SourceForge.net | 4 Jul 03:57

[ fink-Package Submissions-1370163 ] firefox 1.5

Package Submissions item #1370163, was opened at 2005-11-30 10:49
Message generated for change (Comment added) made by danielj7
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=414256&aid=1370163&group_id=17203

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
>Group: Added to Fink
>Status: Closed
>Resolution: Accepted
Priority: 5
Submitted By: Hanspeter Niederstrasser (nieder)
>Assigned to: Daniel Johnson (danielj7)
Summary: firefox 1.5

Initial Comment:
This is the first release of the new Firefox 1.5. 
Tested on 10.3, but should work on 10.4-T as well,
since I successfully tried a nightly build a few weeks
ago on 10.4-T (very minor changes since).

Unstable currently has 1.0.7, and I would like to have
that moved to stable on both branches, since all
recurring bugs w/ it seem to have been fixed (except
for bold text).

----------------------------------------------------------------------

>Comment By: Daniel Johnson (danielj7)
Date: 2006-07-03 21:57

Message:
Logged In: YES 
user_id=789616

OK! I just added it to 10.3 and 10.4. I didn't bother with
10.4-transitional since that's going away soon.

I removed the Architecture field since it's default value is
'all architectures.' I added 1000 to the revision in 10.4
(and changed GCC to 4.0) to allow for upgrading 10.3->10.4.
I also left the freetype219 dependency in the 10.4 package
to simplify maintenance and added Conflicts/Replaces on
firefox1.5 to the firefox 1.0.8 packages (and bumped
revision) to allow switching between them.

I tried building yelp-viewer-firefox against this and it
works perfectly even with just the firefox1.5-shlibs installed.

Thanks for the hard work!

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-06-22 14:21

Message:
Logged In: YES 
user_id=927770

Hopefully 68342th time is the charm...

New info file:

o moves lib/%N/chrome lib/%N/components lib/%N/greprefs
lib/%N/res into -shlibs (as per IRC)

o removes libnssckbi.dylib from Shlibs field because it's a
bundle

o BuildDepends on freetype219

The stuff for freetype219 in the info file and in the .patch
file are only needed on 10.3 in case someone is using
Apple's X11, so they can be removed for 10.4

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-06-21 14:47

Message:
Logged In: YES 
user_id=927770

1) using ac_add_options
--with-freetype-prefix=@PREFIX@/lib/freetype219, it builds
fine w/ xorg on 10.3 and 10.4/PPC.  However, I can't find
any evidence of runtime linking to ft219, and it only shows
up in the build logs while building in the same directory
that you crashed earlier.  All other dirs use
-I/usr/X11R6/include/freetype2 during the build.

I don't have the 10.3 disks, so I can't check to see if this
 works on Apple's X11 on 10.3.

If it can be tested by someone, I'll add 'BuildDepends:
freetype219' to the 10.3 version only.  I chose freetype219
over freetype2 just because it felt 'cleaner' to me, if that
really means anything (no -hinting for one).

2) good (nspr.m4)

3) pkgconfig added to BuildDepends and removed from Depends 

4) upgraded to orbit2 w/out any appreciable issues that I
can tell.  I was using orbit only because that's what the
build dependencies on the Mozilla page said was needed.

5) As far as I know, the dylibs not directly in
libs/firefox%N (like those under /components) and stuff in
/res are used by the browser only.  Do yelp or other
packages that Depends on firefox die if only firefox-shlibs
is installed?

6) fixed the order of -l flags in firefox-config

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2006-06-21 00:55

Message:
Logged In: YES 
user_id=535292

1. On 10.3 Apple X11, I needed to add

  ac_add_options --with-freetype-prefix=@PREFIX@/lib/freetype2

to the mozilla/.mozconfig file in firefox1.5.patch with freetype2-hinting-dev installed to get it
compiled. Looks like ./configure is pretty convoluted: the --with-freetype* flags control the $ft_*
and $FT2_* variables; --with-ft* does nothing.

That would mean an added dependency of 

  BuildDepends:freetype2-hinting-dev|freetype2-dev

would be needed. The analogous Depends:freetype2-hinting-shlibs|freetype2-shlibs does *not* seem
needed, since the places that -lfreetype are using it inherited from x11 things (-L/usr/X11R6/lib,
probably from gnome2 pkgconfig) and no -L/sw/lib/freetype2/lib is ever passed.

2. The .m4 looks great.

3. pkgconfig is probably only really needed at compile-time, and at most otherwise would be more
appropriate for the %N-dev package (where the .pc files are).

4. Is the old orbit still used anywhere? The only orbit flags I see are from gconf2, which brings the new
orbit2 instead. And even those flags appear in compiling but not in linking. Unless ffox dlopen()s the
orbit shared libs, maybe only need BuildDepends:orbit2-dev and no runtime dep. The flags probably only
appear at all due to pkgconfig artifacts.

5. Seems like some (maybe a lot? dunno) of the lib/ files in %N should be in %N-shlibs. Are the components/ and
plugins/ needed to render docs and do standard user interaction? Is chrome/ needed for display theming?
Are defaults/ needed for generic layout? Lots of res/ sounds pretty important too. These things may be
loaded at runtime (for example, libxpcom_core looks at plugins/). And they appear to be in a directory
that is named to avoid conflicts with past/future firefox versions, so there won't be filename collisions.

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-06-20 21:51

Message:
Logged In: YES 
user_id=927770

I've tried linking to freetype219 (w/ and w/out
pango1-xft2-219) in the past with xorg on 10.3, and always
got errors, even when forcing ft219.  I haven't tried with
Apple's X11.  I can try also just using freetype2 (2.1.4)
but I believe the whole FT issue came up when the Freetype
people f'ed up btw 2.1.7 and 2.1.8 and renamed symbols (in
FTC_Image_Cache_*) 

Dmacks: do you have any of the freetype(2|219) packages
installed?  Even if you do, a bunch of $*PATHs need to be
modified to get ./configure to see them.  The output from
./configure does say which variant of freetype-config it is
using.

Is there a way to know the version of Freetype that comes in
a particular flavor of X11?  'freetype-config --version'
gives a value that doesn't sync up with the Fink version scheme.

Furthermore, ./configure has freetype disabled.  huh.

----------------------------------------------------------------------

Comment By: Daniel Johnson (danielj7)
Date: 2006-06-20 20:32

Message:
Logged In: YES 
user_id=789616

The firefox-xpcom.pc changes are correct, but in
firefox-config it should be _xpcom_libs="-lxpcom
-lxpcom_core $_nspr_libs" since xpcom_core needs symbols
from $_nspr_libs which should therefore come later or else
won't be found.

As far as freetype is concerned, it builds correctly using
the freetype from Apple's X11 on 10.4, but the freetype from
10.3 is pretty old and may not be compatible. firefox1.5
might need an explicit freetype dependency, at least on 10.3.

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2006-06-20 19:15

Message:
Logged In: YES 
user_id=535292

Build fails on my 10.3 with Apple's X11:

gcc -o cairo-ft-font.o -c  -DMOZILLA_INTERNAL_API -DOSTYPE=\"Darwin7.9.0\" -DOSARCH=\"Darwin\"
-DBUILD_ID=0000000000 -I/usr/X11R6/include -I/usr/X11R6/include/freetype2 -I.
-I./../../libpixman/src -I../../../../dist/include/zlib
-I../../../../dist/include/libpixman -I../../../../dist/include/cairo
-I../../../../dist/include -I../../../../dist/include/nspr -I/sw/include -I/sw/include 
-I../../../../dist/sdk/include -I/usr/X11R6/include   -fPIC -I/sw/include  -I/usr/X11R6/include
-Wall -W -Wno-unused -Wpointer-arith -Wcast-align -Wno-long-long -no-cpp-precomp -fno-common
-pipe  -DNDEBUG -DTRIMMED -O2  -I/sw/include  -I/usr/X11R6/include -include
../../../../mozilla-config.h -DMOZILLA_CLIENT -Wp,-MD,.deps/cairo-ft-font.pp cairo-ft-font.c
cairo-ft-font.c: In function `_get_bitmap_surface':
cairo-ft-font.c:729: error: `FT_PIXEL_MODE_MONO' undeclared (first use in this function)

followed by a whole bunch more undeclared FT_ symbols. Do I have some unexpected package installed that's
causing ./configure to give unexpected results, or do I need a newer freetype version than Apple X11 provides?

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-06-20 09:45

Message:
Logged In: YES 
user_id=927770

Here's the new patch with xpcom_core and nspr.m4 fixes. 
Built and ran fine on 10.4/PPC and 10.3.

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2006-06-19 15:43

Message:
Logged In: YES 
user_id=535292

I don't think that the firefox build runs the autotools, so
.m4 are not used by firefox itself during the build, and
underquoting or other bugs in private .m4 files don't matter
at all. So I'm only concerned that the ones that are visible
to end users, the public .m4 actually installed (probably
%p/share/aclocal/ in %N-dev), get fixed.

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-06-19 14:52

Message:
Logged In: YES 
user_id=927770

Ack.  I got bitten by mails being sent to the spam folder.

dmacks: is it just nspr.m4 that needs to be patched?  Fix
has been filed for upstream trunk
(https://bugzilla.mozilla.org/show_bug.cgi?id=298457) and
it's several other files as well.

Danielj7: would the new firefox-xpcom.pc Libs: syntax be
"Libs: -L${libdir} -lxpcom -lxpcom-core"?

And _xpcom_libs="-lxpcom $_nspr_libs -lxpcom_core" for
firefox-config?  (mostly, does order matter?  I'm building
now to test, but I won't notice 3rd party linking errors in
building just firefox1.5)

----------------------------------------------------------------------

Comment By: Daniel Johnson (danielj7)
Date: 2006-06-12 18:06

Message:
Logged In: YES 
user_id=789616

1.5.0.4 builds and runs on 10.4/Intel.

However, there is an issue with linking other packages to
firefox1.5-(shlibs|dev). I changed yelp-viewer-firefox to
use this instead of the old firefox and got undefined symbol
errors. The missing symbols are in libxpcom_core.dylib which
isn't listed in any of the pkgconfig files or
firefox-config. Adding -lxpcom_core to the Libs: line of
firefox-xpcom.pc allowed yelp-viewer-firefox to build
successfully and it runs! -lxpcom_core also should be added
to the _xpcom_libs= line of firefox-config.

With these changes and dmacks' I think this package is ready
to go. Then it should be possible to start migrating
firefox-using packages to firefox1.5, which will allow them
to finally work on Intel. Yay!

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2006-06-12 13:43

Message:
Logged In: YES 
user_id=535292

The nspr.m4 installed here (and also in the existing firefox
package) has a bit of sloppy syntax that causes noise for
other packages when they use automake1.9. Could you please
patch:

AC_DEFUN(AM_PATH_NSPR,

to be:

AC_DEFUN([AM_PATH_NSPR],

to shut up the "underquoted definition" warning?

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-06-12 10:20

Message:
Logged In: YES 
user_id=927770

Updated to 1.5.0.4.  Builds and runs on 10.3 w/ the intel
changes by danielj7.

Is there anything else missing?

----------------------------------------------------------------------

Comment By: Daniel Johnson (danielj7)
Date: 2006-05-28 08:10

Message:
Logged In: YES 
user_id=789616

Turning XP_MACOSX back on was the first thing I tried to get
it to work. It was a while ago, and I don't remember what
happened, but it wasn't good. I went through the source
looking for all occurrences of XP_MACOSX and what it mostly
controls is the use of Carbon code for drawing, events,
loading native Mac plugins etc. Trying to mix that with X11
code would do Bad Things. The only possibly useful stuff
would be the Mac-specific browser migration code for Safari,
Camino, OmniWeb and such. That only gets compiled if
XP_MACOSX is set. However, it's probably not worth the
effort of disentangling it from the Carbon code. It looks
like the upstream developers never considered someone would
want to run FireFox under X11 on Mac OS X. :)

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-05-27 17:26

Message:
Logged In: YES 
user_id=927770

Great work Daniel.  In the original patch, there are several
places that undefine XP_MACOSX, particularly in
mozilla/configure.  I won't be able to check on this until
Monday for PPC, but can you check if adding back -DXP_MACOSX
to mozilla/configure and removing your changes also works?

Reason is that the original change in mozilla/configure goes
back to when I copied the patch from mozilla-1.4 (or earlier
maybe) and I just took it as a given that that was an
absolute requirement since then, so I wonder if some of
those changes can be undone now (I know some other changes
can't: anything in nsprpub causes a segfault in the file
dialog)...

----------------------------------------------------------------------

Comment By: Daniel Johnson (danielj7)
Date: 2006-05-27 12:15

Message:
Logged In: YES 
user_id=789616

Yes! I figured out how to get it to work on Intel!

There is some special x86 assembly code that needs to be
used on Darwin/x86 but only gets included if XP_MACOSX is
defined, which it isn't in this package. I changed two
#ifdef XP_MACOSX to #ifdef __APPLE__ and now it works!

I've included an updated patch file.

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-05-24 12:19

Message:
Logged In: YES 
user_id=927770

o Fixed the atk/libidl2 Depends
o Removed pkgconfig from -dev Depends
o /sw/lib/firefox1.5/xpidl from -dev links to
libIDL-2.0.dylib.  In fact, it links to dylibs from glib2,
libgettext3, and libiconv as well, so I added those to the
-dev Depends.[1]
o added -v to shell scripts

[1] although this is superflous because it depends on the
parent package, which already depends on those packages.

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2006-05-24 00:14

Message:
Logged In: YES 
user_id=535292

1.5.0.3 builds fine on intel (with change of GCC:3.3 -> 4.0), but dumps core while running. Same backtrace
as danielj7 on 2006-03-21 20:54

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2006-05-23 20:50

Message:
Logged In: YES 
user_id=535292

By eye, the renaming looks reasonable. I see you've added
explicit Depends to %N-shlibs, which is good, but you've
added atk1 and libidl2, which are just headers
packages--that's bad (BuildDependsOnly violation)--instead
of their associated -shlibs packages.

Does %N-dev really need Depends:libidl2-shlibs? Also,
pkgconfig is a compile-time tool, so I don't think you need
to Depends on it ever--BuildDepends is sufficient.

For sanity, we now recommend that -v be used in addition to
-e when a *Script field uses an explicit #!/bin/sh.

I'm test-building it on intel now...

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-05-23 13:35

Message:
Logged In: YES 
user_id=927770

Here's the latest 1.5.0.3.

Main changes:

o package is now firefox1.5, not just firefox
o added conflicts/replaces with firefox(-dev). 
conflicts/replaces is currently only used in -dev splitoff,
but is now also required for main package
o reduced the Patch complexity by adding to PatchScript to
replace @executable_path with %p/lib/%n
o changed relevant instances in the info file of the text
'firefox' to %n (or %N).
o added a whole lot more Depends to %n-shlibs because otool
-L says the dylibs are linking against files from these
other packages.
o fixed both places in patch to move from firefox to firefox1.5

To be done by reviewer:

- make sure the changes above are OK, particularly that no
instances of firefox -> %n/%N were missed
- test on ppc/10.4 and intel.  I now have a ppc/10.4 system,
but it'll be a while before I can get it building

Finally, mozilla.info in crypto/all-trees has to add
conflicts/replaces to firefox1.5 and firefox1.5-dev

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-05-19 13:26

Message:
Logged In: YES 
user_id=927770

I'll look into this next week after my talk.  Shouldn't be
hard, since we currently _remove_ versioning info from the
install path.  Just have to make sure to pick a good
numbering convention for the future and add it into the
patch.  Probably something like %n1.5 and avoid having to
change this for minor revisions (as the raw source does).

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2006-05-19 12:37

Message:
Logged In: YES 
user_id=535292

Regarding danielj's comment on 2006-03-21 20:54, it's also evident from the removal of the
libxlibrgb.dylib entry in the %N-shlibs Shlibs field that the binary layout has changed. So even if the
linking problem you had was due to some other cause, who knows what other compiled packages may exist that
do directly link to that .dylib, and that will break if firefox-shlibs is upgraded such that the file goes
away. By policy then, the -shlibs package (and the -dev that matches it) should be renamed and the files in
the new -shlibs renamed or moved so that the new and old -shlibs packages can coexist.

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-04-17 15:36

Message:
Logged In: YES 
user_id=927770

Thanks to danielj7, I now have 1.5.0.2 building on 10.3. 
Attaching first try info and patch files.

/me smacks SF for not allowing multiple attachments per post

----------------------------------------------------------------------

Comment By: Daniel Johnson (danielj7)
Date: 2006-04-14 14:27

Message:
Logged In: YES 
user_id=789616

Just to let you know, I tried building this with the new
1.5.0.2 sources, which is known to work on Intel Macs (I did
have to tweak the patch a bit to compile), and it still
crashes on launch in exactly the same way all previous
versions did. :(

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-03-28 17:44

Message:
Logged In: YES 
user_id=927770

Fixes the build on 10.4-T by using NoSetCXX: true.  Tested by rv8 on 10.4-T.  
Should also work on 10.3, but am still building.

----------------------------------------------------------------------

Comment By: Kevin Horton (rv8)
Date: 2006-03-28 11:13

Message:
Logged In: YES 
user_id=526775

It will not build on a G4 Dual, running OS X 10.4.5, with the 10.4-transitional 
tree, using XCode's gcc powerpc-apple-darwin8-gcc-4.0.1 (GCC) 4.0.1 
(Apple Computer, Inc. build 5247).  It fails like this:

g++-3.3 -I/sw/include  -I/usr/X11R6/include -fno-rtti -fno-exceptions -
Wall -Wconversion -Wpointer-arith -Wcast-align -Woverloaded-virtual -
Wsynth -Wno-ctor-dtor-privacy -Wno-non-virtual-dtor -Wno-long-long -
no-cpp-precomp -fno-common -fshort-wchar -pipe  -DNDEBUG -
DTRIMMED -O2 -fPIC -arch ppc -o libxpcom_core.dylib  pldhash.o 
nsCOMPtr.o nsComponentManagerUtils.o nsDebug.o nsID.o 
nsIInterfaceRequestorUtils.o nsINIParser.o nsMemory.o nsTraceRefcnt.o 
nsWeakReference.o nsGREGlue.o nsVersionComparator.o nsTHashtable.o 
nsGenericFactory.o nsXPComInit.o nsStringAPI.o  nsAtomTable.o 
nsAtomService.o nsByteBuffer.o nsCheapSets.o nsCRT.o nsDeque.o 
nsEmptyEnumerator.o nsEnumeratorUtils.o nsFixedSizeAllocator.o 
nsHashSets.o nsHashtable.o nsObserverList.o nsObserverService.o 
nsProperties.o nsPersistentProperties.o nsQuickSort.o nsRecyclingAllocator.o 
nsStaticNameTable.o nsStringEnumerator.o nsSupportsArray.o 
nsSupportsArrayEnumerator.o nsSupportsPrimitives.o nsUnicharBuffer.o 
nsVariant.o nsVoidArray.o nsTextFormatter.o nsTimelineService.o 
nsValueArray.o nsCOMArray.o nsArray.o nsArrayEnumerator.o 
nsHashPropertyBag.o nsAppFileLocationProvider.o nsBinaryStream.o 
nsByteArrayInputStream.o nsDirectoryService.o nsEscape.o nsFastLoadFile.o 
nsFastLoadService.o nsInputStreamTee.o nsLinebreakConverter.o 
nsLocalFileCommon.o nsMultiplexInputStream.o nsPipe3.o nsStreamUtils.o 
nsScriptableInputStream.o nsSegmentedBuffer.o SpecialSystemDirectory.o 
nsStorageStream.o nsStringStream.o nsUnicharInputStream.o 
nsNativeCharsetUtils.o nsLocalFileUnix.o nsCategoryManager.o 
nsComponentManager.o nsNativeComponentLoader.o 
nsStaticComponentLoader.o nsServiceManagerObsolete.o xcDll.o plevent.o 
nsAutoLock.o nsEnvironment.o nsEventQueue.o nsEventQueueService.o 
nsThread.o nsTimerImpl.o nsProcessCommon.o TimerThread.o 
nsProxyEvent.o nsProxyEventClass.o nsProxyEventObject.o 
nsProxyObjectManager.o nsProxyRelease.o nsAllocator.o 
nsVersionComparatorImpl.o nsConsoleMessage.o nsConsoleService.o 
nsDebugImpl.o nsErrorService.o nsExceptionService.o nsMemoryImpl.o 
nsTraceRefcntImpl.o nsInterfaceRequestorAgg.o nsStackFrameUnix.o 
xptcall.o xptiFile.o xptiInterfaceInfo.o xptiInterfaceInfoManager.o 
xptiManifest.o xptiMisc.o xptiTypelibGuts.o xptiWorkingSet.o xptiZipItem.o 
xptiZipLoader.o xpt_arena.o xpt_struct.o xpt_xdr.o 
xptcinvoke_ppc_rhapsody.o xptcstubs_ppc_rhapsody.o 
xptcinvoke_asm_ppc_rhapsody.o xptcstubs_asm_ppc_darwin.o 
nsDependentString.o nsDependentSubstring.o nsPrintfCString.o 
nsPromiseFlatString.o nsReadableUtils.o nsSubstring.o nsSubstringTuple.o 
nsString.o nsStringComparator.o nsStringObsolete.o nsAString.o 
nsObsoleteAStringThunk.o  -L/sw/lib  -Wl,-dead_strip -dynamiclib -
install_name /sw/lib/firefox/libxpcom_core.dylib -compatibility_version 1 -
current_version 1 -L../../dist/lib -lplds4 -lplc4 -lnspr4 -L/sw/lib -ldl -lm     
ld: warning -dylib_install_name /sw/lib/firefox/libxpcom_core.dylib not 
found in segment address table LD_SEG_ADDR_TABLE /sw/var/lib/fink/
prebound/seg_addr_table
ld: warning prebinding disabled because of undefined symbols
ld: Undefined symbols:
_fprintf$LDBLStub
/usr/bin/libtool: internal link edit command failed
make[3]: *** [libxpcom_core.dylib] Error 1
make[3]: Leaving directory `/Volumes/BU/sw/src/fink.build/
firefox-1.5.0.1-7/mozilla/xpcom/build'
make[2]: *** [libs] Error 2
make[2]: Leaving directory `/Volumes/BU/sw/src/fink.build/
firefox-1.5.0.1-7/mozilla/xpcom'
make[1]: *** [tier_2] Error 2
make[1]: Leaving directory `/Volumes/BU/sw/src/fink.build/
firefox-1.5.0.1-7/mozilla'
make: *** [default] Error 2
make: Leaving directory `/Volumes/BU/sw/src/fink.build/firefox-1.5.0.1-7/
mozilla'
### execution of /var/tmp/tmp.1.MxFuy0 failed, exit code 2
Removing build lock...
/sw/bin/dpkg-lockwait -r fink-buildlock-firefox-1.5.0.1-7
(Reading database ... 148636 files and directories currently installed.)
Removing fink-buildlock-firefox-1.5.0.1-7 ...
Failed: phase compiling: firefox-1.5.0.1-7 failed

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-03-28 09:44

Message:
Logged In: YES 
user_id=927770

 Fixed the malformed patch

----------------------------------------------------------------------

Comment By: Kevin Horton (rv8)
Date: 2006-03-28 08:57

Message:
Logged In: YES 
user_id=526775

I tried to build this on PPC/10.4-transitional, and it fails at the patch stage:

sed 's|@PREFIX@|/sw|g' </sw/fink/dists/local/main/finkinfo/firefox.patch | 
patch -p1
patching file .mozconfig
patch: **** malformed patch at line 48: diff -ruN mozilla-orig/browser/app/
mozilla.in mozilla/browser/app/mozilla.in

### execution of sed failed, exit code 2

============
Note: It fails in the same way with or without fink's sed-4.1.4-12 installed.

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-03-24 17:13

Message:
Logged In: YES 
user_id=927770

1.5.0.1 won't work on Mactel.  1.5.0.2 should, whenever
that's released.

http://wiki.mozilla.org/Mac:Intel

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-03-24 09:29

Message:
Logged In: YES 
user_id=927770

I can't do anything about the intel crash since I have no
way to test it :(

I can version the shlibs/includes into firefox-%v (in fact,
the patch currently _un_versions LIBDIR and INCLUDEDIR;
mozilla does the same thing), but I wonder if the build
crash is for a similar reason as the runtime crash; ie the
program built, but not correctly, so you can't run it or
link to its libraries.  It runs fine for me here on 10.3
(using it to type this in fact), but I have no way of
testing any of the other programs that BuildDepend on
firefox (too many dependencies and not enough space).

The API should not have changed between 1.0 and 1.5 as far
as I know, however, I'll ask around to double check.

Anyone with PPC/10.4 please try this, pretty please.

----------------------------------------------------------------------

Comment By: Daniel Johnson (danielj7)
Date: 2006-03-21 20:54

Message:
Logged In: YES 
user_id=789616

The patch file is bad. On line 4 it should read +1,43 not +1,44.

I haven't tested this on a ppc system, but on my MacBook Pro
I get

/sw/lib/firefox/run-mozilla.sh: line 424: 10400 Illegal
instruction     "$prog" ${1+"$@"}

with the following crash log:

Command: firefox-bin
Path:    /sw/lib/firefox/firefox-bin
Parent:  sh [10395]

Version: ??? (???)

PID:    10400
Thread: 0

Exception:  EXC_BAD_INSTRUCTION (0x0002)
Code[0]:    0x0000000d
Code[1]:    0x00000000

Thread 0 Crashed:
0   dyld                	0x8fe136e4
stub_binding_helper_interface + 18
1   <<00000000>> 	0x00000000 0 + 0
2   libxpconnect.dylib  	0x02d60a78 NSGetModule + 29498
3   libxpconnect.dylib  	0x02d5c2dd NSGetModule + 11167
4   libxpcom_core.dylib 	0x002f0c8a XPTC_InvokeByIndex + 246
5   libxpcom_core.dylib 	0x002f0df1 nsXPTCStubBase::Stub4() + 27
6   libxpconnect.dylib  	0x02d70e78 NSGetModule + 96058
7   libxpconnect.dylib  	0x02d7115c NSGetModule + 96798
8   libxpconnect.dylib  	0x02d6fb96 NSGetModule + 91224
9   libxpcom_core.dylib 	0x002d3c7c
nsComponentManagerImpl::AutoRegisterNonNativeComponents(nsIFile*)
+ 266
10  libxpcom_core.dylib 	0x002d4867
nsComponentManagerImpl::AutoRegisterImpl(int, nsIFile*, int)
+ 747
11  libxpcom_core.dylib 	0x002d49dd
nsComponentManagerImpl::AutoRegister(nsIFile*) + 137
12  libxpcom_core.dylib 	0x002a55c1 NS_InitXPCOM3_P + 1273
13  firefox-bin         	0x00002d70
ScopedXPCOMStartup::Initialize() + 62
14  firefox-bin         	0x00004431 XRE_GetBinaryPath + 4245
15  firefox-bin         	0x0000596a XRE_main + 2540
16  firefox-bin         	0x00002584 main + 40
17  firefox-bin         	0x00002502 start + 270
18  firefox-bin         	0x0000241d start + 41

I get the same crash with firefox 1.0.7 and mozilla 1.7.5 so
this is no worse. :) Firefox obviously can be built so that
it works on Intel Macs since the official Mac version does;
there must be something that has to be different during the
build process, but I don't know what.

Also, if I try to rebuild yelp-viewer-firefox after
installing this package, it fails with:

g++ -fno-rtti -g -O2 -Wall -Wno-unused -Wshadow
-Woverloaded-virtual -Wall -Wconversion -Wpointer-arith
-Wcast-align -Woverloaded-virtual -Wsynth
-Wno-ctor-dtor-privacy -Wno-non-virtual-dtor -g -O2 -o yelp
yelp-Yelper.o yelp-yelp-base.o yelp-yelp-bookmarks.o
yelp-yelp-cache.o yelp-yelp-db-pager.o yelp-yelp-error.o
yelp-yelp-gecko-utils.o yelp-yelp-html.o
yelp-yelp-io-channel.o yelp-yelp-pager.o
yelp-yelp-settings.o yelp-yelp-toc-pager.o yelp-yelp-utils.o
yelp-yelp-window.o yelp-yelp-xslt-pager.o
yelp-yelp-marshal-main.o yelp-yelp-main.o
yelp-yelp-info-pager.o yelp-yelp-info-parser.o
yelp-GNOME_Yelp-stubs.o yelp-GNOME_Yelp-skels.o
yelp-GNOME_Yelp-common.o -Wl,-bind_at_load  -L/sw/lib
-L/usr/X11R6/lib /sw/lib/libglade-2.0.dylib
/sw/lib/libgnomeui-2.dylib /sw/lib/libjpeg.dylib
/sw/lib/libbonoboui-2.dylib -lSM -lICE
/sw/lib/libgnome-keyring.dylib
/sw/lib/libgnomecanvas-2.dylib /sw/lib/libgnome-2.dylib
/sw/lib/libesd.dylib /sw/lib/libaudiofile.dylib
/sw/lib/libart_lgpl_2.dylib /sw/lib/libgtk-x11-2.0.dylib
/sw/lib/libgdk-x11-2.0.dylib -lXrandr -lXinerama -lXext
-lXcursor /sw/lib/libatk-1.0.dylib
/sw/lib/libgdk_pixbuf-2.0.dylib
/sw/lib/libpangoxft-1.0.dylib -lXft -lXrender
/sw/lib/libpangox-1.0.dylib -lX11
/sw/lib/libpangoft2-1.0.dylib -lfontconfig -lfreetype
/sw/lib/libpango-1.0.dylib /sw/lib/libgnomevfs-2.dylib
/sw/lib/libbonobo-2.dylib /sw/lib/libgconf-2.dylib
/sw/lib/libbonobo-activation.dylib
/sw/lib/libORBitCosNaming-2.dylib /sw/lib/libORBit-2.dylib
/sw/lib/libgobject-2.0.dylib -lresolv
/sw/lib/libgmodule-2.0.dylib /sw/lib/libgthread-2.0.dylib
/sw/lib/libglib-2.0.dylib /sw/lib/libexslt.dylib
/sw/lib/libxslt.dylib /sw/lib/libxml2.dylib -lpthread -lm
/sw/lib/libpopt.dylib /sw/lib/libintl.dylib
/sw/lib/libiconv.dylib -lc -lz -lbz2 -L/sw/lib/firefox
-lgtkembedmoz -lxpcom -lplds4 -lplc4 -lnspr4 -ldl
/usr/bin/ld: warning multiple definitions of symbol
_locale_charset
/sw/lib/libiconv.dylib(localcharset.o) definition of
_locale_charset
/sw/lib/libintl.dylib(localcharset.o) definition of
_locale_charset
/usr/bin/ld: warning multiple definitions of symbol _poll
/usr/lib/gcc/i686-apple-darwin8/4.0.1/../../../libpthread.dylib(poll.So)
definition of _poll
/sw/lib/firefox/libnspr4.dylib(unix.o) definition of _poll
/usr/bin/ld: Undefined symbols:
nsCOMPtr_base::assign_from_qi(nsQueryInterface, nsID const&)
nsCOMPtr_base::begin_assignment()
nsCOMPtr_base::assign_from_helper(nsCOMPtr_helper const&,
nsID const&)
nsCOMPtr_base::assign_with_AddRef(nsISupports*)
nsCOMPtr_base::~nsCOMPtr_base()
vtable for nsGetInterface
vtable for nsCreateInstanceByContractID
nsCOMPtr_base::assign_from_gs_contractid(nsGetServiceByContractID,
nsID const&)
collect2: ld returned 1 exit status
make[3]: *** [yelp] Error 1
make[2]: *** [all] Error 2
make[1]: *** [all-recursive] Error 1
make: *** [all] Error 2
### execution of make failed, exit code 2
Failed: phase compiling: yelp-viewer-firefox-2.12.2-1001 failed

indicating that firefox's API has changed and that
firefox-shlibs 1.5.0.1 is not a drop-in replacement for
version 1.0.7 and therefore this package needs to be renamed
to firefox15 or some such. The -shlibs files would also need
to be moved to a /sw/lib/firefox15 directory since they
don't have version numbers in their names and would conflict
with the 1.0.7 files (which would have to remain for
backward compatibility).

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-03-21 09:38

Message:
Logged In: YES 
user_id=927770

Adding updated patch and info files that take into account
package fixes that were made to 1.0.7 in unstable.

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2006-02-14 10:09

Message:
Logged In: YES 
user_id=927770

This is update to 1.5.0.1.  It also fixes the ability to open a new url from the 
commandline when an instance of Firefox is already running.  This was done 
by splicing the running instance check from /sw/bin/firefox (1.0.7) into the 
1.5.0.1 file.  Tested on 10.3.

When first tested on 10.4-T several months ago, I had to manually set up a 
PATH during CompileScript, partly cause it insisted on using gcc-4, 
regardless of GCC setting.  In 10.3, I was able to get rid of that and use the 
default CompileScript.  This should be OK for 10.4 since it uses gcc-4, but 
might need to be tested first.

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2005-12-16 10:43

Message:
Logged In: YES 
user_id=535292

Whoops...this only applies to 1.0.7...ne'ermind here...

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2005-12-16 10:42

Message:
Logged In: YES 
user_id=535292

Regarding We've recently observed that 'sudo' wipes the
DYLD_LIBRARY_PATH environment variable.

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2005-12-07 13:41

Message:
Logged In: YES 
user_id=535292

Fixed in 1.0.7. Also tweaked the %N-dev Depends on %N-shlibs
to be =%v-%r instead of >=%v-%r. The -dev packages are never
used at runtime and there's no way someone could want
"either one, I don't care" (one always requests a specific
one) so packages that use them are always/still in control.
-dev are meant to be swapped around:)

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2005-12-07 11:04

Message:
Logged In: YES 
user_id=927770

I would guess they can replace each other, but I have never
installed mozilla, so I can't be sure of it.  The one
possible catch I can think of is that Fink's mozilla is
1.7.5 (the newest is 1.7.12), and Firefox-1.5 came off the
newer mozilla-1.8 branch (firefox-1.0.x is from the 1.7 branch).

So I guess while we're still at 1.0.x, they can Replace: and
if people complain change it back ;)

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2005-12-07 10:50

Message:
Logged In: YES 
user_id=535292

Currently, firefox-dev and mozilla-dev Conflicts with each other but do not Replaces each other. That
means if a user wants one and has the other installed, fink crashes instead of automaticlaly swapping
them. Is there a reason why this automatic swapping is a problem, or should they also Replaces each other?

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2005-12-02 18:55

Message:
Logged In: YES 
user_id=927770

Ugh.  This was 'fixed' by a big reorganization of how
external URL calls are handled

https://bugzilla.mozilla.org/show_bug.cgi?id=280725

The instructions from
http://www.mozilla.org/unix/remote.html work, but are clumsy.

This will take a while...

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2005-12-02 18:01

Message:
Logged In: YES 
user_id=535292

1.0.7 is updated in 10.3 and 10.4T unstable.

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2005-12-02 15:46

Message:
Logged In: YES 
user_id=927770

It's also in 1.5.  The new attached patch is for 1.0.7-4
(just bump %r in the info file).  I use $moz_libdir, rather
than $p/lib/firefox, and you had your diff +/- switched below :)

Your test of "firefox http://some.url & ; firefox
http://another.url &" now loads the 2nd site OK (even with
the various tab options set).

This does NOT work on 1.5 (dunno why yet), will file that
eventually.

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2005-12-02 11:17

Message:
Logged In: YES 
user_id=535292

I just found a runtime bug in 1.0.7, could you see if it is in 1.5 also? If I 'firefox http://some-url &' and
then 'firefox http://some-other-url &', that second firefox command is supposed to detect that there is
already a firefox window open and display some-other-url in it. It does not detect it, so it prompts me for a
profile, and crashes with "already in use" if I pick my usual one (it's in use by the instance it failed to
detect). The problem is that the /sw/bin/firefox script looks for run-mozilla.sh in /sw/bin instead of
/sw/lib/firefox. Solution is to patch /sw/bin/firefox:

 found=0
 progname="$0"
-curdir="/sw/lib/firefox"
+curdir=`dirname "$progname"`
 progbase=`basename "$progname"`
 run_moz="$curdir/run-mozilla.sh"

----------------------------------------------------------------------

Comment By: Christian Schaffner (chris01)
Date: 2005-11-30 12:14

Message:
Logged In: YES 
user_id=286406

Please remember to set the Group field to "Undergoing
Validation" when you create a new Tracker item and
any time you update your submission and are waiting
for feedback. Thanks.

----------------------------------------------------------------------

Comment By: Hanspeter Niederstrasser (nieder)
Date: 2005-11-30 12:07

Message:
Logged In: YES 
user_id=927770

That certainly puts a damper on things.  I guess I can
submit 1.0.7 for stable after libgettext3 gets moved and
just check this in now to unstable.

----------------------------------------------------------------------

Comment By: Daniel Macks (dmacks)
Date: 2005-11-30 11:47

Message:
Logged In: YES 
user_id=535292

libgettext3 won't be in stable for a few weeks, so firefox
1.0.7 can't move:(

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=414256&aid=1370163&group_id=17203

Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642

Gmane