case for logging though. And the people using Apple's log service only see one additional use case, which is for logging on the developers machine. But these two use cases aren't that important to software that is sold for use. If a developer (individual or team) has software that they have to support, and users report defects, that is where things like NSLog ASL fail, and a true logging framework is needed. When trying to figure out a customers issue, you can have them manually open the system log file and search for specific info. You also can't have the app always logging debug level detail, as this not only consumes large amounts of disk space and impacts performance greatly, but is almost useless to the developer (I've seen apps produce logs in the gigabyte size). Frameworks like Log4Cocoa are a means of doing just about anything you want with logging. In production software, it's valuable because it allows a compiled application to change it'[s logging behavior with either a restart (if using a config file), or by a few settings in the a preferences window and no need to restart. The app then can produce a log file with very detailed logging around the suspected classes/ subsystems that are of interest, while leaving the rest of the system only logging at an error level (for example). The app can then even send that log file to he developer automatically, or the user can compress it an e-mail it. These kinds of things are not possible with NSLog or ASL, without building a log framework around them. If the software you are writing is for in-house use, or your own use, then perhaps NSLog ASL are what's best for you. And perhaps for most Mac software that is the case. But there are development projects out there that simply need more. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
On Jan 24, 2009, at 1:19 AM, Antonio Nunes wrote:
I need to be able to force the requested allocation of a cocoa class
to always return an instance of my subclass. I have looked into ways
of doing that.
First things first
Why?
Or, more specifically, why do you want to make some bit of framework
code which currently allocates an instance of A allocate an instance
of SubA instead?
This is an inherently fragile path to follow.
I have a solution that seems to work, by using a category on the
class to replace the regular invocation, but from the archives I
understand that is a Bad Idea that can lead to all manner of
unpredictable and hard to trace errors. The current implementation
looks like this:
@implementation PDFDocument (PDFDocument_Alloc)
+ (id)alloc
{
if ([[self class] isEqual:[PDFDocument class]]) {
return [ANPDFDocument alloc];
} else {
return [super alloc];
}
}
@end
This is an excellent demonstration of why category based overrides are
rife with danger. The above code is making a boatload of assumptions.
Not the least of which, it assumes that PDFDocument doesn't already
override +alloc. If it does, the above breaks it.
The code also assumes that everything that uses PDFDocument is
perfectly fine with a subclass of PDFDocument instead. Likely a safe
assumption for external-to-the-PDF-framework clients, but maybe not so
much for the internal implementation details of the PDF framework
itself.
Finally, it assumes that PDFDocument, in and of itself, has nothing
within itself that uses, say, -isMemberOfClass: to guide behavior.
There are even more ways that the above might break stuff, but you get
the idea.
However, if you do decide to go down this path, you'll want to test
across every architecture and bit-width as allocation behavior is one
of those things that might optimized at such a specific level.
And you would be counting on the allocation behavior to never change
over releases or updates.
As I wrote, this appears to work fine. But if indeed this technique
is better avoided, what would be the best alternative? I have looked
at class_replaceMethod, but am having difficulties implementing it.
What would be the best way to implement a safe alternative to the
method listed above?
I would use method_exchangeImplementations(), if absolutely positively
necessary. It is about the most innocuous of an otherwise noxious
encapsulation breaking implementation pattern.
The real question, though, is what are you trying to do?
b.bum
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
On 24 Jan 2009, at 18:09, Bill Bumgarner wrote:
Or, more specifically, why do you want to make some bit of framework
code which currently allocates an instance of A allocate an instance
of SubA instead?
Because of, what I assume to be, a bug in PDFClerkThumbnailView which
when one drags multiple pages from or within it, puts the pages in a
PDFDocument instead of in the subclass of PDFDocument, which is where
the pages live. Dragging a single page works as expected and the page
is of the correct subclass. Dragging multiple pages leads to errors. I
filed a bug report for this a while back, but I really need a safe
workaround until the issue is fixed. (Assuming Apple considers it a
bug too. I have never received feedback on it.)
This is an excellent demonstration of why category based overrides
are rife with danger. The above code is making a boatload of
assumptions.
I know. That's why I wrote that I understand it is not the way to go,
but yet that it _appears_ to work.
I would use method_exchangeImplementations(), if absolutely
positively necessary. It is about the most innocuous of an
otherwise noxious encapsulation breaking implementation pattern.
Thanks. I found that a few hours ago and was able to create the
exchange. However, writing a correct replacement method is another
thing altogether. The following doesn't work:
Early in application startup:
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(alloc));
Method replacedMethod = class_getClassMethod([PDFDocument class],
@selector(allocReplacement));
method_exchangeImplementations(originalMethod, replacedMethod);
Addition to PDFDocument:
+ (id)allocReplacement
{
if ([[self class] isEqual:[PDFDocument class]]) {
return [ANPDFDocument alloc];
} else {
return [super alloc];
}
}
I get the impression it replaces alloc for everything, not just for
the PDFClerkDocument class. So the app crashes immediately after the
implementation exchange. Which doesn't surprise me really. But: Is
there a way to make this work using method_exchangeImplementations, or
maybe for the particular problem at hand, is there a better way of
making the PDFThumbnailView use the correct PDFDocument (sub)class
when it deals with multiple selections.
Just to make this clear: I would really rather not create this
workaround, but I can't really afford not to let users drag multiple
pages in the PDFClerkThumbnailView either.
-António
There is nothing as strong as real gentleness, and
there is nothing as gentle as real strength.
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Knowing when a NSArrayController is ready
It looks like you can set to prepare content AND call fetchWithRequest: http://developer.apple.com/DOCUMENTATION/Cocoa/Conceptual/CoreData/ Articles/cdBindings.html If the automatically prepares content flag (see, for example, setAutomaticallyPreparesContent:) is set for a controller, the controller's initial content is fetched from its managed object context using the controller's current fetch predicate. It is important to note that the controller's fetch is executed as a delayed operation performed after its managed object context is set (by nib loading)—this therefore happens after awakeFromNib and windowControllerDidLoadNib:. This can create a problem if you want to perform an operation with the contents of an object controller in either of these methods, since the controller's content is nil. You can work around this by executing the fetch manually with fetchWithRequest:merge:error:. You pass nil as the fetch request argument to use the default request, as illustrated in the following code fragment. On Jan 24, 2009, at 12:59 PM, Matt Neuburg wrote: On Sat, 24 Jan 2009 09:07:59 +, jonat...@mugginsoft.com jonat...@mugginsoft.com said: On 24 Jan 2009, at 02:43, David LeBer wrote: This may be a stupid question, but I've had reason to wonder a number of times. If I have an NSArrayController in a nib, bound to a MOC, configured with an Entity name and set to prepare content: Is there a way to be notified when it has finished loading and arranging it's objects? Hi David Adding an observer on the controller's arrangedObjects property should do the trick. [arrayController addObserver:self forKeyPath:@arrangedObjects options:0 context:(void *)ControllerContentContext]; In a persistent document based app, for example, the controller's MO content is not yet available in - (void)windowControllerDidLoadNib: (NSWindowController *)windowController. Adding the observer means you can set up any dependent code when the binding/core data machinery has finished doing its thing. Is that really true? I'm not saying you're wrong; I'm really asking. When I posed the same question a couple of years back, I was told that the solution is, do *not* set to prepare content; instead, load the data yourself by calling fetchWithRequest: [arrayController fetchWithRequest:nil merge:NO error:error]; When that call returns, the data is ready. I just now tried adding addObserver: to the mix, just as an experiment, and logging reveals that I get two corresponding calls to observeValueForKeyPath:, *before* fetchWithRequest: has returned. The two calls are quite far apart (half a second or so). Well, I sure wouldn't want to run my dependent code half a second too early. So I'm not convinced that this approach is entirely safe, at least not without some further modification... m. -- matt neuburg, phd = m...@tidbits.com, http://www.tidbits.com/matt/ A fool + a tool + an autorelease pool = cool! One of the 2007 MacTech Top 25: http://tinyurl.com/2rh4pf AppleScript: the Definitive Guide - Second Edition! http://www.amazon.com/gp/product/0596102119 ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/dave.fernandes% 40utoronto.ca This email sent to dave.fernan...@utoronto.ca ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
Le 24 janv. 09 à 18:51, Antonio Nunes a écrit :
On 24 Jan 2009, at 18:09, Bill Bumgarner wrote:
Or, more specifically, why do you want to make some bit of
framework code which currently allocates an instance of A allocate
an instance of SubA instead?
Because of, what I assume to be, a bug in PDFClerkThumbnailView
which when one drags multiple pages from or within it, puts the
pages in a PDFDocument instead of in the subclass of PDFDocument,
which is where the pages live. Dragging a single page works as
expected and the page is of the correct subclass. Dragging multiple
pages leads to errors. I filed a bug report for this a while back,
but I really need a safe workaround until the issue is fixed.
(Assuming Apple considers it a bug too. I have never received
feedback on it.)
This is an excellent demonstration of why category based overrides
are rife with danger. The above code is making a boatload of
assumptions.
I know. That's why I wrote that I understand it is not the way to
go, but yet that it _appears_ to work.
I would use method_exchangeImplementations(), if absolutely
positively necessary. It is about the most innocuous of an
otherwise noxious encapsulation breaking implementation pattern.
Thanks. I found that a few hours ago and was able to create the
exchange. However, writing a correct replacement method is another
thing altogether. The following doesn't work:
Early in application startup:
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(alloc));
Method replacedMethod = class_getClassMethod([PDFDocument class],
@selector(allocReplacement));
method_exchangeImplementations(originalMethod, replacedMethod);
Addition to PDFDocument:
+ (id)allocReplacement
{
if ([[self class] isEqual:[PDFDocument class]]) {
return [ANPDFDocument alloc];
} else {
return [super alloc];
}
}
First, as it was said, the designated allocation method is
allocWithZone and not alloc.
Then, you call alloc in a recursive way in your method. alloc was
replaced by your allocReplacement, so calling alloc call
allocReplacement, even in you replacement implementation.
see http://developer.apple.com/samplecode/MethodReplacement/listing3.html
Then, alloc is a class method, so self is already the class. No need
to do [self class]
It should be
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZone:));
Method replacedMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZoneReplacement:));
method_exchangeImplementations(originalMethod, replacedMethod);
+ (id)allocWithZoneReplacement:(NSZone *)aZone {
if (self == [PDFDocument class]) {
return [ANPDFDocument allocWithZoneReplacement:aZone];
} else {
return [super allocWithZoneReplacement:aZone];
}
}
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Long load time of managed object context
Date: Fri, 16 Jan 2009 13:03:24 +0100 From: Matteo Manferdini mat.maili...@gmail.com Subject: Long load time of managed object context I'm testing performance in my application, to see how it behaves with a big load of data inside of it. My application is core data based and uses array controllers in entity mode to fetch data. What I discovered is that my app fetched all the managed objects present in core data, so I set the fetch predicate of array controllers at startup. This reduced dramatically the number of objects fetched. The problem though is that my app still takes a lot of time to start when is full of data. I ran the app within instruments, and the core data fetches instrument confirms that this is not the problem (my app fetches only 20-30 objects when starts). But running it under Shark I can see that for 4-5 seconds after startup the only call on the stack is to the app managed object context. So I think it has something to do with interaction between managed object context and data stored, even if I don't fetch all the data. If I empty the app or leave little data inside of it, it starts very quickly. Does somebody know why this happens? Am I missing something obvious? Why does it take so much to the managed object context to load if I only fetch few objects? I always read that core data scales well and the programming guide states that 10.000 objects are not much for core data, but in my app it makes a big difference, so I'm wondering where I am wrong. You don't say what format you're using, but if you're using XML, the entire XML file has to be loaded and parsed, even before you fetch any objects. m. -- matt neuburg, phd = m...@tidbits.com, http://www.tidbits.com/matt/ pantes anthropoi tou eidenai oregontai phusei Among the 2007 MacTech Top 25, http://tinyurl.com/2rh4pf AppleScript: the Definitive Guide, http://tinyurl.com/2ouo3b Take Control of Customizing Leopard, http://tinyurl.com/2t9629 TidBITS, Mac news and reviews since 1990, http://www.tidbits.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Knowing when a NSArrayController is ready
On 24 Jan 2009, at 17:59, Matt Neuburg wrote: On Sat, 24 Jan 2009 09:07:59 +, jonat...@mugginsoft.com jonat...@mugginsoft.com said: On 24 Jan 2009, at 02:43, David LeBer wrote: This may be a stupid question, but I've had reason to wonder a number of times. If I have an NSArrayController in a nib, bound to a MOC, configured with an Entity name and set to prepare content: Is there a way to be notified when it has finished loading and arranging it's objects? Hi David Adding an observer on the controller's arrangedObjects property should do the trick. [arrayController addObserver:self forKeyPath:@arrangedObjects options:0 context:(void *)ControllerContentContext]; In a persistent document based app, for example, the controller's MO content is not yet available in - (void)windowControllerDidLoadNib: (NSWindowController *)windowController. Adding the observer means you can set up any dependent code when the binding/core data machinery has finished doing its thing. Is that really true? I'm not saying you're wrong; I'm really asking. When I posed the same question a couple of years back, I was told that the solution is, do *not* set to prepare content; instead, load the data yourself by calling fetchWithRequest: [arrayController fetchWithRequest:nil merge:NO error:error]; I have been utilising the NSPersistentDocument sample code with bindings and prepareContent enabled. Going back to my code I see that I was actually observing the content of an NSObjectController not the arrangedObjects of an NSArrayController. Apologies for that. This observation seems to work okay, but there may well be issues as you suggest observing arrangedObjects. When that call returns, the data is ready. I just now tried adding addObserver: to the mix, just as an experiment, and logging reveals that I get two corresponding calls to observeValueForKeyPath:, *before* fetchWithRequest: has returned. The two calls are quite far apart (half a second or so). Well, I sure wouldn't want to run my dependent code half a second too early. So I'm not convinced that this approach is entirely safe, at least not without some further modification... m. I am also having horrible performance problems. A data set of 1500 items with a total on disk size of 1.8MB is taking more than 30 secs to load. Maybe NSArrayController -fetchWithRequest will improve things. -- matt neuburg, phd = m...@tidbits.com, http://www.tidbits.com/matt/ A fool + a tool + an autorelease pool = cool! One of the 2007 MacTech Top 25: http://tinyurl.com/2rh4pf AppleScript: the Definitive Guide - Second Edition! http://www.amazon.com/gp/product/0596102119 Jonathan Mitchell Central Conscious Unit http://www.mugginsoft.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
Le 24 janv. 09 à 19:20, Jean-Daniel Dupas a écrit :
Le 24 janv. 09 à 18:51, Antonio Nunes a écrit :
On 24 Jan 2009, at 18:09, Bill Bumgarner wrote:
Or, more specifically, why do you want to make some bit of
framework code which currently allocates an instance of A allocate
an instance of SubA instead?
Because of, what I assume to be, a bug in PDFClerkThumbnailView
which when one drags multiple pages from or within it, puts the
pages in a PDFDocument instead of in the subclass of PDFDocument,
which is where the pages live. Dragging a single page works as
expected and the page is of the correct subclass. Dragging multiple
pages leads to errors. I filed a bug report for this a while back,
but I really need a safe workaround until the issue is fixed.
(Assuming Apple considers it a bug too. I have never received
feedback on it.)
This is an excellent demonstration of why category based overrides
are rife with danger. The above code is making a boatload of
assumptions.
I know. That's why I wrote that I understand it is not the way to
go, but yet that it _appears_ to work.
I would use method_exchangeImplementations(), if absolutely
positively necessary. It is about the most innocuous of an
otherwise noxious encapsulation breaking implementation pattern.
Thanks. I found that a few hours ago and was able to create the
exchange. However, writing a correct replacement method is another
thing altogether. The following doesn't work:
Early in application startup:
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(alloc));
Method replacedMethod = class_getClassMethod([PDFDocument class],
@selector(allocReplacement));
method_exchangeImplementations(originalMethod, replacedMethod);
Addition to PDFDocument:
+ (id)allocReplacement
{
if ([[self class] isEqual:[PDFDocument class]]) {
return [ANPDFDocument alloc];
} else {
return [super alloc];
}
}
First, as it was said, the designated allocation method is
allocWithZone and not alloc.
Then, you call alloc in a recursive way in your method. alloc was
replaced by your allocReplacement, so calling alloc call
allocReplacement, even in you replacement implementation.
see http://developer.apple.com/samplecode/MethodReplacement/listing3.html
Then, alloc is a class method, so self is already the class. No need
to do [self class]
It should be
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZone:));
Method replacedMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZoneReplacement:));
method_exchangeImplementations(originalMethod, replacedMethod);
+ (id)allocWithZoneReplacement:(NSZone *)aZone {
if (self == [PDFDocument class]) {
return [ANPDFDocument allocWithZoneReplacement:aZone];
} else {
return [super allocWithZoneReplacement:aZone];
}
}
Sorry, this is not
return [super allocWithZoneReplacement:aZone];
but
return [self allocWithZoneReplacement:aZone];
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
Le 24 janv. 09 à 19:20, Bill Bumgarner a écrit :
Thanks. I found that a few hours ago and was able to create the
exchange. However, writing a correct replacement method is another
thing altogether. The following doesn't work:
Early in application startup:
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(alloc));
Method replacedMethod = class_getClassMethod([PDFDocument class],
@selector(allocReplacement));
method_exchangeImplementations(originalMethod, replacedMethod);
Addition to PDFDocument:
+ (id)allocReplacement
{
if ([[self class] isEqual:[PDFDocument class]]) {
return [ANPDFDocument alloc];
} else {
return [super alloc];
}
}
I get the impression it replaces alloc for everything, not just for
the PDFClerkDocument class. So the app crashes immediately after
the implementation exchange. Which doesn't surprise me really. But:
Is there a way to make this work using
method_exchangeImplementations, or maybe for the particular problem
at hand, is there a better way of making the PDFThumbnailView use
the correct PDFDocument (sub)class when it deals with multiple
selections.
Just to make this clear: I would really rather not create this
workaround, but I can't really afford not to let users drag
multiple pages in the PDFClerkThumbnailView either.
Ahh... yes -- my suggestion was slightly misleading.
class_getClassMethod() will walk the class hierarchy and return a
super's implementation, if the method itself is not found in the
subclass.
This is expected, and not a problem with exchange implementation.
You'll need to detect whether or not the method already exists in
the targeted class. You can use class_copyMethodList() to grab the
methods specific to the class and then hunt about within to see if
the targeted method is implemented.
Not require here, but my way to do this is the following one:
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZone:));
Method superMethod =
class_getClassMethod(class_getSuperclass([PDFDocument class]),
@selector(allocWithZone:));
if (superMethod == originalMethod) {
// PDF Document does not override allowWithZone:
}
If so, the method_exchangeImplementations() pattern will work. If
not, you'll need to add the method to the class via class_addMethod().
Normally, when doing this kind of thing, you absolutely positively
must call the original implementation of the method -- whether it be
in the class or in the superclass -- to ensure that you preserve any
existing behaviors.
Yes, but to call the original impl, you don't use super. You call
[self allocReplacement]
+alloc, though, is slightly different in that classes typically only
override +alloc when they want to do some kind of custom allocation
behavior and, by doing so, they would almost never (but not never)
call through to super's implementation.
And so, exchanging alloc is not different than exchanging any other
method.
If you are actually replacing +alloc (which, in this case, it seems
not), well that can be very very tricky.
b.bum
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Knowing when a NSArrayController is ready
On or about 1/24/09 10:17 AM, thus spake jonat...@mugginsoft.com jonat...@mugginsoft.com: I am also having horrible performance problems. A data set of 1500 items with a total on disk size of 1.8MB is taking more than 30 secs to load. Maybe NSArrayController -fetchWithRequest will improve things. I would ask you the same question I just asked someone else on the list: are you using the XML persistent storage format? Because, if so, you must expect some delay, since even before you fetch any objects the entire XML file must be loaded into memory and parsed. I use XML because I want human readability for my data, so startup speed is a sacrifice I am prepared to make. My data consists of about 5000 items, but they are very simple; dealing with complex objects would take considerably longer, one assumes. In order to cave-man-instrument the startup procedure via NSLog, I do NOT prepare content or even bind my array controller's managed object content in the nib; instead, I do it all manually in awakeFromNib. Here are some NSLog timings: 2009-01-24 10:37:09.157 [788] awakening from nib 2009-01-24 10:37:09.269 [788] binding to MOC 2009-01-24 10:37:09.359 [788] creating whole shebang 2009-01-24 10:37:09.360 [788] adding persistent store 2009-01-24 10:37:13.877 [788] calling fetch 2009-01-24 10:37:14.192 [788] done calling fetch Notice the big delay between adding persistent store and calling fetch; that consists a single line of code (calling addPersistentStoreWithType:). So the delay there is nothing but the XML file being loaded and parsed. m. -- matt neuburg, phd = m...@tidbits.com, http://www.tidbits.com/matt/ pantes anthropoi tou eidenai oregontai phusei Among the 2007 MacTech Top 25, http://tinyurl.com/2rh4pf AppleScript: the Definitive Guide, http://tinyurl.com/2ouo3b Take Control of Customizing Leopard, http://tinyurl.com/2t9629 TidBITS, Mac news and reviews since 1990, http://www.tidbits.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Using the security framework
I think I figured out how SD does it: When you unlock SD!, it calls AEWP() on SDAgent. Then, SDAgent calls setuid(0) to make itself root. With it as root, when it calls SDCopy, or SDDiskTool, it calls it with AEWP, and since it's root, it doesn't need the user's password to do this! This is just my guess. But it sounds fairly right :) Cheers, Joe Turner On Jan 24, 2009, at 6:47 AM, Michael Ash wrote: On Fri, Jan 23, 2009 at 10:59 PM, Joe Turner joetur...@me.com wrote: Okay, so, it seems everyone was right :) I went to cocoabuilder to find some of the responses to this, that I never got. Anyways, it seems I can just call AEWP() once, and it should stay suid. So, my last question to everyone is, how do I know if the tool is suid. I mean, I could implement some sort of lock-unlock thing, and then I would be pretty sure, but is there anyway to know for sure? And, is there a way to take SUID away from it? So, if I want to lock it, it would work. man 2 stat to check the bit, man 2 chmod to set it. Mike ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/joeturner%40me.com This email sent to joetur...@me.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
24 jan 2009 kl. 16.56 skrev Steve Christensen: And I'd suggest getting your hands on a 10.4 system of your own, particularly if current or future customers have that as a requirement. My customer does so I'm doing all my development with Xcode 2.5 on a 10.4.11 system. I make sure everything works on 10.4 first, then also make sure that nothing weird happens on 10.5. I don't know if it will be much of a help, but Xcode 2.5 can be installed side by side with Xcode 3 on Leopard. smime.p7s Description: S/MIME cryptographic signature ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
Your best bet is to purchase a used Mac that can run Tiger and use it for test builds. There is a huge installed userbase of Tiger still, remember it has been the longest single incarnation of OS X. (as a result of this long life of Tiger, there are lots of affordable used macs that can do the work. An iBook G4 or even a dual-processor G4 PowerMac, or a G5 iMac would be totally affordable and reasonable options that would handle it well.) You second best bet, install Xcode 2.5 on Leopard. This will be very near to the same as what can actually run on Tiger and will help you be sure that you're developing using Nibs and items that can compile on Tiger. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
On Jan 24, 2009, at 8:48 AM, Matt Long wrote: Not sure what you mean by capable of running tiger. If you have a machine capable of running Leopard, it should be able to run tiger. Macs typically doesn't support operating system releases earlier than the one with which it shipped, because earlier releases don't include the appropriate hardware drivers. -- Chris ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
NSRunloop performSelector needs CFRunLoopWakeUp ?
I have a few hundred tasks to handle, and am using NSRunLoop's
performSelector to get the scheduled:
-(void)nextTask:(SEL)sel
{
[[NSRunLoop mainRunLoop] performSelector:sel target:self argument:nil order:0
modes:[NSArray arrayWithObject:NSDefaultRunLoopMode]];
CFRunLoopRef crf = [[NSRunLoop mainRunLoop] getCFRunLoop];
CFRunLoopWakeUp(crf);
}
One process sends the first message to the worker object. When the
first task is done, nextTask: is used to schedule the next one.
Well, it was bizarre - before the CFRunLoopWakeUp addition, only 6 or
7 of the tasks would run, then the process would stall. NSLogs showed
the performSelector message, then nothing.
But, wait, there's more! If I clicked on a window, or clicked on a
menu, then 7 more tasks would run, then another stall. Every click
caused 7 more messages.
In desperation I added the CFRunLoopWakeUp() call; now it runs as I
would expect.
There is an oblique reference to CFRunLoopWakeUp in the Threading
Programming Guide (where documentation on Run Loop Management has
moved), but the usages is for a different purpose (it would seem).
What seems really odd is that I have used this technique in the past
without issue. The current app is single threaded.
Can anyone shed any light on why the CFRunLoopWakeUp() call is
necessary? Is there some other issue I'm missing?
David
PS: Xcode 3.1.2, Leopard 32bit app
PSS: yes, I did google on NSRunLoop performSelector
CFRunLoopWakeUp, and did find the below note from 2002. However,
there is not mention of this in the Threading Programming Guide. Note
that [NSRunLoop performSelector] results in the creation of a very
short timer:
If you add a timer to the main thread's run loop (which presumably
you've squirreled away somewhere), be sure to use CFRunLoopWakeUp() on
the CFRunLoopRef of the NSRunLoop. Cocoa won't do that for you
currently (and it's not clear it should -- might degrade the performance
unnecessarily).
Chris Kane
Cocoa Frameworks, Apple
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
On 24 Jan 2009, at 19:20, Bill Bumgarner wrote:
If so, the method_exchangeImplementations() pattern will work. If
not, you'll need to add the method to the class via class_addMethod().
Let's see if I understand this correctly: you mean I need to add
alloc, or as suggested, allocWithZone: zone to the PDFClerkDocument
class, and after that, I suppose, exchange the method implementations?
Make use of Bill's and Jean-Daniel's suggestions I now have:
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZone:));
Method superMethod =
class_getClassMethod(class_getSuperclass([PDFDocument class]),
@selector(allocWithZone:));
if (superMethod == originalMethod) {
// PDF Document does not override allocWithZone:
IMP replacement = class_getMethodImplementation([PDFDocument class],
@selector(replacementAllocWithZone:));
BOOL success = class_addMethod([PDFDocument class],
@selector(allocWithZone:),
replacement,
@@:@);
}
I verified that the method is added correctly. However, the
replacement method appears not to be called, so I must be doing
something wrong.
From my understanding, the second argument to class_addMethod is the
name the added method will have, so that should be
replacementAllocWithZone, the third argument is the IMP of the
replacement, and the last argument is the array of characters
describing the of the method arguments, resp an object as the return
type, self and cmd and the NSZone object. Is this correct?
-António
---
Some things have to be believed to be seen.
--Ralph Hodgson
---
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: NSRunloop performSelector needs CFRunLoopWakeUp ?
On Sat, Jan 24, 2009 at 2:42 PM, David Hoerl dho...@mac.com wrote:
I have a few hundred tasks to handle, and am using NSRunLoop's
performSelector to get the scheduled:
-(void)nextTask:(SEL)sel
{
[[NSRunLoop mainRunLoop] performSelector:sel target:self argument:nil
order:0
modes:[NSArray arrayWithObject:NSDefaultRunLoopMode]];
CFRunLoopRef crf = [[NSRunLoop mainRunLoop] getCFRunLoop];
CFRunLoopWakeUp(crf);
}
[snip]
What seems really odd is that I have used this technique in the past without
issue. The current app is single threaded.
I'm kind of confused as to how 10.5 extended NSRunLoop. On the one
hand, there's this +[NSRunLoop mainRunLoop] method that seems to imply
you can play with the main runloop from other threads. On the other,
the documentation is still littered with warnings that NSRunLoop is
not thread safe. Talk about mixed messages!
Also, -[NSObject
performSelectorOnMainThread:withObject:waitUntilDone:] now uses
NSRunLoopCommonModes -- instead of NSRunLoopDefaultModes -- when it
places the selector in the runloop. And NSRunLoopCommonModes is a new
constant in 10.5.
So I guess my question is what happens when you change your code to this:
-(void)nextTask:(SEL)sel
{
[self performSelectorOnMainThread:sel withObject:nil waitUntilDone:NO];
}
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Targetting Tiger
On Jan 24, 2009, at 9:22 AM, Thomas Davie wrote: On 24 Jan 2009, at 17:48, Matt Long wrote: Not sure what you mean by capable of running tiger. If you have a machine capable of running Leopard, it should be able to run tiger. Fraid not, the two machines I have available to develop on are too new to run tiger – there's no drivers for them. It's going to be a nightmare to debug the environment if you don't have a dev environment to match. I've (hate to admit it) tried this before. It did not work out well and the customer was pretty frustrated as I was trying to get them to try my debug builds and report back to me. Not one of my brighter moments. ;-) Yep, I can imagine, the working on tiger criteria kinda got sprung on me, I'm thinking at the moment that the best solution is to buy an ancient mac, and test it on that. Is an iMac from 2008 ancient? That's what I'm using with Tiger. steve ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
On 24 Jan 2009, at 22:17, Steve Christensen wrote: On Jan 24, 2009, at 9:22 AM, Thomas Davie wrote: On 24 Jan 2009, at 17:48, Matt Long wrote: Not sure what you mean by capable of running tiger. If you have a machine capable of running Leopard, it should be able to run tiger. Fraid not, the two machines I have available to develop on are too new to run tiger – there's no drivers for them. It's going to be a nightmare to debug the environment if you don't have a dev environment to match. I've (hate to admit it) tried this before. It did not work out well and the customer was pretty frustrated as I was trying to get them to try my debug builds and report back to me. Not one of my brighter moments. ;-) Yep, I can imagine, the working on tiger criteria kinda got sprung on me, I'm thinking at the moment that the best solution is to buy an ancient mac, and test it on that. Is an iMac from 2008 ancient? That's what I'm using with Tiger. I didn't claim that all machines that can run Tiger are ancient – I implied that all cheep machines capable of running tiger are ancient. Bob___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Using the security framework
So, you are saying that I must create an install tool, that installs my utility that will run as root? On Jan 24, 2009, at 2:23 PM, Chris Hanson wrote: The proper way to construct everything you've described is discussed in the Authorization Services Programming Guide. Authorization Services Programming Guide http://developer.apple.com/documentation/Security/Conceptual/authorization_concepts/index.html The current, most up-to-date and correct example of constructing an application with a helper that performs privileged operations securely on its behalf is BetterAuthorizationSample which replaces the previous MoreAuthSample example code. BetterAuthorizationSample http://developer.apple.com/samplecode/BetterAuthorizationSample/index.html It is critical to your users' security that you understand everything presented in the above programming guide and example before you attempt to write code that needs to run with elevated privileges. You must not simply use AuthorizationExecuteWithPrivileges to run arbitrary code or scripts as root. You must install your helper tool securely. There's a lot to do if you want to run code with elevated privileges, and it must be done correctly; the rationale for the design recommended by the Programming Guide and BetterAuthorizationSample is documented in Design and Implementation Rationale.txt within the example: BetterAuthorizationSample - Design and Implementation Rationale http://developer.apple.com/samplecode/BetterAuthorizationSample/listing4.html Unfortunately, when it comes to system security there's no I'm just… or I simply want to… All i's must be dotted and all t's crossed, or you risk compromising your users' security. Fortunately there's extensive documentation on how to manage the complexity involved, and you can provide a decent user experience if you do the cooking by the book. -- Chris ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Switching NSDocument based application to tabbed views interface
On 22 Jan 2009, at 1:00 PM, Klaus Backert wrote: It's the difference between Search for: All of the words and Search for: Any of the words. You can choose this in the Advanced search. The default is _Any_, because of which more than 40.000 results are found in this case. Well, that's embarrassing. Sometimes I don't look closely enough. Google spoiled me. — F ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Creating a managed object without adding it to the context?
Hi, How do I programmatically create a managed object (with respect to some entity in a particular context of course) but WITHOUT adding it to the context? The reason I want this is to be able to take that new managed object and add it via an NSArrayController (one that is connected to said context of course) instead. That way I will get all the benefits of the array controller, such as the newly inserted object getting selected in a table view that is bound to the array controller. Regards, U _ News, entertainment and everything you care about at Live.com. Get it now! http://www.live.com/getstarted.aspx___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
Le 24 janv. 09 à 22:13, Antonio Nunes a écrit :
Sorry, sent this before I had finished making the necessary changes,
I don't think I need to exchange any implementations if I add the
method with the correct selector (allocWithZone:)
===
On 24 Jan 2009, at 19:20, Bill Bumgarner wrote:
If so, the method_exchangeImplementations() pattern will work. If
not, you'll need to add the method to the class via
class_addMethod().
Let's see if I understand this correctly: you mean I need to add
alloc, or as suggested, allocWithZone: zone to the PDFClerkDocument
class, and after that, I suppose, exchange the method implementations?
No, you don't have to do this if you use exchangeMethod correctly.
You may want to add one to not affect other objects, but it is not
required.
Just in case:
Calling the original implementation in a method you have exchanged is
done using
[self replacedMethodName]
and in a method added at runtime, you have to call
[super originalMethodName]
Make use of Bill's and Jean-Daniel's suggestions I now have:
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZone:));
Method superMethod =
class_getClassMethod(class_getSuperclass([PDFDocument class]),
@selector(allocWithZone:));
if (superMethod == originalMethod) {
// PDF Document does not override allocWithZone:
IMP replacement = class_getMethodImplementation([PDFDocument
class], @selector(replacementAllocWithZone:));
BOOL success = class_addMethod([PDFDocument class],
@selector(allocWithZone:),
replacement,
@@:@);
}
I verified that the method is added correctly. However, the
replacement method appears not to be called, so I must be doing
something wrong.
From my understanding, the second argument to class_addMethod is the
name the added method will have, so that should be
replacementAllocWithZone, the third argument is the IMP of the
replacement, and the last argument is the array of characters
describing the of the method arguments, resp an object as the return
type, self and cmd and the NSZone object. Is this correct?
Yes it is, but if you don't want to bother with the type string, just
query it using runtime functions:
And you want to add a class method, not an instance method. You have
to add this method to the 'PDFDocument class' meta-class.
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZone:));
Method superMethod =
class_getClassMethod(class_getSuperclass([PDFDocument class]),
@selector(allocWithZone:));
Method replacedMethod = class_getClassMethod([PDFDocument class],
@selector(replacementAllocWithZone:));
if (superMethod == originalMethod) {
class_addMethod(object_getClass([PDFDocument class]),
@selector(allocWithZone:),
method_getImplementation(replacedMethod),
method_getDescription(replacedMethod)-types);
}
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Creating a managed object without adding it to the context?
On 2009 Jan 24, at 14:13, Ulai Beekam wrote: How do I programmatically create a managed object (with respect to some entity in a particular context of course) but WITHOUT adding it to the context? I do not believe this is possible. A managed object without a context would not be managed. The reason I want this is to be able to take that new managed object and add it via an NSArrayController (one that is connected to said context of course) instead. That way I will get all the benefits of the array controller, such as the newly inserted object getting selected in a table view that is bound to the array controller. You indicated in your first sentence that you know which managed object context the new object is destined for. I don't see the problem. Just insert it when you create it. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
On Jan 24, 2009, at 11:34 AM, Marcus wrote: 24 jan 2009 kl. 16.56 skrev Steve Christensen: And I'd suggest getting your hands on a 10.4 system of your own, particularly if current or future customers have that as a requirement. My customer does so I'm doing all my development with Xcode 2.5 on a 10.4.11 system. I make sure everything works on 10.4 first, then also make sure that nothing weird happens on 10.5. I don't know if it will be much of a help, but Xcode 2.5 can be installed side by side with Xcode 3 on Leopard. That's true, but then you couldn't -test- your code on a 10.4 system to make sure it actually works. steve ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Knowing when a NSArrayController is ready
On 24 Jan 2009, at 18:41, Matt Neuburg wrote: On or about 1/24/09 10:17 AM, thus spake jonat...@mugginsoft.com jonat...@mugginsoft.com: I am also having horrible performance problems. A data set of 1500 items with a total on disk size of 1.8MB is taking more than 30 secs to load. Maybe NSArrayController -fetchWithRequest will improve things. I would ask you the same question I just asked someone else on the list: are you using the XML persistent storage format? Because, if so, you must expect some delay, since even before you fetch any objects the entire XML file must be loaded into memory and parsed. I use XML because I want human readability for my data, so startup speed is a sacrifice I am prepared to make. My data consists of about 5000 items, but they are very simple; dealing with complex objects would take considerably longer, one assumes. In order to cave-man-instrument the startup procedure via NSLog, I do NOT prepare content or even bind my array controller's managed object content in the nib; instead, I do it all manually in awakeFromNib. Here are some NSLog timings: 2009-01-24 10:37:09.157 [788] awakening from nib 2009-01-24 10:37:09.269 [788] binding to MOC 2009-01-24 10:37:09.359 [788] creating whole shebang 2009-01-24 10:37:09.360 [788] adding persistent store 2009-01-24 10:37:13.877 [788] calling fetch 2009-01-24 10:37:14.192 [788] done calling fetch Notice the big delay between adding persistent store and calling fetch; that consists a single line of code (calling addPersistentStoreWithType:). So the delay there is nothing but the XML file being loaded and parsed. m. Hi Matt I changed my store type to SQLite and noted an immediate improvement in load times. The biggest boost by far though came from setting NSObjectController setUsesLazyFetching:YES. The 10.5 release notes state that this can give a big boost and for an NSArrayController bound to an NSTableView it seems very true. The only down side is that searches are noticeably slower as the data has to be retrieved from the store, if not already present, at search time. I did not use -fetchWithRequest but left - prepareContent = YES. All bindings were set in the nib. Core data has improved a lot of the years so perhaps the prohibition on - prepareContent no longer applies. But I agree that pulling all this stuff out of the nib lets you get a lot close to the action. Obviously the complexity of the model and the number/complexity of derived properties will be a factor too. My model is very simple, the most complex bit being extracting an NString rep of an RTF data object for search purposes. The NSTableView data now appears in about 1/5 of the time it did previously! With regard to your earlier point I put an observer on the NSArrayController's arrangedObjects and received a single notification as expected. So I don't know why you were getting two notifications. Thanks for your response - it put me on the right track! -- matt neuburg, phd = m...@tidbits.com, http://www.tidbits.com/matt/ pantes anthropoi tou eidenai oregontai phusei Among the 2007 MacTech Top 25, http://tinyurl.com/2rh4pf AppleScript: the Definitive Guide, http://tinyurl.com/2ouo3b Take Control of Customizing Leopard, http://tinyurl.com/2t9629 TidBITS, Mac news and reviews since 1990, http://www.tidbits.com Jonathan Mitchell Central Conscious Unit http://www.mugginsoft.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Using the security framework
On Jan 24, 2009, at 1:41 PM, Joe Turner wrote: So, you are saying that I must create an install tool, that installs my utility that will run as root? I am saying that, in order to maintain your users' system security, you must follow the guidance in the Authorization Services Programming Guide and BetterAuthorizationSample example code when implementing software that needs to run with elevated privileges. Among other things, to be truly secure you must use a secure installation mechanism. Do not write your own install tool — it can't be made secure without itself being installed via a secure installation mechanism. Instead, use Installer.app for your installations since it's included with the operating system and not modifiable with normal user privileges. You can still ship your application as a drag-install; for example, your application itself can include an installer package to do the actual installation of the tool it uses and its launchd plist, and open the package in Installer.app if the tool needs to be installed. -- Chris ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Creating a managed object without adding it to the context?
On 2009 Jan 24, at 14:13, Ulai Beekam wrote: How do I programmatically create a managed object (with respect to some entity in a particular context of course) but WITHOUT adding it to the context? I do not believe this is possible. A managed object without a context would not be managed. The second does not imply the first. You can simply pass in nil to the designated initializer. The object will not be managed. No saving, no undo, no MOC notifications, etc. The biggest issue is that without a MOC, no one is observing relationship changes, so it's trivially easy to corrupt an object graph by failing to maintain inverses by hand. In theory, if an entity had no relationships this would work fine. The reason I want this is to be able to take that new managed object and add it via an NSArrayController (one that is connected to said context of course) instead. That way I will get all the benefits of the array controller, such as the newly inserted object getting selected in a table view that is bound to the array controller. You indicated in your first sentence that you know which managed object context the new object is destined for. I don't see the problem. Just insert it when you create it. You can simply call -insertObject on the MOC with the new MO when you are ready. However, any changes you make to the object between the time of +alloc and -insertObject: will be unobserved. All the same issues about using -awakeFromInsert instead of -initWithEntity:... to perform initializations on modeled properties apply here. - Ben ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
iPhone Compositing
CoreImage is not available on iPhone so the CALayer property compositingFilter is undefined. The docs say that by default a layer is composited using source- over. Is this true for iPhone as well? David Blanton ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
iPhone Vibrate
I want to be sensitive to battery usage in the iPhone. When calling AudioServicesPlayAlertSound ( kSystemSoundID_Vibrate ); How much battery does this use? I may be calling this 'alot' in my application. David Blanton ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
RE: Creating a managed object without adding it to the context?
You indicated in your first sentence that you know which managed object context the new object is destined for. I don't see the problem. Just insert it when you create it. My problem is that when I add directly to the managed object context, the item is not getting selected automatically in the array controller (by the way, I see the selection visually by means of the table view bound to the array controller) even if I have checked the select on insert checkbox on the array controller in Interface Builder. That's why I asked about this. But yes, when I come to think about it now as you say it, it kind of makes little sense to be creating a managed object that is not in the context. _ News, entertainment and everything you care about at Live.com. Get it now! http://www.live.com/getstarted.aspx___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: iPhone Vibrate
On Jan 24, 2009, at 5:31 PM, David Blanton wrote: I want to be sensitive to battery usage in the iPhone. When calling AudioServicesPlayAlertSound ( kSystemSoundID_Vibrate ); How much battery does this use? I may be calling this 'alot' in my application. Reposting messages will not get them answered any quicker. In fact, you risk them not being answered at all. Having said that, this is really the wrong forum for iPhone-specific questions such as this one. It's best to ask on the dedicated iPhone developer forums when you log in to your iPhone dev account. ___ Ricky A. Sharp mailto:rsh...@instantinteractive.com Instant Interactive(tm) http://www.instantinteractive.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Log4Cocoa
I'm in the process of kicking it's tires, and I'm liking it so far. Logging frameworks are one of those things that I'm surprised we don't see more Open Source solutions. Probably because logging frameworks aren't necessarily glorious, but they're incredibly utilitarian and in my opinion necessary. I look forward to playing more with Log4Cocoa! On 24-Jan-09, at 9:10 AM, Rob Ross wrote: I have never used the Log4Cocoa implementation, but I use the Java version (log4j) every day. It's got a great API so if Log4Cocoa uses most of the same API I would definitely say use it. It's simple, lightweight, and you can learn enough to cover 80% of your needs in a just a few hours. It's also configureable enough to most likely handle that remaining 20% of things you want to do with it. Rob On Jan 21, 2009, at 2:20 PM, Robert Kukuchka wrote: Apparently this Log4Cocoa project is based off of a Java project which is highly configurable. I'm trying to do an evaluation on the code from SF: http://sourceforge.net/projects/log4cocoa/ Files were updated in 08, but mailing list last updated in 06. Not sure if it's dead or what... On 21-Jan-09, at 1:48 PM, jonat...@mugginsoft.com wrote: On 21 Jan 2009, at 21:40, Robert Kukuchka wrote: I was hoping to find something with built in log level support and non-recompile options to turn logging modules on / off On .NET you can accomplish this with the Enterprise Library Logging Application Block, which is very highly configurable. I looked around for similar functionality in Cocoa frameworks about a year ago and did not find anything remotely similar. But maybe it is hiding out there in the bushes somewhere... On 21-Jan-09, at 1:15 PM, Kyle Sluder wrote: On Wed, Jan 21, 2009 at 2:58 PM, Robert Kukuchka rkukuc...@gmail.com wrote: I'm looking into logging frameworks and see references to this project. Does anyone know if this project is still running? I was hoping to find some examples of how to get things setup. Anyone here use it within a commercial application? Our Mac team is small, so if I don't have to roll my own, I'd rather not. Why not use ASL, since it's built into the OS? http://developer.apple.com/documentation/Darwin/Reference/ManPages/man3/asl.3.html --Kyle Sluder ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/jonathan%40mugginsoft.com This email sent to jonat...@mugginsoft.com Jonathan Mitchell Central Conscious Unit http://www.mugginsoft.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/rkukuchka%40gmail.com This email sent to rkukuc...@gmail.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/rob.ross%40gmail.com This email sent to rob.r...@gmail.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Coverflow like background drawing?
UIKit/AppKit are not thread-safe meaning we should never do drawing on a background thread. Yet, if we have say 5 views and for performance reasons we want the main view to draw first and then the other 4 to draw in the background like how CoverFlow will draw its icons/images seemingly in the background, how does one do that? Alex Kac - President and Founder Web Information Solutions, Inc. The optimist proclaims that we live in the best of all possible worlds; and the pessimist fears this is true. -- James Clabell ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Coverflow like background drawing?
Le 25 janv. 09 à 01:10, Alex Kac a écrit : UIKit/AppKit are not thread-safe meaning we should never do drawing on a background thread. Yet, if we have say 5 views and for performance reasons we want the main view to draw first and then the other 4 to draw in the background like how CoverFlow will draw its icons/images seemingly in the background, how does one do that? Coverflow does not draw in the background, it load image in background (and maybe resize them to improve drawing). Drawing 5 or more images should not be slow enough to have to bother with threading. And if it's not fast enough using Cocoa, convert your image into OpenGL texture and use OpenGL to draw them. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Targetting Tiger
On Sat, Jan 24, 2009 at 8:48 AM, Matt Long matt.l...@matthew-long.com wrote: Not sure what you mean by capable of running tiger. If you have a machine capable of running Leopard, it should be able to run tiger. Not true. In general, any Mac requires the latest OS available at the time it was released. So machines released after the release of Leopard usually require Leopard. -- Clark S. Cox III clarkc...@gmail.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Opening Invisible Files
On 24/01/2009, at 11:40 PM, Gerriet M. Denkmann wrote: I want my NSOpenPanel to show all files - same as the Finder does, if one adds AppleShowAllFiles = YES to: ~/Library/Preferences/ com.apple.finder.plist. Is there a way to accomplish this? Kind regards, Gerriet. Yes, just set AppleShowAllFiles to YES in your app's defaults. -- Rob Keniger ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Opening Symbolic Links
On 24/01/2009, at 11:46 PM, Gerriet M. Denkmann wrote: I have a text file, called Text. I did in Terminal: ln -s Text SymlinkToText When I try to open SymlinkToText in TextEdit SymlinkToText is shown in the Open Panel, but when I select it, the Open-button is disabled. Is this a bug or a feature (in 10.5.6) ? I believe this did work in Tiger, but have no machine to test this. Call [yourOpenPanel setResolvesAliases:YES] to make an open panel resolve aliases/symlinks. -- Rob Keniger ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Not Resolving Aliases
On Sat, Jan 24, 2009 at 9:54 AM, Gerriet M. Denkmann gerr...@mdenkmann.de wrote: When I do: [ sharedWorkspace openFile: @/path/to/symlink withApplication: @SomeApp ] then SomeApp does NOT get /path/to/symlink but the content of the symlink. Same for Aliases. Is there a way to open a path without resolving aliases? I haven't tried it, but I would bet that if you use LaunchServices directly then it will do what you say instead of trying to do what you mean. Mike ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: NSCaledarDate's deprecation
On Sat, Jan 24, 2009 at 4:14 PM, Nick Zitzmann n...@chronosnet.com wrote: On Jan 24, 2009, at 9:40 AM, Paul Bruneau wrote: Why spend so much time and effort to remove some code that is going to keep running fine for years? Because you don't know that. Stuff can change at any time and break your application in unexpected ways in the future, though you do get advance warning. Some people around here might remember the chaos that ensued when Apple shipped Mac OS 7.0 18 years ago. Mac OS 7 introduced real 32-bit addressing, which completely blew away every app that assumed that the top 8 bits of a pointer were unused by the OS (in Mac OS 6 and earlier, the OS used 4-byte pointers, but the top byte wasn't doled out by the OS). Bad analogy. That top byte was always marked as reserved. Applications that noticed it was unused and used it were violating Apple's API contract. Apple has, as far as I know, never actually removed or disabled a Mac OS X API without also crossing a binary compatibility boundary. (E.g. they removed the GUI bits of Carbon, but only if you cross into 64-bit land.) NSCalendarDate isn't even deprecated yet. Apple is not going to remove it in any way that would break currently shipping code. They *may* remove it the next time they cross a binary compatibility boundary, but that would only happen after they *officially* deprecate it (instead of this will be deprecated weasel word nonsense) and then your currently shipping code will still be unaffected; only if you try to make your code cross that boundary will you encounter problems. In conclusion, rewriting a bunch of code that uses a class that isn't even deprecated yet, just discouraged, is pointless unless you get some other benefit from doing so. With NSCalendarDate, that other benefit is being able to use non-Gregorian calendars. If that helps you, great, if you're married to Gregorian for reasons other than using NSCalendarDate then there's absolutely no reason to switch now. Mike ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Using the security framework
On Sat, Jan 24, 2009 at 6:08 PM, Chris Hanson c...@me.com wrote: Among other things, to be truly secure you must use a secure installation mechanism. Do not write your own install tool — it can't be made secure without itself being installed via a secure installation mechanism. Instead, use Installer.app for your installations since it's included with the operating system and not modifiable with normal user privileges. I'm afraid I don't understand this advice. Could you explain what sort of vulnerability would exist in a custom install tool that would not exist when using Installer.app to install a custom package? Mike ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Nib memory management under Garbage Collection
On 16/01/2009, at 2:26 PM, mmalc Crawford wrote: Since the collector follows strong references from root objects, and treats as garbage all objects that cannot be reached from a root object, you must ensure that there are strong references to all top- level objects in a nib file (including for example, stand-alone controllers)—otherwise they will be collected. You can create a strong reference simply by adding an outlet to the File's Owner and connecting it to a top-level object. (In practice this is rarely likely to be an issue.) Nib Files: http://developer.apple.com/documentation/Cocoa/Conceptual/GarbageCollection/Articles/gcEssentials.html Just following up on this thread for the archives, there is a definitive guide for Nib Memory Management under GC and non-GC on this page: http://developer.apple.com/DOCUMENTATION/Cocoa/Conceptual/LoadingResources/CocoaNibs/chapter_3_section_4.html#/ /apple_ref/doc/uid/1051i-CH4-SW6 -- Rob Keniger ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Creating a managed object without adding it to the context?
On 2009 Jan 24, at 15:51, Ulai Beekam wrote: My problem is that when I add directly to the managed object context, the item is not getting selected automatically in the array controller (by the way, I see the selection visually by means of the table view bound to the array controller) even if I have checked the select on insert checkbox on the array controller in Interface Builder. I'm still confused about what you're having trouble with. But, inserting a managed object and selecting it in a table controlled by an array controller is a very standard design, like this... http://developer.apple.com/samplecode/DepartmentAndEmployees/index.html When you run that sample project, click Add, fill in a name and you see your new Employee gets selected in the table. Just study how it's done there. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: NSRunloop performSelector needs CFRunLoopWakeUp ?
On Jan 24, 2009, at 10:18 PM, Michael Ash wrote: On Sat, Jan 24, 2009 at 3:42 PM, David Hoerl dho...@mac.com wrote: [...] The current app is single threaded. Can anyone shed any light on why the CFRunLoopWakeUp() call is necessary? Is there some other issue I'm missing? Look at the big master list of what Cocoa classes are thread safe: The OP wrote that his app is single threaded, so the lack of NSRunLoop thread safety shouldn't be a problem. It sounds like the event loop is blocking the runloop, so maybe posting a custom event would work as well as tickling the runloop...but that's just a guess. -- Adam smime.p7s Description: S/MIME cryptographic signature ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: NSRunloop performSelector needs CFRunLoopWakeUp ?
On Sat, Jan 24, 2009 at 11:07 PM, Adam R. Maxwell amaxw...@mac.com wrote: On Jan 24, 2009, at 10:18 PM, Michael Ash wrote: On Sat, Jan 24, 2009 at 3:42 PM, David Hoerl dho...@mac.com wrote: [...] The current app is single threaded. Can anyone shed any light on why the CFRunLoopWakeUp() call is necessary? Is there some other issue I'm missing? Look at the big master list of what Cocoa classes are thread safe: The OP wrote that his app is single threaded, so the lack of NSRunLoop thread safety shouldn't be a problem. It sounds like the event loop is blocking the runloop, so maybe posting a custom event would work as well as tickling the runloop...but that's just a guess. Man, missing one tiny sentence is enough to make a guy look like an idiot. Oh well, thanks for pointing it out. I think you're right about the event. Reading the documentation: This method sets up a timer to perform the aSelector message on the current thread's run loop at the start of the next run loop iteration. I think the words the next run loop iteration are key here. It will invoke your method at the next run loop iteration... but it won't make that iteration happen any sooner! For what the OP wants, I think the best thing would be to just avoid that method. Use -performSelector:withObject:afterDelay: (with a 0 delay) or a non-repeating NSTimer (with a 0 interval). Mike ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Fwd: ibplugin - framework - application
I may have found a solution. I built a separate framework containing the same objects and incorporated it into my app. The app functions without the plugin or plugin framework being installed in library/ frameworks. The framework and plugin need to be installed in order to open the nib file to work with the interface, but I guess that makes sense. Rich Begin forwarded message: From: Richard Humiston rhumis...@comcast.net Date: January 23, 2009 9:09:35 PM EST To: cocoa-dev@lists.apple.com Subject: ibplugin - framework - application I built an ibplugin and framework for a custom control using the ib plugin programming guide documentation. The plugin works fine. Shows up in the library. I can drag it to a window, configure with a custom inspector, etc. When I create an Xcode project and use the control in a nib file and build the application, everything works as expected. I noticed if I remove the frameworks from library/ frameworks, the nib file is not abel to load. If I put the frameworks back in library/frameworks, the app functions again. I also noticed that the plugin needs the framework in order to function in IB. I would like to incorporate the frameworks into my app rather than have it in library/frameworks, but then the plugin won't function in IB while I'm building the interface. It looks like most of the posts here are concerned with ibplugins and frameworks as they apply to IB. Is there a solution to this problem? thank you, Rich ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
building a file path
hello,
i'm writing some code that saves a file on the iphone,
there's no user intervention / file dialog involved.
so once i've found the documents directory
code
NSArray *paths =
NSSearchPathForDirectoriesInDomains(NSDocumentDirectory,
NSUserDomainMask, YES);
if ([paths count] 1)
{
//failed
return;
}
NSString* path = [paths objectAtIndex:0];
/code
should I just manually build a file path string by adding /myfolder/
myfile.xml' to 'path'
or is there a correct way to do it that doesn't involve hard coding
the slashes into the filepath?
cheers
thanks for your help
mathew
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
addSubview bottleneck
Hi,
I am constructing an application that draws rooted trees (graphs with a
distinguished 'first' node) by creating an NSView subclass for each node and
adding them as subviews of the graph view.
There is a big problem when the number of nodes in the tree is large: adding
the subviews takes a very long time. For example, one of my trees has
around 60,000 nodes. It takes about 1.5 seconds to create all of the
objects, and about 200 seconds to add them as subviews.
Can anyone suggest a way to speed the process up? Multithreading didn't
really improve matters.
Each node keeps track of its child nodes in an NSArray, and the subview
addition is recursive:
- (void) add
{
NSAutoreleasePool *autoreleasepool = [[NSAutoreleasePool alloc] init];
[[parent superview] addSubview:self];
for (Vertex * child in children)
[child performSelectorOnMainThread:@selector(add) withObject:nil
waitUntilDone:NO];
//[child add];
[autoreleasepool release];
}
Thanks,
Josh
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: building a file path
You should use the path methods on NSString: NSString * filePath = [[path stringByAppendingPathComponent:@myfolder] stringByAppendingPathComponent:@myfile.xml]; Cheers, Dave On 23 Jan, 2009, at 11:00 PM, mathew wrote: should I just manually build a file path string by adding /myfolder/ myfile.xml' to 'path' or is there a correct way to do it that doesn't involve hard coding the slashes into the filepath? ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Coverflow like background drawing?
CovertFlow source code is in the Developer/SampleCode directory. You should be able to see it there. On 24-Jan-09, at 7:10 PM, Alex Kac wrote: UIKit/AppKit are not thread-safe meaning we should never do drawing on a background thread. Yet, if we have say 5 views and for performance reasons we want the main view to draw first and then the other 4 to draw in the background like how CoverFlow will draw its icons/images seemingly in the background, how does one do that? Alex Kac - President and Founder Web Information Solutions, Inc. The optimist proclaims that we live in the best of all possible worlds; and the pessimist fears this is true. -- James Clabell ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/scott%40cocoadoc.com This email sent to sc...@cocoadoc.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Using the security framework
On Jan 24, 2009, at 6:29 PM, Michael Ash wrote: On Sat, Jan 24, 2009 at 6:08 PM, Chris Hanson c...@me.com wrote: Among other things, to be truly secure you must use a secure installation mechanism. Do not write your own install tool — it can't be made secure without itself being installed via a secure installation mechanism. Instead, use Installer.app for your installations since it's included with the operating system and not modifiable with normal user privileges. I'm afraid I don't understand this advice. Could you explain what sort of vulnerability would exist in a custom install tool that would not exist when using Installer.app to install a custom package? Because Installer.app is installed by the operating system you can - if you've taken appropriate security measures to begin with - be reasonably certain that it hasn't been tampered with. When writing your own install tool, you have a bootstrapping problem: You will eventually need to have the user authorize some untrusted code to run as root - code that could have been modified behind the user's back. An installer package could also have been writable by the user, but modern packages can be signed so their integrity can be checked. -- Chris ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Using the security framework
But you can also code sign nowadays On Jan 24, 2009, at 11:54 PM, Chris Hanson wrote: On Jan 24, 2009, at 6:29 PM, Michael Ash wrote: On Sat, Jan 24, 2009 at 6:08 PM, Chris Hanson c...@me.com wrote: Among other things, to be truly secure you must use a secure installation mechanism. Do not write your own install tool — it can't be made secure without itself being installed via a secure installation mechanism. Instead, use Installer.app for your installations since it's included with the operating system and not modifiable with normal user privileges. I'm afraid I don't understand this advice. Could you explain what sort of vulnerability would exist in a custom install tool that would not exist when using Installer.app to install a custom package? Because Installer.app is installed by the operating system you can - if you've taken appropriate security measures to begin with - be reasonably certain that it hasn't been tampered with. When writing your own install tool, you have a bootstrapping problem: You will eventually need to have the user authorize some untrusted code to run as root - code that could have been modified behind the user's back. An installer package could also have been writable by the user, but modern packages can be signed so their integrity can be checked. -- Chris ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/joeturner%40me.com This email sent to joetur...@me.com ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Forcing allocation of a subclass
On 24 Jan 2009, at 23:19, Jean-Daniel Dupas wrote:
Calling the original implementation in a method you have exchanged
is done using
[self replacedMethodName]
and in a method added at runtime, you have to call
[super originalMethodName]
[...]
if you don't want to bother with the type string, just query it
using runtime functions:
And you want to add a class method, not an instance method. You have
to add this method to the 'PDFDocument class' meta-class.
Method originalMethod = class_getClassMethod([PDFDocument class],
@selector(allocWithZone:));
Method superMethod =
class_getClassMethod(class_getSuperclass([PDFDocument class]),
@selector(allocWithZone:));
Method replacedMethod = class_getClassMethod([PDFDocument class],
@selector(replacementAllocWithZone:));
if (superMethod == originalMethod) {
class_addMethod(object_getClass([PDFDocument class]),
@selector(allocWithZone:),
method_getImplementation(replacedMethod),
method_getDescription(replacedMethod)-types);
}
Thanks for supplying the last bits to the puzzle Jean-Daniel. Initial
testing seems to indicate the replacement is working. This is what I
have now:
Early during app startup is the code adding the replacement method to
the class, as quoted above.
In a category on PDFDocument is the replacement method itself, which
now reads:
@implementation PDFDocument (PDFDocument_Alloc)
+ (id)replacementAllocWithZone:(NSZone *)zone
{
if (self == [PDFDocument class]) {
return [ANPDFDocument replacementAllocWithZone:zone];
} else {
return [super allocWithZone:zone];
}
}
@end
Many thanks to both Bill and Jean-Daniel for helping me through this,
while pointing out the potential pit-falls. Here's to hoping an update
to Leopard will soon allow me to disable this unfortunate workaround
for it and future updates.
-António
---
Don't believe everything you think
---
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Creating a managed object without adding it to the context?
Am 24.01.2009 um 23:51 Uhr schrieb Jerry Krinock: On 2009 Jan 24, at 14:13, Ulai Beekam wrote: How do I programmatically create a managed object (with respect to some entity in a particular context of course) but WITHOUT adding it to the context? I do not believe this is possible. A managed object without a context would not be managed. It is possible. Just pass nil as context: NSEntityDescription *entityDescription = [[theManagedObjectModel entitiesByName] objectForKey:@EntityName]; NSManagedObject *newObject = [[NSManagedObject alloc] initWithEntity:entityDescription insertIntoManagedObjectContext:nil]; The reason I want this is to be able to take that new managed object and add it via an NSArrayController (one that is connected to said context of course) instead. That way I will get all the benefits of the array controller, such as the newly inserted object getting selected in a table view that is bound to the array controller. You indicated in your first sentence that you know which managed object context the new object is destined for. I don't see the problem. Just insert it when you create it. I am using the method shown above when I don't want to insert an unfinished object. I.e. I have some sort of input form and want to bind the fields to my model's properties. But I don't want the new object to show up in any list view before data input is finished. So I create a new object that is not yet part of the managed object context and insert it after editing is finished. (You also need some place to store that object that you can bind to while editing; like a property of a window controller or such.) Andreas ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Creating a managed object without adding it to the context?
Am 25.01.2009 um 00:51 Uhr schrieb Ulai Beekam: My problem is that when I add directly to the managed object context, the item is not getting selected automatically in the array controller (by the way, I see the selection visually by means of the table view bound to the array controller) even if I have checked the select on insert checkbox on the array controller in Interface Builder. IIRC that happens when you insert the object directly into the managed object context without going through the array controller. I.e. by using NSManagedObject's -initWithEntity:insertIntoManagedObjectContext: instead of NSArrayController's -addObject: That's why I asked about this. But yes, when I come to think about it now as you say it, it kind of makes little sense to be creating a managed object that is not in the context. I beg to differ. :) See my previous post. Andreas ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: addSubview bottleneck
Am 24.01.2009 um 23:29 Uhr schrieb Twisted Theory:
I am constructing an application that draws rooted trees (graphs
with a
distinguished 'first' node) by creating an NSView subclass for each
node and
adding them as subviews of the graph view.
There is a big problem when the number of nodes in the tree is
large: adding
the subviews takes a very long time. For example, one of my trees has
around 60,000 nodes. It takes about 1.5 seconds to create all of the
objects, and about 200 seconds to add them as subviews.
Can anyone suggest a way to speed the process up? Multithreading
didn't
really improve matters.
Yes. Do not use subviews. Instead create your own view subclass and
use cells.
Each node keeps track of its child nodes in an NSArray, and the
subview
addition is recursive:
- (void) add
{
NSAutoreleasePool *autoreleasepool = [[NSAutoreleasePool alloc]
init];
[[parent superview] addSubview:self];
for (Vertex * child in children)
[child performSelectorOnMainThread:@selector(add)
withObject:nil
waitUntilDone:NO];
//[child add];
[autoreleasepool release];
}
Why do you create a new autorelease pool here? Since this method seems
to be called once for each node, you will create an additional
autorelease pool every time. Also I don't see any objects created here
that would go into that pool. So, unless I missed something important,
you are creating and destroying 6 additional, totally useless
objects.
But you can get rid of this anyway if you use a custom view and just
use your model for drawing in -drawRect:. No need to duplicate that
hierarchy.
Andreas
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
