I believe the $() syntax is a bash extension not universally supported
by all /bin/sh implementations.
It's actually Korn Shell, but you have a good point on older systems
that default to Bourne Shell, back ticks it is.. :-)
___
mythtv-dev maili
On Wed, 2005-11-23 at 20:48 -0700, David Snider wrote:
> Probably just a typo with your pipe, but I would write the shell code
> as test x$(which sdl-config 2>&1 | grep -v ^no) != x. I like the $
> ( ) syntax rather than the back ticks ` ` as it is a little easier to
I believe the $() syntax
So far the contenders are test x`which sdl-config | 2>&1 grep -v
^no` != x and a shell function that implements a properly quiet
which. David Snider leans towards the first and I lean towards the
second.
Probably just a typo with your pipe, but I would write the shell code
as test x$(whic
On 23-Nov-05, at 7:27 PM, Nigel Pearson wrote:
test expects one word on either side of the != so all that "no
sdl-config in ..." crap gives it heartburn.
Aah. Common scripting problem. Easy to fix. Try this:
if test "x`which sdl-config`" != x ; then
Yup, I know that one too but you n
In other words, you can't be in myth-svn and have update find all the
lower directories and update them. You'd have to go into each and do
separate updates which opens you to the small but not zero chance of
mismatching revisions.
"svn update --revision ABCD URL" would eliminate any ch
On 23-Nov-05, at 4:23 PM, David Snider wrote:
IMHO I don't think that the grep option is too ugly, it should work
across most platforms doesn't require additional manual coding.
Agreed. I think you'd also have to do the stderr /dev/null redirection
for whiches that send error messages there.
The problem is with line 325 of mythplugins/configure:
if test x`which sdl-config` != x ; then
test expects one word on either side of the != so all that "no sdl-
config in ..." crap gives it heartburn. Down at line 691 theres a
which that redirects stderr to /dev/null. I assume that is
Just an FYI, the latest osx-packager script you posted here works
very well for me, thanks.
On Nov 23, 2005, at 12:04 AM, George Nassas wrote:
On 23-Nov-05, at 1:23 AM, Nigel Pearson wrote:
I updated the script to do all the checkouts in one shot.
Indeed. Looking good, but I think
On 23-Nov-05, at 1:23 AM, Nigel Pearson wrote:
I updated the script to do all the checkouts in one shot.
Indeed. Looking good, but I think we can do better.
This version will always do a svn co, and never an update.
I suspect it should still try and update if the source dirs
are alread
Thanks for this update. It has a few minor problems with already
checked out code, but I will commit most of your changes tonight.
True. And your "get the latest revision" trick isn't bad,
although I think I would prefer to do it via an svn info.
svn info only operates on local c
One remaining issue is the behaviour of the retail OSX which
script. It always gives output but the plugins configure script
depends on it being silent when a file isn't found. If, like me,
you've got DarwinPorts then you can get a nicer one in /sw/bin
which is why I have that added to th
On 21-Nov-05, at 1:57 AM, Nigel Pearson wrote:
True. And your "get the latest revision" trick isn't bad,
although I think I would prefer to do it via an svn info.
svn info only operates on local copies so you'd have to check out the
latest and then do an info to find out what you got
Sure, a clueless or malicious developer could commit intentional
breakage
Well, it is more careless than malicious, but that does happen :-)
...
Point is mixing revisions is wrong.
True. And your "get the latest revision" trick isn't bad,
although I think I would prefer to do
On 20-Nov-05, at 9:43 PM, Nigel Pearson wrote:
but I don't think it guarantees that the resultant code
will interoperate. r1234 of mythtv might not talk to r1234
of mythplugins, and the situation might not change until r2345.
Since svn revision numbers are repository-wide it does guarantee
co
Unfortunately it doesn't explicitly say if the volume is case-
sensitive and I'd rather not hardcode which filesystems are. Then
it occurred to me I could detect sensitivity by touching a file
under one name and testing for its existance under another. Simple
and it works great.
I was unde
On 20-Nov-05, at 10:41 PM, David Snider wrote:
Commenting on another post, to detect the filesystem on a Mac OSX
system, you would use the diskutil command.. for example:
diskutil info
Unfortunately it doesn't explicitly say if the volume is case-sensitive
and I'd rather not hardcode which
My exact shell command is osx-packager -verbose -clean -plugins ALLAll compiles well, any errors are probably scrolled off of the screen.. I post nightly builds to my website @ www.thesniderpad.com under the downloads section.. You can check a few out to see if I am missing something..Commenting
I imaged just the plugins for performance reasons.
Understand, though that is probably a reason
to _not_ have automagic disk images created :-)
...
Enabling the backend works and is probably a good idea if it's getting
time to start packaging the backend for OSX. Is anyone working on
On 20-Nov-05, at 7:27 PM, Nigel Pearson wrote:
Thanks for the improvements to the script -
automating it does have some merit. Some discussion:
1) This checkout problem will probably be fixed soon.
Not sure of the merit of adding this much code in.
Sounds like you're given a good r
Anyway, on the build front, David Snider turned me on to a really
great trick where one can create a case-sensitive filesystem for just
the mythplugins source.
Yes, I have been using the same trick. A 1GB image seems about
right for the third party dependencies and whole source tree.
TVRec::GetScreenGrab() is a static function and could be moved to
the preview generator.
Until we get a "proper" solution,
I have just put this in previewgenerator.cpp :
#ifdef USING_BACKEND
...
#else
VERBOSE(VB_IMPORTANT,
"Backend code not compiled in - cannot generate p
Add new options for selecting particular branches, tags or revisions.
A few tweaks to get this to work on my machine:
1) I suspect only a rarified group will run svn out of
/Volumes/Users/nigel/bin/svn ;)
Do'h. Yes, we badly need to put auto path detection in there.
(or, maybe just c
On 20-Nov-05, at 5:24 PM, David Snider wrote:
Interesting, I never noticed that before, just as an FYI, my Mac Mini
that I'm using for compiles seems to work fine and it is using the
/usr/bin/which
...
$ which blah
no blah in /bin /sbin /usr/bin /usr/sbin
Yup, that's the problem. test expect
On Nov 20, 2005, at 3:11 PM, George Nassas wrote:
On a similar note, which "which" are people using? I find that the
default Apple/Darwin/bsd one in /usr/bin gives a chatty multiword
message when it doesn't find the file and this interacts badly with
configure statements like test x`whic
On 20-Nov-05, at 4:39 PM, David Snider wrote:
One note, would it be hard to add some intelligence about where the
svn binary is? Perhaps have it look in /sw/bin, then /usr/bin, then
/usr/local/bin, or even just do a "which svn"? Mine happens to be
/usr/local/bin/svn...
On a similar note, w
The script worked great for me, I used the following options:
$ ./osx-packager.pl -clean -plugins ALL -verbose
One note, would it be hard to add some intelligence about where the
svn binary is? Perhaps have it look in /sw/bin, then /usr/bin, then /
usr/local/bin, or even just do a "which svn
On Nov 19, 2005, at 7:55 PM, George Nassas wrote:
On 19-Nov-05, at 3:21 PM, Geoffrey Kruse wrote:
One thing I would like to add is the ability to build a self
contained jobque / commflag app bundle that could be run on the mac.
It doesn't sound very hard but I'm wondering why you'd want to
On 19-Nov-05, at 3:21 PM, Geoffrey Kruse wrote:
One thing I would like to add is the ability to build a self contained
jobque / commflag app bundle that could be run on the mac.
It doesn't sound very hard but I'm wondering why you'd want to do this.
Is your mac way faster than your backend PC
On 19-Nov-05, at 4:41 PM, Daniel Kristjansson wrote:
TVRec::GetScreenGrab() is a static function and could be moved to
the preview generator.
But as an aside I never intended the --disable-backend and
--disable-frontend params to be used in perpetuity. They were only
I noticed about GetScreen
On Sat, 2005-11-19 at 14:49 -0500, George Nassas wrote:
> On 19-Nov-05, at 7:46 AM, [EMAIL PROTECTED] wrote:
> With these changes I'm able to to complete a checkout and partial
> build. Things go awry when creating libmythtv because the preview
> generator requires TVRec::GetScreenGrab but that f
Nice work, I think it is all working after these changes (except for
not building because of an off backend #define). One thing I would
like to add is the ability to build a self contained jobque /
commflag app bundle that could be run on the mac. I have looked
through the script but I c
On 19-Nov-05, at 7:46 AM, [EMAIL PROTECTED] wrote:
Add new options for selecting particular branches, tags or revisions.
A few tweaks to get this to work on my machine:
1) I suspect only a rarified group will run svn out of
/Volumes/Users/nigel/bin/svn ;)
2) handle the case where no revisi
32 matches
Mail list logo