Re: qt-peer thread issue
> "Donald" == Donald Gim <[EMAIL PROTECTED]> writes: Donald> I'm working on porting "Jamvm+Classpath+Qt/Embedded. Donald> Do I need any configuration on the Qt(4.4.3) or Donald> Classpath(0.97.2) to avoid this message? My impression is that the Qt peers are not completely robust. AFAIK they have not been worked on in quite some time. So, I would say that chances are good that you've run into a bug. Tom
Re: trouble w/ synchronized close method in PushbackInputStream
> "Robert" == Robert Dodier <[EMAIL PROTECTED]> writes: Robert> PushbackInputStream.close is synchronized. Apparently there is a Robert> deadlock: if I cut out the "synchronized" keyword, the program runs Robert> to completion. I am guessing that there is a thread sleeping in read() Robert> or something like that; I didn't investigate further. Could you file this in bugzilla? That way at least it won't get lost. Or, send a patch :-) Tom
Re: Fwd: Missing ReleaseArrayElements
> "Robert" == Robert Lougher <[EMAIL PROTECTED]> writes: Robert> Sorry to nag about this, but it would be good to get this applied. Robert> Without it, the Swing Demo leaks like a sieve on JamVM :) Funny -- by coincidence I was just prepping this when I got this message :-) I checked it in. It is simpler if you send a ChangeLog entry as well. Tom
[commit-cp] classpath ChangeLog native/jni/gtk-peer/gnu_jav...
CVSROOT:/cvsroot/classpath Module name:classpath Changes by: Tom Tromey 08/10/17 18:52:30 Modified files: . : ChangeLog native/jni/gtk-peer: gnu_java_awt_peer_gtk_CairoGraphics2D.c Log message: 2008-10-17 Robert Lougher <[EMAIL PROTECTED]> * native/jni/gtk-peer/gnu_java_awt_peer_gtk_CairoGraphics2D.c (Java_gnu_java_awt_peer_gtk_CairoGraphics2D_cairoDrawGlyphVector): Release 'fonts'. CVSWeb URLs: http://cvs.savannah.gnu.org/viewcvs/classpath/ChangeLog?cvsroot=classpath&r1=1.9731&r2=1.9732 http://cvs.savannah.gnu.org/viewcvs/classpath/native/jni/gtk-peer/gnu_java_awt_peer_gtk_CairoGraphics2D.c?cvsroot=classpath&r1=1.26&r2=1.27
[commit-cp] classpath ChangeLog tools/gnu/classpath/tools/j...
CVSROOT:/cvsroot/classpath Module name:classpath Changes by: Tom Tromey 08/10/16 17:20:24 Modified files: . : ChangeLog tools/gnu/classpath/tools/jar: Creator.java Indexer.java WorkSet.java Log message: * tools/gnu/classpath/tools/jar/WorkSet.java (initSet): Use foreach. Change argument type. (WorkSet): Change argument type. * tools/gnu/classpath/tools/jar/Indexer.java (indexJarFile): Use foreach. * tools/gnu/classpath/tools/jar/Creator.java (writeCommandLineEntries): Use foreach. (getAllEntries): Likewise. CVSWeb URLs: http://cvs.savannah.gnu.org/viewcvs/classpath/ChangeLog?cvsroot=classpath&r1=1.9730&r2=1.9731 http://cvs.savannah.gnu.org/viewcvs/classpath/tools/gnu/classpath/tools/jar/Creator.java?cvsroot=classpath&r1=1.8&r2=1.9 http://cvs.savannah.gnu.org/viewcvs/classpath/tools/gnu/classpath/tools/jar/Indexer.java?cvsroot=classpath&r1=1.7&r2=1.8 http://cvs.savannah.gnu.org/viewcvs/classpath/tools/gnu/classpath/tools/jar/WorkSet.java?cvsroot=classpath&r1=1.2&r2=1.3
Re: FW: Help a noob?
> "Twisti" == Christian Thalinger <[EMAIL PROTECTED]> writes: Twisti> You don't need to write that one, the shipped one works. I compile GNU Twisti> Classpath and CACAO on IRIX for a long time: Do you have a configure change that didn't make it upstream to Classpath? Just curious :) Tom
Re: FW: Help a noob?
> "Geoffrey" == Greene, Geoffrey N <[EMAIL PROTECTED]> writes: >> But that's weird. This is supposed to be a REPLACEMENT for java...why >> would it need javac? How can I get around the need for javac? And >> why is it needed anyway? Basically, all the 1.5 compilers in existence are written in java. So, you need a java runtime to run them. A simple way to do a bootstrap on a new machine is to build the .class files on a working machine, then copy them over. An even simpler way is to see if somebody already has them built and will make them available for download :-) Once you have the Classpath .class files, you can get a VM running. Then you have enough to rebuild everything from scratch on the new machine. >> Also, I notice the "WARNING...in the code below: checking jni_md.h >> support... configure: WARNING: no" jni_md.h is built in a funny way. Basically we look for a pre-made one in $srcdir/include that matches the current CPU/OS combination. So, to do a new port you would have to write this file. This is not hard, ordinarily. Tom
Re: Savannah has Mercurial!
> "Andrew" == Andrew John Hughes <[EMAIL PROTECTED]> writes: Andrew> For me the speed of diffs/status checks in CVS is appalling Andrew> and has a significant impact on my work habits. For diff I can't help you, but for status I recommend the cvs utils. I've used these upwards of ten years, I can't really picture using cvs without them. FWIW, I'm ambivalent about changing Classpath's VC system. We haven't had many branches historically (do we have any now?), bypassing one of cvs's big problems. The diff thing is annoying, though I guess I'm used to it. One consideration is whether there is Eclipse support for the new system. Assuming, that is, that any active Classpath developers are using Eclipse. Tom
Re: Integer.parseInt("+42") gives -42
> "David" == David Daney <[EMAIL PROTECTED]> writes: David> I am not an expert in this realm, but this may be small enough David> so that an assignment is not necessary. Yes, I agree, particularly because there is really only one fix for this -- delete the '+' code. Actually, the proposed patch doesn't seem to go far enough in that direction... AFAICT a leading '+' is not allowed at all; there's no reason to check for it specially. Tom
Re: How to build Classpath without Classpath
> "Gustavo" == Gustavo Guillermo Perez <[EMAIL PROTECTED]> writes: Gustavo> Hey I can do that, but what about rhe lib folder of gnu classpath Gustavo> libraries are not binary compatible from glibc to uClibc Yeah, the idea is to compile the class files only, then build the rest on the new platform. This may require build hackery. And, it may require a bit of extra hackery in the initial build, depending on what classes the desired VM overrides, etc. Anyway, whatever is required, it sounds simpler to me than the long processes outlined upthread. And, on the plus side, if the needed Classpath build tweaks are done cleanly, we can check them in for future users. Tom
Re: How to build Classpath without Classpath
> "Gustavo" == Gustavo Guillermo Perez <[EMAIL PROTECTED]> writes: >> ecj-binary -> gcj-4.3 -> classpath 0.9x (x >= 5) -> cacao >> 0.98+hg20071001 (or newer) -> ecj(-bootstrap) Gustavo> Yes very well, I having nightmares cause I can't have a Gustavo> binary build of gcc 4.3 or 4.4 without errors on uClibc, I Gustavo> guess cross compiling is a solution but gcc 4.2.X and gcc Gustavo> 4.1.X works natively. Having java on embedded systems is more Gustavo> painful than last year. It seems to me that it should be possible to compile Classpath on *any* machine with a working java environment, then copy the classes around. I haven't looked at the cacao or jamvm build systems in a while, but supporting something like this would make it simpler to do ports. For instance, we could just have a glibj.zip download somewhere -- bootstrap problem solved. Tom
note on readdir_r
While looking into http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24170, I noticed that Classpath is also using readdir_r. There is no reason to use this. Ulrich wrote a blog entry a while back explaining why: http://udrepper.livejournal.com/18555.html I'm not planning to fix this in Classpath, sorry. It should just be a matter of deleting the readdir_r case and configure checks, though. Tom
Re: Classpath Tools
> "Andrew" == Andrew John Hughes <[EMAIL PROTECTED]> writes:
Andrew> Does anyone know why our tools use -Xbootclasspath to load as in:
Andrew> exec @VM_BINARY@ -Xbootclasspath/p:"${tools_cp}"
Andrew> gnu.classpath.tools.javah.Main "$@"
Andrew> The option is not supported by either gcj or JikesRVM (although the
Andrew> latter also uses it to build). Can we swap it for just plain
Andrew> -classpath?
It could just be historical. It is hard to remember; I spent a little
time looking at cvs annotate and the mailing list archives, but I
didn't find anything.
BTW, gjavah.in is marked -kb, which is wrong.
Tom
[commit-cp] classpath ChangeLog java/util/LinkedHashSet.java
CVSROOT:/cvsroot/classpath Module name:classpath Changes by: Tom Tromey 07/08/15 16:29:10 Modified files: . : ChangeLog java/util : LinkedHashSet.java Log message: * java/util/LinkedHashSet.java (LinkedHashSet): Fix typo. CVSWeb URLs: http://cvs.savannah.gnu.org/viewcvs/classpath/ChangeLog?cvsroot=classpath&r1=1.9361&r2=1.9362 http://cvs.savannah.gnu.org/viewcvs/classpath/java/util/LinkedHashSet.java?cvsroot=classpath&r1=1.6&r2=1.7
[commit-cp] classpath/java/lang Class.java
CVSROOT:/cvsroot/classpath Module name:classpath Changes by: Tom Tromey 07/08/15 16:27:00 Modified files: java/lang : Class.java Log message: * java/lang/Class.java (internalGetFields): Use LinkedHashSet. CVSWeb URLs: http://cvs.savannah.gnu.org/viewcvs/classpath/java/lang/Class.java?cvsroot=classpath&r1=1.52&r2=1.53
[commit-cp] classpath ChangeLog
CVSROOT:/cvsroot/classpath Module name:classpath Changes by: Tom Tromey 07/08/15 16:25:41 Modified files: . : ChangeLog Log message: * java/lang/Class.java (internalGetFields): Use LinkedHashSet. CVSWeb URLs: http://cvs.savannah.gnu.org/viewcvs/classpath/ChangeLog?cvsroot=classpath&r1=1.9360&r2=1.9361
Re: J2SE compliant RE for Arm Linux
> "Søren" == Søren Boll Overgaard <[EMAIL PROTECTED]> writes: Kinda old thread, but wanted to clear this up... Søren> So far, none of the the platforms I've been looking at (except Søren> PERC, which is prohibitively priced) support J2SE 5.0. Thus, my Søren> question is, can you provide me with a pointer to a J2SE 5.0 Søren> JRE (GNU classpath based or otherwise) which will run on an ARM Søren> Linux platform? GCC svn supports 1.5 and, probably, runs on ARM. Tom
Re: URLLoader / RemoteURLLoader
> "Martin" == Martin Schlienger <[EMAIL PROTECTED]> writes: Martin> Well I am actually patching URLClassLoader. RemoteURLLoader Martin> seemed to provide some connection to http protocol. When is it Martin> exactly used? The various Loader classes are used to handle particular URL types. RemoteURLLoader is used for things other than file: and jar:. Martin> Caused by: java.lang.NoClassDefFoundError: gnu/java/net/protocol/http/Headers That class is part of Classpath. Did you remove it as part of your size reduction? All the files in that package are needed for http: to work. Martin> Is this sci-fi to work inside URLClasseLoader to load classes Martin> from http URL ? Nope, it should be pretty easy. Really you shouldn't have to modify URLClassLoader at all. Tom
Re: URLLoader / RemoteURLLoader
> "Martin" == Martin Schlienger <[EMAIL PROTECTED]> writes: Martin> -I noticed that URLLoader and RemoteURLLoader are also Martin> intended to load remote directories. May someone give me any Martin> hint to use these one? These are only for use as internal helpers of URLClassLoader. Martin> The very best would be that we launch our software with jamvm Martin> -classpath http: //url-to-my-classpath-classes/mypackage.jar You may want to look at JNLP. Martin> Another question: what is the gnu/java/net/loader/Load_ prefix Martin> for? there is no class that begins with Load_ . This is in cases a VM wants to add support for some URL type it provides. libgcj uses this to handle gcjlib: URLs. Tom
Re: gcjwebplugin usefullness? 64 bit plugin?
> "Dâniel" == Dâniel Fraga <[EMAIL PROTECTED]> writes: Dâniel> I tried very hard to use gcjwebplugin, but it will fail Dâniel> loading the applet I'm interested (from my bank site). Please file a bug report with the details, if you have them. Try running appletviewer from the command line, perhaps you'll get a nicer stack trace that way. Dâniel> Ok. The applet from classpath homepage works perfecly, but if I Dâniel> change to another site, Firefox crashes (segfault). Details here would also be good. Dâniel> The main problem today on 64 bit is the lack of plugin... So Dâniel> someone could help? Or explain why we still doesn't have a 64 bit Dâniel> plugin? The JDK is not completely open yet -- the libraries have not been released. We expect this will happen soon (before the end of June, last I heard). And the web plugin may not appear in the next code drop, either; we don't know yet. Tom
Re: Classpath 0.95 build issue
> "Audrius" == Audrius Meskauskas <[EMAIL PROTECTED]> writes: Audrius> This is becoming the permanent topic already. Maybe we could add Audrius> something to README or hackers guide? Sounds good to me. Dâniel, would you mind making a patch for this? Perhaps a change to the FAQ is the best bet. Tom
Re: Compilation Time | More questions ClassLoader
> "Martin" == Martin Schlienger <[EMAIL PROTECTED]> writes:
Martin> Current behavior acts like this, macro-algorithm:
[..]
Martin> Class loadClass(String className){
Martin> if (bootstrap) {
Martin> c= VMClassLoader.loadClass(className);
Martin> Basically all we have done is add a step at
Martin> bootstrap. Normally all classes to perform a
Martin> RemoteClassLoader.loadClass() can be loaded with the first
Martin> step VMClassLoader.loadClass(). In the end we want to have
Martin> only these classes in the local core API and the rest of the
Martin> API distant.
Yup. I'd suggest that instead of thinking of the solution in terms of
writing a class loader, think about it in terms of writing some code
that simply returns a byte[] that is the class contents.
Then call this downloading code from VMClassLoader.loadClass.
How you implement the downloading is up to you, but it turns out to be
pretty simple to re-use URLClassLoader for this... just treat the
.class file at the URL as a resource, not a class, and don't use the
loadClass family of calls on your delegate URLClassLoader at all.
Martin> We have done some basic tests and this seemed to work. However, when
Martin> trying with a more complete program, it appears that it throws
Martin> IllegalAccessException making me think that some security is added
Martin> when using the defineClass() on core API.
For this sort of a thing more details are needed. At least a full
stack trace. But if you follow the above this may be moot.
Tom
Re: Compilation Time | More questions ClassLoader
> "Martin" == Martin Schlienger <[EMAIL PROTECTED]> writes: Martin> - Is there any place where I can find something like a UML sequence Martin> diagram of classloading mechanisms. Nope, sorry. Martin> It seems there is a lot of back and forth between Martin> URLClassLoader, ClassLoader, VMClassLoader and all is a little Martin> bit confusing. Most of this is delegation. ClassLoaders are arranged in a dynamic hierarchy and each delegates loading to its parent before trying to load itself. This chain is rooted in VMClassLoader, which is not a standard class but rather the way we've chosen to represent the bootstrap loader inn Classpath . Martin> - Any suggestion to make a class loaded by another loader to have the Martin> same rights as the ones loaded with the bootstrap ( Martin> VMClassLoader.loadClass() ). Since we are loading some of the classes Martin> from glibj.zip with another loader that can access a remote file, but Martin> ATM we have some IllegalAccessException with some of these. Since we Martin> use the default ClassLoader's defineClass(), we may have to hack this Martin> one (no success ATM). The VM* classes in Classpath are a bit funny. They define an API but they don't use inheritance for implementations -- instead a given VM can replace one of the classes entirely and it is simply expected to conform to the defined API. So, you can change VMClassLoader to do whatever you like. For instance, in libgcj we delegate some operations to a URLClassLoader that we construct after VM startup. I'm not really sure what you mean about IllegalAccessException and the default defineClass. As I recall, ultimately all classes are defined via a method on VMClassLoader. Tom
Re: Compilation Time | More questions -Charsets
> "Martin" == Martin Schlienger <[EMAIL PROTECTED]> writes: Martin> Actually since we saw that support for different charsets were Martin> constantly added to GNU/Classpath, this may have been linked with Martin> GNU/Classpath and not jamvm. Martin> Indeed, gnu.java.nio.charset.Provider loads multiple default charsets. Martin> We modify this one as well. It seems that UTF8 , 8859_1 and US_ASCII Martin> are enough to boot. What makes a charset considered mandatory in Martin> GNU/Classpath? Oh, sorry, I thought you meant charsets being searched for at startup or something like that. But instead you seem to be talking about registration of the NIO charsets... is that right? There is no rule here as far as I know. We just register all the charsets we have. Adding a charset is done either because it is specified by the standard, or because someone needed it. I suppose registration could be done more lazily somehow. I haven't looked into it. Tom
Re: couldn't able to install classpath: java.lang.OutOfMemoryError
> ">" == r srinivasaraju <[EMAIL PROTECTED]> writes: >> but when it comes to executing "gmake", its giving some error. >> Internal compiler error >> java.lang.OutOfMemoryError: Java heap space There's not much we can do about this. Maybe you don't have enough memory. Maybe you need to increase the heap available to whatever JVM is running ecj. We don't know, sorry. Tom
Re: Compilation Time | More questions -Charsets
> "Martin" == Martin Schlienger <[EMAIL PROTECTED]> writes: Martin> Now we are investigating the charsets classes since for sure we don't Martin> need them all on our minimal system. JamVM tries to load a bunch of Martin> them when initializing and we would stick to one (8859 or UTF8 for Martin> example). You may give us a hint again about how to make jamvm/gnu Martin> classpath work with only one charset. I don't know about jamvm in particular. In most places, though, something like 'file.encoding' chooses the encoding to use. The locale may also affect this. So as a first attempt I would make sure that these things are set properly. Tom
Re: Compilation Time
> "Martin" == Martin Schlienger <[EMAIL PROTECTED]> writes: Martin> Each time we change the ClassLoader.java class, compilation Martin> lasts quite a long and I am not really convinced that Martin> everything need to be recompiled. I am quite a beginner with Martin> Makefile but it seems that no real check is done and Martin> everything is compiled by default. Martin> Maybe you can give me some tricks on that or tell me what method you Martin> use when hacking Classpath. IMNSHO, the best way to do incremental development on Classpath is to use Eclipse. It will only recompile the minimum necessary when you make a change. There a HOWTO for setting this up on the wiki -- it isn't as trivial as we would like, but it is doable, and once you have it set up you rarely have to tweak it. Tom
Re: [cp-patches] Patch: FYI: hard-code SAX fallback
> "Jeroen" == Jeroen Frijters <[EMAIL PROTECTED]> writes: >> I think for proper operation we have to remove the various XML service >> files from META-INF; see the earlier thread. But in order for this to >> work we also have to fix SAX to properly fall back to the Classpath SAX >> parser. Jeroen> Any update on this? I'm convinced that we must remove the Jeroen> META-INF/services files (at least for XML, but my guess is Jeroen> that we should remove all of them). I've been super busy with F7 work, but I will try to get to this soon-ish. I'll at least remove all the XML-related ones. The others probably need a code audit first. Tom
Re: Runtime.exec() race condition...
> "David" == David Daney <[EMAIL PROTECTED]> writes: David> I just committed a patch to libgcj that fixes a race condition that David> could allow file descriptors to leak to sub-processes: David> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31228 David> I don't have the time to port the fix to the default classpath VM David> implementation, but would encourage any interested classpath hackers David> to do so. Would you mind reporting this in bugzilla? I looked and Classpath does suffer from this bug. Tom
FYI: un dos-ify ChangeLog
Today I found out that ^M characters were added to the end of all the lines of ChangeLog. I undid this and added a local variable section for the benefit of those using Emacs. Tom
Re: Regressions from gcj 4.2 to 4.3 involving XML
>>>>> "Mario" == Mario Torre <[EMAIL PROTECTED]> writes: Mario> Il giorno ven, 09/03/2007 alle 12.06 -0700, Tom Tromey ha scritto: >> I think we ought to delete these service files from Classpath. Mario> Yeah, but please, not every service file, unless really needed. Mario> I have the java.util.prefs.PreferencesFactory there. The setting is set Mario> at configure time (usually the GConf backend), but in case of errors the Mario> fallback backend is always the FileBasedPreference one (that is, not Mario> what we want usually, but a safe fallback indeed). I was specifically only thinking about removing the various XML-related service files. That said, I think we should probably not use this approach to configure Classpath itself, at least not for areas where the user might need to be able to override the default. In those situations, maybe compiling in a default system property would be better. Historically we might have used Configuration.java for this sort of thing. But this brings up another subject: I think we should get rid of this class. In libgcj we already had to make a fake one so we could have system-independent .class files. That seems like a nice enough thing that Classpath should do it too. Tom
Re: Regressions from gcj 4.2 to 4.3 involving XML
> "Andrew" == Andrew Haley <[EMAIL PROTECTED]> writes: Andrew> I've been mystified why Xerces suddenly stopped working with this Andrew> release of gcj. I think I've found the root cause that triggered Andrew> this. Thanks for tracking this down. Andrew> Now, the handling of "core" URLs is fixed, so we do find Andrew> "/META-INF/services/javax.xml.parsers.DocumentBuilderFactory" in Andrew> libgcj, which is incompatible with Xerces. I think we ought to delete these service files from Classpath. It seems to me that, due to class loader delegation, we will always pick up any service file that is built in to glibj.zip. But, this isn't what we want. And, since we have fallbacks hard-coded into the factory classes, we don't actually need these service files anyway. Comments? Tom
Re: plugin for mozilla/firefox
> "Dima" == Dima Sorkin <[EMAIL PROTECTED]> writes: Dima> On a system with classpath installed, is there is a web browser plugin ? It depends on the VM. gij in particular doesn't use the system Classpath, it has its own copy. So, it would depend on how gij was configured and built. Tom
Re: Non-POSIX port
> "Chris" == Chris Cole <[EMAIL PROTECTED]> writes: Chris> I have the first cut of a port to an exotic platform (non-POSIX). Cool. Chris> I'd like to investigate the possibility of getting my changes Chris> adopted into the project. What's the best way to start that Chris> process? First thing is to get started on copyright assignment paperwork. Mark should probably do this but if he can't I will send you the form. Next is to start looking at how to merge it in. It may be a little difficult since Classpath doesn't yet have native code for a non-POSIX port (afaik). I'm sure we can figure out something sensible though. Maybe you could describe your port a bit more. What files did you add, where did you put them, do you use configure, etc. A non-POSIX port may also need some .java changes. ISTR a place or two where we have POSIX-y assumptions, like File or .so name mapping or something. Tom
builder cvs note
I noticed that the nightly scripts on builder.classpath.org still weren't using the CVS repository. I checked in all the little diffs that had accumulated in ~cpdev/Nightly/scripts, and then changed that directory to be a (pserver) checkout from the CVS repository. I've also set up the builder to run Mauve on gij (the whole reason I ended up looking at this in the first place :). This means test cycles may take a little longer now. Please make sure to check in all script changes, and please try not to have any local divergences in cpdev's script directory. This makes it simpler to make changes. Thanks, Tom
Re: [Jamvm-general] Compiling Jamvm 1.4.5 - errors
> "Twisti" == Christian Thalinger <[EMAIL PROTECTED]> writes: >> http://www.nabble.com/-cp-patches--RFC%3A-fix-annotation-inheritance-tf2471335.html#a6890685 Twisti> It would be nice to have this finally commited. IIRC, this was one Twisti> reason I stopped implementing annotations. Sorry for the delay on this. I forgot all about it. I will check in the patch shortly. Note that the Class part of the patch already went in at some point. Tom
Re: GJDoc
> "Andrew" == Andrew John Hughes <[EMAIL PROTECTED]> writes: >> Another plan, call it (d), is to wait until Sun GPLs their javadoc. Andrew> Is this on the cards? I'm pretty sure it will happen on the schedule (though it was somewhat more vague than a "schedule" :-) that Sun published. My recollection is, sometime this spring. Tom
Re: GJDoc
> "Andrew" == Andrew John Hughes <[EMAIL PROTECTED]> writes: Andrew> Do we have a plan for this? Nope. Andrew> c) move to an alternative solution. Andrew> Comments? For Fedora 7 I believe Tom Fitzsimmons is looking at sinjdoc. http://www.cag.lcs.mit.edu/~cananian/Projects/GJ/ Unfortunately gjdoc sort of dropped off the bottom of our priority list when looking at 1.5 :-( Another plan, call it (d), is to wait until Sun GPLs their javadoc. Tom
Re: Using Classpath on the Sun VM
> "Blake" == Blake Meike <[EMAIL PROTECTED]> writes: Blake> The follow-on, of course, is "is there a cookbook/suggested-toolset Blake> for porting apps?" Is there someone with experience doing this, who Blake> could share hints. (... do I have the wrong list?) There's some info on the wiki about common application problems, mostly involving use of 'sun.*' APIs. Usually I bring up a new application in the most obvious way: try it and then debug what doesn't work :-) Blake> Why is rt.jar/glibj.zip a single library? I'm not sure if there's a particular reason. Probably a combination of history, ease of setting up the build, ease of not having to figure out a sensible way to divide it, plus compatibility with (perhaps some old version of :) the JDK. Blake> Would it not make integration easier if it were divided into Blake> the Classpath jar/zip, and a Kaffe glue jar/zip? Jes askin'... Perhaps one of the Kaffe maintainers can reply. Tom
Re: Using Classpath on the Sun VM
> "Blake" == Blake Meike <[EMAIL PROTECTED]> writes: Blake> I have a largish application that runs well under the Sun JRE. I'm Blake> interested in porting it to Kaffe/Classpath. My first attempts have Blake> failed in some surprising ways, and I'm anticipating an extended Blake> struggle ;) :) Blake> I would like to know if anyone has tried using the Classpath class Blake> libraries on the Sun JVM. Not that I know of. Blake> Error occurred during initialization of VM Blake> java/lang/NoClassDefFoundError: java/lang/ref/FinalReference Offhand I would guess that this is a Sun-specific internal class that the VM wants. Since Classpath doesn't provide it you either have to modify the VM, or figure out what this class does and add it to Classpath. I'd imagine this is just the first of many such cases... Tom
Re: [PATCH] MD4/MD5 fixes
> "Andrew" == Andrew Haley <[EMAIL PROTECTED]> writes: >> Perhaps we should replace these uses with Integer.rotateLeft, for >> clarity. Andrew> In MD5? No! Please! Andrew> At least, not until this stuff gets inlined everywhere... Yeah. I thought about this more after I wrote the note. I think we should treat this idiom like "f != f" as a NaN test -- something that Classpath developers have to know or learn. At least for gcj inlining Integer methods is a pain for various reasons. Ideally we would fix this as well, though (at least for C++ ABI code). Tom
Re: [PATCH] MD4/MD5 fixes
> "David" == David Daney <[EMAIL PROTECTED]> writes: >> -aa = aa << 3 | aa >>> -3; >> +aa = aa << 3 | aa >>> (32 - 3); >> David> Java masks all shift counts to the number of bits being operated on. David> Your patches are a nop. David> I have not tried them, so I don't know if they quiet the warnings David> emitted by ecj. But if they do, it *might* be worthwhile applying David> them on those grounds. Historically we've heard about jikes warning for this code. But, the code is perfectly valid, and IMO: (foo << N) | (foo >>> -N) ... makes a nice "bit rotation" idiom. Perhaps we should replace these uses with Integer.rotateLeft, for clarity. Tom
Re: Can javax/swing/text/html/default.css be moved to resources/javax/swing/text/html/default.css?
> "Jeroen" == Jeroen Frijters <[EMAIL PROTECTED]> writes: Jeroen> Subject says it all. Can someone who understands the Classpath build Jeroen> infrastructure please have a look at this? It can be done with only a minor change to lib/Makefile.am. I can't get to it very soon though. I wonder whether we ought to move the various png files as well. Tom
Re: Mauve vs 1.5
> "Mark" == Mark Wielaard <[EMAIL PROTECTED]> writes: [ CharArrayWriter and Appendable ] Mark> Now this is of course easily fixed by using -1.5 so the compiler knows Mark> about covariant return types and makes all these tests that define Mark> classes that extend some Writer class compile again. Yes, let's do that... Mark> But now we have another problem. Shown by anything that has implements a Mark> retrofitted Comparable interface like Integer: Mark> 1. ERROR in gnu/testlet/java/lang/Integer/compareTo.java line 98: Mark> harness.check(zero.compareTo(o) == 0); Mark>^ Mark> The method compareTo(Integer) in the type Integer is not applicable for the arguments (Object) In this code, 'o' is just 'zero' cast to an Object. You could just change it to use compareTo(zero), but that doesn't test the same thing... You could change it to: harness.check(((Comparable) zero).compareTo(o) == 0); This will check using the raw Comparable and preserve the meaning of the test, more or less. Tom
Re: [cp-patches] FYI: Add Enum.finalize()
> "Twisti" == Christian Thalinger <[EMAIL PROTECTED]> writes: Twisti> 87% precent of the heap allocations with a finalizer have an empty Twisti> finalizer. The class responsible for this is java/util/zip/Inflater. I think we can simply delete Inflater.finalize. Tom
Re: Minimizing glibj.zip size
> "Trevor" == Trevor Harmon <[EMAIL PROTECTED]> writes: Trevor> It would be nicer if there were a configuration program in classpath, Trevor> like what the Linux kernel and BusyBox have, that lets you fine-tune Trevor> the components to include in the build. Yeah... it has been discussed many times on the list. There are some difficult constraints for a solution which could be checked in, and I think this has probably prevented anybody from implementing it. For gcj there is micro-libgcj: http://ulibgcj.sourceforge.net/ ... but this is more of an ad hoc reduction, not a general tool. Tom
Re: Minimizing glibj.zip size
> "Paul" == Paul Jenner <[EMAIL PROTECTED]> writes: Paul> What about lib/standard.omit? Paul> It would have to manually edited and the dependencies sorted out by the Paul> person building but I thought it should work. Yeah, you can add things to standard.omit; sorry I didn't mention this. However this file only controls what files are passed to javac on the command line -- typical java compilers will also read (and generate code for) whatever other classes might be needed. So, this method may pull in more code than you would like. Tom
Re: Minimizing glibj.zip size
> "Trevor" == Trevor Harmon <[EMAIL PROTECTED]> writes: Trevor> Is there some way of configuring classpath so that it Trevor> doesn't build and install these non-essential packages? No, there's not a standard built-in way. Trevor> Currently I'm working around the problem by manually deleting Trevor> files using "zip - d ..." and was hoping for a better way. That sounds reasonable enough. There's also tools for trimming class files & jars for embedded deployment, though I don't know whether any are free software. Tom
Re: GNU Classpath 0.93 "Dreamland" released
> "Dâniel" == Dâniel Fraga <[EMAIL PROTECTED]> writes: Dâniel> directory does not exist: ../vm/reference:..:../external/w3c_dom:../external/sax:../external/relaxngDatatype:../external/jsr166:.:: Dâniel> Strange, because all the above directories exists. Yes, that is odd. You aren't on Windows, are you? That could explain why ecj would treat the above as a single directory and not a list of directories. Tom
Re: GNU Classpath 0.93 "Dreamland" released
> "Dâniel" == Dâniel Fraga <[EMAIL PROTECTED]> writes: Dâniel> Thanks. But can i use gcjx? I use Linux from scratch, so I have Dâniel> to install separately. Which is better? ecj or gcjx? Thank you. Use ecj. The gcjx project has been cancelled. Tom
RFC: cvs rm'ing some ancient files
While doing a recent classpath->libgcj merge, I found a number of things in Classpath CVS which are very old and, AFAICT, unused. I'm going to remove them. Please speak up if you need any of this. I'll defer the deletions a couple days to give you a chance to respond... * test * testsuite Both unused and obsoleted by mauve anyhow * compat Unused, ancient * native/testsuite Old guile stuff? * native/vmi Old native code glue stuff Tom
Re: Generics Merge Done, NOW please go nuts!
> "Andrew" == Andrew John Hughes <[EMAIL PROTECTED]> writes: Andrew> HEAD is now generics ready. Please test and let me know Andrew> about anything that broke ASAP. We're open for business again, Andrew> so, as Mark says, 'please go nuts!' Thanks for doing these merges, and for all your work keeping the generics branch up-to-date these past two years. I believe when I started I had the naive idea that we wouldn't need to update the branch... and now I'm especially happy you stepped in to handle this :-) For Classpath hackers in general: on the generics branch we stuck to a "shallow" genericization, limited more or less to what Japi would see. I think it would be appropriate now to start on deeper work, i.e., moving the use of generics into the bodies of methods and not just on method signatures. Tom
Re: Please go nuts! -- Time For Generics
Andrew> So here's my first bash at a plan for the generics merge. Suggestions, Andrew> wild screams of despair, etc. much welcomed. Andrew> I'd like to have Sunday (or at least most of the day) as a no commit Andrew> day. It's usually fairly quiet anyway. During that time, I'll: Your plan sounds good to me. Andrew> a) tag the trunk and create a branch for maintaining 1.4-level stuff Is someone planning to continue work on that? If not a tag is sufficient; we can always branch later. Tom
Re: ASM and gnu.bytecode
Mark> We must make sure to properly document the way someone can grab Mark> the upstream sources in case we want to pull in bug fixes later. Tom> I'll handle this as part of the import. I've got the import working here. I'm going to wait for 0.93 to branch before committing it. Meanwhile, if someone wants it, I can mail the patch and the new sources. The patch also fixes a buglet I noticed -- we put the tool resources into glibj.zip, but not tools.zip. I think it makes sense to put them in the latter as well; it makes tools.zip more self-contained. Tom
Re: Migration to generics opens up exciting optimization opportunities from annotations
> "Ian" == Ian Rogers <[EMAIL PROTECTED]> writes: Ian> I'm interested in adding these optimization to the Jikes RVM, but Ian> clearly the annotations need adding to Classpath. I was wondering what Ian> kind of response this would get? I think it would be fine. My only concerns are the usual boring maintenance ones -- clear documentation for the annotations, how classpath hackers can avoid breaking things (without "too much" overhead), etc. Tom
Re: ASM and gnu.bytecode
> "Mark" == Mark Wielaard <[EMAIL PROTECTED]> writes: Mark> Thanks. I cannot find that as a source download. But it seems they have Mark> at least tagged their CVS with ASM_2_2_3 so we could pull the code from Mark> there. Updating to the latest release would also be an option for us. Moving our code from using 2.2.3 to using 3.0 or whatever probably would not be overly hard. Mark> We must make sure to properly document the way someone can grab Mark> the upstream sources in case we want to pull in bug fixes later. I'll handle this as part of the import. Tom
Re: ASM and gnu.bytecode
Mark asked me to send some more info about this: Tom> Ideally we could just import the ASM sources. I thought this idea was Tom> rejected, but I can't find a link. I'd like to revisit this, since Tom> this is the simplest way to solve the problem. The code is available from asm.objectweb.org. The license is here: http://asm.objectweb.org/license.html I would import whatever version currently works. Later we could import newer versions, as desired, and update our code to match. I'd import the code into classpath/tools/external (a new directory created for this purpose) and update the build scripts to match. I wouldn't rename the classes or anything like that. I would just import them using their upstream names. This is ok, I think, because the resulting classes would only end up in tools.zip -- not in glibj.zip. Users setting CLASSPATH could still have problems, but this doesn't seem like a major issue. Tom
Re: ASM and gnu.bytecode
> "Andrew" == Andrew Haley <[EMAIL PROTECTED]> writes: Andrew> Having gcj depend not only on ASM but also on a *specific version* of Andrew> ASM is intolerable. If gnu.bytecode will do the job, we should use Andrew> it. I suppose it would be best to import some bytecode library source into Classpath's tools subdirectory. Then we'd be insulated from upstream changes. Yay static linking! Ideally we could just import the ASM sources. I thought this idea was rejected, but I can't find a link. I'd like to revisit this, since this is the simplest way to solve the problem. It would also be easy to just put the right asm.jar on an ftp site somewhere, and have a script to download it. I already checked in a similar script (to gcc svn) to make it easy for GCC developers to get ecj. Failing that we can import GNU Bytecode and switch to that. FWIW the only reason this is my last choice is the amount of work involved. Tom
Re: building 0.93 on Mac OS X 10.3.9
> "jb" == Jean-Baptiste Nivoit <[EMAIL PROTECTED]> writes: jb> i thought i'd share my experiences building classpath 0.93 on a mac. Thanks! jb> [...] jb> hope this helps & this is the proper list to report this. I'd suggest either filing a bug report in the Classpath bugzilla, or submitting a patch :-) Tom
Re: SystemProperties secure?
> "Roman" == Roman Kennke <[EMAIL PROTECTED]> writes: Roman> We are using the SystemProperties class throughout the Classpath code to Roman> access system properties and avoid the security checks in Roman> java.lang.System. However, I come to think that this is no good the way Roman> it is. This class is public and nothing prevents use of this class from Roman> application code. As I recall things in gnu.classpath should not be available to application code. The system class loader, or something, has to enforce this. I'm having some trouble with the details but I know Jeroen knows the details here... Tom
Re: issue with DatagramSocket
> "Yohann" == Yohann Martineau <[EMAIL PROTECTED]> writes: Yohann> I have a problem with a DatagramPacket which receives incorrect Yohann> bytes. I made a capture with ethereal to see what is actually sent Yohann> over the network and I see a difference between what is captured with Yohann> ethereal and what is displayed with the following code: Could you send a self-contained test case? And indicate what platform, VM, and version of Classpath you are using? That might help. Tom
Re: Signals, sockets and threads
> "Guilhem" == Guilhem Lavaux <[EMAIL PROTECTED]> writes: Guilhem> 1) POSIX syscalls where originally written with no threads in Guilhem> mind. glibc uses a trick to store errno in a thread local Guilhem> storage using a special function. So the expected behaviour Guilhem> of storing errno just after the accept is true only in this Guilhem> case. If you use other type of threaded system then it may Guilhem> not be right (or for example if the library is buggy). Do these systems exist? I was under the impression that this errno behavior was required by the posix threads spec. But I admit I don't have a definitive URL :-) Tom
Re: Signals, sockets and threads
> "Twisti" == Christian Thalinger <[EMAIL PROTECTED]> writes: Twisti> Right. Robert told me that too, but why not using SIGHUP now Twisti> everywhere? Back then I think I thought that SIGUSR2 would be a better default choice. I don't think I had very good reasons for this though. Tom
Re: gkeytool error
> ">" == Haoyang Lin <[EMAIL PROTECTED]> writes: >> I checked the tools.zip. It contains no >> gnu/classpath/tools/keytool/Main. There is Main.java in >> /classpath-0.92/tools/gnu/classpath/tools/keytool/. What should I >> change to make gkeytoll working well? It is hard to say since we don't know much about how you built or installed. I would suggest starting at the beginning; make sure you built Classpath in the way that your VM expects, make sure you ran 'make install', etc. Tom
Re: Signals, sockets and threads
> "Twisti" == Christian Thalinger <[EMAIL PROTECTED]> writes: Twisti> I've implemented Robert's suggestion today and it seems to work (at Twisti> least on linux). I used SIGHUP for now. Why does GCJ use SIGUSR2 on Twisti> non-linux and freebsd OSes? Any good reason for that? Here's the code. Looks like we originally wanted SIGUSR2 everywhere but were working around a LinuxThreads bug. // The signal to use when interrupting a thread. #if defined(LINUX_THREADS) || defined(FREEBSD_THREADS) // LinuxThreads (prior to glibc 2.1) usurps both SIGUSR1 and SIGUSR2. // GC on FreeBSD uses both SIGUSR1 and SIGUSR2. # define INTR SIGHUP #else /* LINUX_THREADS */ # define INTR SIGUSR2 #endif /* LINUX_THREADS */ Tom
Re: gkeytool error
> ">" == Haoyang Lin <[EMAIL PROTECTED]> writes: >> I did : gkeytool >> errors: >> java.lang.NoClassDefFoundError: gnu/classpath/tools/keytool/Main >> <> >> glibj.zip and tools.zip exist in /usr/local/classpath/share/classpath/ >> What should I do? Please help. First thing is, check tools.zip to see if that class is actually contained in it. Perhaps you have an out of date install or the like. Tom
Re: Jboss/Java1.5 on HPPA in linux
> "Trevor" == Trevor Glen <[EMAIL PROTECTED]> writes: Trevor> Has anyone had any luck with a 1.5 vm and/or classpath on a Trevor> hppa linux box? What VM are you using? My impression is that hppa is probably the least support architecture out there... Beyond that, we've had success running jonas on gcj and other VMs. We're working on jboss right now. So, if it can't be done today, it will probably be doable soon... at least on the more supported platforms. Tom
Re: file locking problem in Java_gnu_java_nio_VMChannel_lock
> "Edwin" == Edwin Steiner <[EMAIL PROTECTED]> writes: Edwin> The attached patch fixes this case. It is still Edwin> not safe for len > signed 32 bit. Looks ok to me, I'm going to check it in. I wrote a ChangeLog entry for you -- but please, next time, send one with a patch. Thanks! Tom
Re: VMPlainDatagramSocketImpl still needed?
> "Robert" == Robert Schuster <[EMAIL PROTECTED]> writes: Robert> Casey said that he found Robert> gnu.java.nio.VMPlainDatagramSocketImpl to be obsolete. I Robert> quick glance at gnu.java.nio.PlainDatagramSocketImpl revealed Robert> that it does not use this class any more. In general I think if a VM class is no longer used from the main body of Classpath, it is obsolete and should be removed. Tom
Re: float printing in Formatter
> "Twisti" == Christian Thalinger <[EMAIL PROTECTED]> writes: Twisti> A few days back I tried to implement float printing in Twisti> Formatter. When looking at the existing int/long code, I saw Twisti> that toString() functions are used to print the value to a Twisti> buffer. That seems to be ok for such values, but handling Twisti> floats like this won't work, I guess. So the question is, Twisti> should we stick to our native float-printing implementation Twisti> (dtoa) or would it be better to have something, more flexible, Twisti> in Java? I suspect Formatter will need its own FP-formatting code. Tom
Re: Request for enlighment (newbie struggling with classpath and gij)
> "Zoltan" == Bartko, Zoltan <[EMAIL PROTECTED]> writes: Zoltan> I read somewhere on gcc's list Classpath 0.92 was integrated with the Zoltan> newest gcc snapshots, so I downloaded gcc-4.2-20060930, compiled and Zoltan> installed it. Ok. Zoltan> I executed gij -jar OmegaT.jar in the command line. gij Zoltan> complained: Cannot load AWT toolkit: Zoltan> gnu.java.awt.peer.gtk.GtkToolkit. You probably did not configure gcc correctly. At a guess you did not pass '--enable-java-awt=gtk'. Tom
Re: native methods in inner classes (long)
> "Jeroen" == Jeroen Frijters <[EMAIL PROTECTED]> writes: >> JNIEXPORT void JNICALL Java_OC_IC_natInit Jeroen> This looks like a bug in javah. [1] claims they fixed it. Thanks for finding this. I was worried that I'd have to change everything to check the InnerClasses attribute before doing JNI name mangling... would've been gross. Tom
Re: Debian dependencies for Classpath package
Chris> I'm especially surprised about the mysql dependency. Is all this really Chris> necessary? Not really; it depends on what kind of classpath you want to build. This looks like the setup for a maximal classpath with all the peers and stuff... though I don't see 'file' in there (needed for libmagic). Tom
Re: A question @Mark Wilaard (and other developer)
>>> I think three different OpenSource implementations would be the >>> worst case scenario, if Sun makes its Java OpenSource. FWIW, I tend to agree overall. Our experience merging libgcj with classpath was that the merge improved both -- sometimes the libgcj code was better, sometimes the classpath code was better. Parts of the class libraries are platform dependent or worth replacing depending on the VM or can have different implementations with different properties. But, most of the libraries are not like this. In these cases I think the most desirable result is a single implementation. And, there's nothing forcing an all-or-nothing approach; we can do as much or as little makes sense. That said, the details matter; cf the harmony situation. If Sun's terms are acceptable I would consider petitioning the FSF to let us freely merge code back and forth. Whether this will be the case, and whether Sun would be interested in the kind of disruption that this might entail, remains to be seen. Tom
Re: possible problem with gcjh (long)
> "Raif" == Raif S Naffah <[EMAIL PROTECTED]> writes: Raif> i'm attaching an OC$.java source and the generated header using Raif> gcjh. same behaviour --the dollar symbol is replaced by its Raif> unicode codepoint. Could you send the header file that the JDK's javah creates for this class? I'm curious about that. Tom
Re: possible problem with gcjh (long)
> "Raif" == Raif S Naffah <[EMAIL PROTECTED]> writes: Raif> $ gcjh -jni -classpath . -o OC_IC.h OC$IC gcjh is probably just seeing 'OC' here, unless you actually quoted the '$' on the command line. This explains why the gcjh-generated OC_IC.h is mostly empty -- the class OC has no native methods. Try OC\$IC and you will get a different result. Raif> $ /opt/jdk1.5.0_08/bin/javah -jni -classpath . -o OC_IC.h OC$IC I'd guess that javah is also reading inner classes... gcjh will probably never do this, I'm afraid, but we should probably modify the javah in Classpath to be compatible here. That shouldn't be hard. Could you file a PR? I don't have the JDK installed here so I didn't try to reproduce this... Raif> JNIEXPORT void JNICALL Java_OC_IC_natInit Hmm, gcjh generates Java_OC_00024IC_natInit. I'm surprised by the above, I (obviously I suppose :) would have picked the gcjh interpretation instead. Perhaps some experimentation is in order... what happens if you have a class with an explicit '$' in its name that has a native method? Tom
Re: Bootstrapping GNU Classpath on Fedora Core 5
> "Paul" == Paul Jenner <[EMAIL PROTECTED]> writes: Paul> IMHO the INSTALL file in the distribution should contain enough detail Paul> on requirements for people to be able to build. I'll take a look at Paul> updating that if necessary and maybe compare notes with you offline Paul> Alexander? Yeah, that would be handy. Updating it for all the distros when things change may be a bit painful, but it is worth trying. Paul> If mozilla devel dependency is not available on the system, I think it Paul> should be picked up at configure time and either error or disable plugin Paul> then, not have configure step succeed but compile fail later. I like configure arguments in situations like this to be 3-state: if nothing is specified, default to what is on the system, but if the feature is explicitly requested, error out during configure if the dependencies are not available. This lets us maintain "best effort" -- handy for builds in eclipse, for instance -- but also lets power users request a specific set of features and get a real failure if the request can't be satisfied. Tom
Re: Bootstrapping GNU Classpath on Fedora Core 5
> "Alexander" == Alexander Shopov <[EMAIL PROTECTED]> writes: Alexander> Is there a small tutorial on how to bootstrap GNU classpath Alexander> on Fedora Core 5? Not exactly, I'm afraid. In particular I don't think we document what devel packages you need. For general info there is the INSTALL file, and some stuff on the wiki. http://developer.classpath.org/mediation/ClasspathFirstSteps Mostly it is a matter of figuring out what devel packages you need. Once you've got those you should be in good shape. Currently I'm using an ordinary FC5 install on my main development machine, and I do builds every day, so I'm confident it can be done :-) Alexander> 4. ./configure --prefix=/home/ash/GNUCP --with-jikes --disable-gtk-peer IMO better to use ecj instead of jikes. Alexander> Another thing - which package provides the dependencies for Alexander> gtk-peer? gtk2-devel Alexander> gcjwebplugin.cc:47:19: error: npapi.h: No such file or directory yum install mozilla-devel Tom
Re: [cp-patches] RFC: merge ssl-nio-branch into generics-branch
> "Casey" == Casey Marshall <[EMAIL PROTECTED]> writes: Casey> The ssl-nio work is complete enough to the point where it can be merged Casey> into the generics branch. Now that 0.92 has been released, this is a Casey> good time to do that merge. I've accomplished three of the four goals I Casey> set out on this project, which were [1]: Congratulations! Casey> Unless there are any issues with this, I'll try to commit the patch this Casey> weekend. Sounds good to me. Tom
Re: glibj-generics-latest.zip corrupt?
> "Stuart" == Stuart Ballard <[EMAIL PROTECTED]> writes: Stuart> [EMAIL PROTECTED]:~/defunct/japigenerics$ wget Stuart> http://builder.classpath.org/dist/glibj-generics-latest.zip I downloaded this an ran 'unzip' on it, and it worked fine. I also looked at the file on builder, and it is ok there too. Tom
Re: Running JOnAS 4.7.4 that uses 1.5 features
> "Gary" == Gary Benson <[EMAIL PROTECTED]> writes: Gary> I wonder if this is what was not happening in gcj. The patch I wrote Gary> basically added the contents of java.endorsed.dirs to one of JOnAS's Gary> many classloaders. We implemented endorsed dirs specifically for jonas -- as I recall this was needed not only for jacorb but also for loading xerces/xalan. My recollection is that with the FC5 gcc (aka 4.1) we didn't need any special jonas hacks in this area... if I'm wrong, I'd like to know so we can fix it :-) Tom
Re: SIGSEGV caused by libgconf-2.so.4
Raif> which i presume is caused by "no" being generated in the file Raif> java.util.prefs.PreferencesFactory under resource/META-INF/services, Raif> instead of "gnu.java.util.prefs.FileBasedFactory". Try this. (untested) Tom Index: configure.ac === RCS file: /cvsroot/classpath/classpath/configure.ac,v retrieving revision 1.175 diff -u -r1.175 configure.ac --- configure.ac2 Aug 2006 19:32:45 - 1.175 +++ configure.ac4 Aug 2006 00:39:11 - @@ -89,7 +89,7 @@ AS_HELP_STRING([--enable-default-preferences-peer], [fully qualified class name of default Preferences API Backend])) DEFAULT_PREFS_PEER=$enable_default_preferences_peer -if test "$DEFAULT_PREFS_PEER" = ""; then +if test "$DEFAULT_PREFS_PEER" = "no" || test "$DEFAULT_PREFS_PEER" = yes; then DEFAULT_PREFS_PEER=gnu.java.util.prefs.FileBasedFactory fi dnl AC_SUBST(DEFAULT_PREFS_PEER)
Re: jboss-4.0.4
> "Rob" == Robert Lougher <[EMAIL PROTECTED]> writes: Rob> No, but it doesn't look difficult to implement. If I understand it Rob> correctly it seems to be as simple as just prepending the value of Rob> java.endorsed.dirs to the bootpath? You need to search the endorsed directories for .jar and .zip files and add each one to the boot class path. Tom
Re: jboss-4.0.4
> "Twisti" == Christian Thalinger <[EMAIL PROTECTED]> writes: Twisti> On shutdown I get a NPE: Twisti> 13:07:37,901 WARN [JRMPInvoker] Stopping failed Twisti> jboss:service=invoker,type=jrmp Twisti> java.lang.NullPointerException Twisti>at java.rmi.server.UnicastRemoteObject.unexportObject(UnicastRemoteObject.java:240) Could you file a PR for this? I don't know RMI very well (does anybody?) but this seems like a real Classpath bug to me. Tom
Re: Xara backend over cairo?
> ">" == êðõ <[EMAIL PROTECTED]> writes: >> The benchmark shows good perfomance than Cairo. >> http://www.xaraxtreme.org/about/performance.html >> But I don't know whether we can also use it for class path (though it is >> basically a vector rendering engine). If it is free software and does what we need for Java2D, I don't see why not. Naturally we would need a volunteer to start work on this :-) Tom
Re: classpath-0_92-branch created
> "Andrew" == Andrew John Hughes <[EMAIL PROTECTED]> writes: Andrew> I've synced the generics branch up to the branch point for 0.92. Just Andrew> apply any release branch patches both to generics and the release Andrew> branch. Are you going to hold off on any other head->generics merges until 0.92 ships? Tom
RE: Testing JDK bugs?
Jeroen> We obviously agree on this, but I'm trying to point out that things Jeroen> aren't always as "obvious" as one would think at first glance. How about distilling this wisdom into a section in the Classpath hacking guide? Tom
Re: OpenTasks?
> "Hari" == Hari Shreedharan <[EMAIL PROTECTED]> writes: Hari> I had mailed couple of weeks back on opentasks pending on GNU Hari> Classpath.Kindly tell me what are all the open tasks except for Swing, Hari> especially if any in networking and RMI. For RMI I would suggest looking in the Classpath bugzilla. If there is a known issue, it would be there. There may also be networking things in bugzilla. I don't recall. However there are also some new 1.5 additions in java.net that haven't yet been written. See the JAPI comparison pages on builder.classpath.org. Tom
Re: Mauve regressions (Was: [cp-patches] FYI: Component.setName() fix)
> "Mark" == Mark Wielaard <[EMAIL PROTECTED]> writes: Mark> The results aren't that. bad (all generated against cacao on fc5 x86-64) Mark> Lots of new PASSes. One new fail: I don't think new fails matter much. I tend to ignore them myself; I think they're also ignored for the purpose of baseline computation. Mark> FAIL: java.lang.InheritableThreadLocal.simple I updated Classpath and Mauve and ran this test case in my Eclipse workspace -- everything passed. Tom
Re: Mauve regressions (Was: [cp-patches] FYI: Component.setName() fix)
> "Francis" == Francis Kung <[EMAIL PROTECTED]> writes: Francis> Most of those tests use the Robot in some way, and I've Francis> noticed that tests using the robot require that the computer Francis> be dedicated solely to the tests... running it in the Francis> background while doing other work can cause the tests to Francis> return incorrect results. I thought the builder was running xvfb so that the AWT tests were isolated on their own X server... ? Tom
Re: VM interface changes for splash-screen support
> "Sven" == Sven de Marothy <[EMAIL PROTECTED]> writes:
Sven> So it seems AWT in 1.6 will include support for splash-screens.
Sven> This will require a little (optional) VM support, so here's a proposal
Sven> and some prototype code for this I'd like feedback on.
This seems pretty reasonable to me.
How is the splash screen taken down? Do we need another native
function for that?
Sven> struct splashscreenhandle
Sven> {
Sven> void *nativeWindowHandle;
Sven> void *nativeImageHandle;
Sven> };
This type should probably be given a name starting with cp_.
Tom
Re: Harmony, Suns work in progress OSS Java, GNU Classpath and the Java SE front
>> An other point is GNU Classpath at the moment. >> GNU Classpath 0.92 was ok. But all the snapshots after it have more >> and more bugs. Please file bug reports. Tom
Re: Is anyone working on InetAdderess.isReachable()?
> "David" == David Daney <[EMAIL PROTECTED]> writes: David> If not, I will probably give it a go. I haven't heard of anyone working on this. Tom
Re: Trouble using libxmlj in GNUJAXP classpath 0.90
> "Herve" == Herve Chang <[EMAIL PROTECTED]> writes: Herve> => building classpath from sources do not give me any xmlj.dll file Herve> needed for my usage (I'm using classpath on a windows mobile Herve> environment). If someone have already built the shared library Herve> xmlj.dll (or either xmlj.so file), I would kindly appreciate these Herve> infos :-) FWIW I haven't heard of anybody doing this. Herve> => If not, I am wondering if I am the first trying to use XML or Herve> XML/DOM in classpath :-) ? I don't recall doing it myself, but I am sure it has been done. Herve> a) IMO, the UnsatisfiedLinkError catch block is not at the right Herve> place. It is currently in gnu.xml.dom ImplementationSource class (as Herve> Tom said) which is a client class. It should rather be directly in the Herve> class that calls System.load() (gnu.xml.libxmlj.util.XMJ) something Herve> like It is a bit goofy, but I don't think it is incorrect. Herve> b) Morover, another problem is that, the success/failure of libxmlj Herve> loading is not considered. There should be a flag to indicate that Herve> either native loading has succeeded or not. Otherwise, there would be Herve> some ExceptionInInitializerError (as in the stacktrace), as some Herve> others class may rely on libxmlj. There shouldn't be any direct dependencies on this code -- if there are then that is a bug. The whole point, AIUI, is that this provider is optional. Herve> Caused by: java.lang.ExceptionInInitializerError Herve> at java.lang.VMClass.forName (Native Method) Herve> at java.lang.Class.forName (Class.java:161) Herve> at gnu.xml.dom.ImplementationSource. (ImplementationSource.java) Herve> at gnu.xml.dom.ImplementationSource. (ImplementationSource.java) Herve> at java.lang.VMClassLoader.loadClass (Native Method) Herve> ...25 more Herve> Caused by: java.lang.UnsatisfiedLinkError: Native library `xmlj' not Herve> found (as file `xmlj') in gnu.classpath.boot.library.path and Herve> java.library.path Herve> at java.lang.Runtime.loadLibrary (Runtime.java:763) Herve> at java.lang.System.loadLibrary (System.java:529) Herve> at gnu.xml.libxmlj.util.XMLJ.init (XMLJ.java:92) Herve> at gnu.xml.libxmlj.dom.GnomeDocumentBuilder. Herve> (GnomeDocumentBuilder.java) Herve> at gnu.xml.libxmlj.dom.GnomeDocumentBuilder. Herve> (GnomeDocumentBuilder.java) Herve> at java.lang.VMClass.forName (Native Method) Herve> ...29 more Ah, I see the problem. ImplementationSource is catching UnsatisfiedLinkError, but Class.forName is actually wrapping it in ExceptionInInitializerError. By my reading of the VM Spec: http://java.sun.com/docs/books/vmspec/2nd-edition/html/Concepts.doc.html#24237 (see point #9) I think this is a VM bug. In particular Error and its subclasses should not be wrapped in ExceptionInInitializerError. Tom
Re: Trouble using libxmlj in GNUJAXP classpath 0.90
> "Herve" == Herve Chang <[EMAIL PROTECTED]> writes:
Herve> So, as I want to use DOM/XML in classpath, I'd like to have your
Herve> comments/feedbacks about:
Herve> - Is the use of libxmlj mandatory? (according to my comments below,
Herve> seems to be yes)
I don't think it is. See below.
Herve> AFAIU, The DOM Boostrapping fallback class
Herve> gnu.xml.dom.ImplementationSource loads
Herve> gnu.xml.libxmlj.dom.GnomeDocumentBuilder which contains the
Herve> following static code:
Note the code in ImplementationSource:
try
{
Class t = Class.forName("gnu.xml.libxmlj.dom.GnomeDocumentBuilder");
acc.add(t.newInstance());
}
catch (Exception e)
{
// libxmlj not available
}
catch (UnsatisfiedLinkError e)
{
// libxmlj not available
}
In particular catching UnsatisfiedLinkError should guarantee that if
libxmlj is not found, then we can still proceed.
This definitely isn't ideal (IMO) but it seems like it ought to work
ok.
Tom
Re: classpath code coverage reports now online
> "Tony" == Anthony Balkissoon <[EMAIL PROTECTED]> writes: Tony> There are code coverage reports for classpath-mauve now at Tony> builder.classpath.org/~cpdev/coverage (there is a link from the main Tony> builder.classpath.org page). Thanks, this is really awesome. Tom
Re: [cp-patches] FYI: X peers
> "Roman" == Roman Kennke <[EMAIL PROTECTED]> writes: Roman> Mark proposed to put jars that are (optionally) needed for classpath on Roman> http://builder.classpath.org/ somewhere. What do others think? This seems like a reasonable idea to me. Tom
Re: [Gcjwebplugin-devel] Re: NSAPI/ OJI/ Applet plugin
> "Philippe" == Philippe Laporte <[EMAIL PROTECTED]> writes: .. replying to somewhat old email. Philippe> I've reviewed the outstanding PRs on this. Philippe> There was lots of activity in April/Eraly May around the Philippe> WebPlugin, but as far as Security is concerned, the class Philippe> lib seems not to have been touched, and so I guess none of Philippe> these PRs have been updated. The PRs have a pretty large granularity -- each one is a significant amount of work. Gary has been working steadily on the throwpoint test cases. My impression is that many of the checks are actually in place, so this doesn't translate directly to a lot of commits to Classpath. However, the tests are important in their own right, as they give us some assurance that we won't regress. All this work of his comes under the umbrella of a single PR -- hence no action there. We could certainly use help in this area. There's more to do, as I'm sure you know (I saw you were CCd on all the security PRs). Philippe> Is Security seen as vital for running Applets, in the usage contexts Philippe> currently being addressed? Right now we have this whitelist thing, which works nicely -- better than I expected :-). But, yes, security is vital for the long term. Tom
Re: Status of NIO socket implementation
> "James" == James Greenfield <[EMAIL PROTECTED]> writes: James> I'm struggling to find a decent summary of the status of the James> Classpath's NIO socket implementation. James> Can anyone summarize for me (or point me at the summary I've missed)? It works well enough to run Azureus. I searched in bugzilla and found 4 bugs, one of which is a feature request for an nio-related gcj optimization (i.e., not that relevant). Tom
javah
I recently rewrote 'gcjh' in java, including all the bits needed for 'javah'-like operation. I'd like to check it in. However, when I wrote it I used the ASM library. My preferred approach to this would be to check in the needed ASM sources (in tools/external), and build the library as part of the Classpath build. That way we wouldn't have any java-based external dependencies. On irc Mark suggested a --with-asm configure option. I could do that, instead, though it would make the eclipse-based build a bit weird. Any other ideas? I'm happy to send the code to anybody who wants it. It depends on the classpath getopt code. Tom
