Re: [pygtk] new pygtk for gtk 2.0 snapshot
On Mon, 14 May 2001, Christian Robottom Reis wrote: On Sun, 13 May 2001 [EMAIL PROTECTED] wrote: Christian Skip, I'm having the same problems with gtk+-1.3.5 and the Christian latest (CVS) pygtk. I took the easy way out and or'd in RTLD_GLOBAL to the two dlopen calls at the bottom of the .../Python/dynload_shlib.c in the Python source tree. I think James has a better fix in the mix but it requires changes to some other Gtk stuff. So you mean the fact pygtk is coercing pango to load a .so.la file is a result of the RTLD_GLOBAL problem? _That's unusual_. The gmodule library (a wrapper for various dynamic linker interfaces) in glib-2.0 is able to load libtool .la files natively. I guess it tried to load whatever.so.la as a last resort, and that came out in the error message. The problem is to do with incomplete dependency lists for the pango shaper modules, etc. Patching python to use RTLD_GLOBAL is a work around for the problem. I've spoken to Owen and he says there will be a fix possibly this week, so I'll sit and wait in the meantime. Well, you know where the bugzilla bug with the patches is if you get impatient :) James. -- Email: [EMAIL PROTECTED] WWW: http://www.daa.com.au/~james/ ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
On Mon, 14 May 2001, James Henstridge wrote: I've spoken to Owen and he says there will be a fix possibly this week, so I'll sit and wait in the meantime. Well, you know where the bugzilla bug with the patches is if you get impatient :) The problem is the patches aren't against 0.16, and I did start out applying by hand. However, Owen suggested I wait because other changes he made were important to pango. What version of pango did you provide the patches for? Take care, -- /\/\ Christian Reis, Senior Engineer, Async Open Source, Brazil ~\/~ http://async.com.br/~kiko/ | [+55 16] 274 4311 ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
James This is a known problem or modify python to pass the James RTLD_GLOBAL flag when dlopen'ing extensions. James The second option is easier (Red Hat has a patch to do it in James their python rpms), Any pointers to RedHat's RTLD_GLOBAL patch would be appreciated. I couldn't find it via their search mechanisms. Thanks, -- Skip Montanaro ([EMAIL PROTECTED]) (847)971-7098 ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
On Tue, 17 Apr 2001, Skip Montanaro wrote:
Any pointers to RedHat's RTLD_GLOBAL patch would be appreciated. I couldn't
find it via their search mechanisms.
It's not the correct fix, but anyway.. I guess you know that already. This
was discussed and Guido himself said it was not a reasonable solution.
Take care,
--
/\/\ Christian Reis, Senior Engineer, Async Open Source, Brazil
~\/~ http://async.com.br/~kiko/ | [+55 16] 274 4311
--- Python-1.5.2/Python/importdl.c.global Sat Jul 17 16:52:26 1999
+++ Python-1.5.2/Python/importdl.c Sat Jul 17 16:53:19 1999
@@ -441,13 +441,13 @@
#ifdef RTLD_NOW
/* RTLD_NOW: resolve externals now
(i.e. core dump now if some are missing) */
- void *handle = dlopen(pathname, RTLD_NOW);
+ void *handle = dlopen(pathname, RTLD_NOW | RTLD_GLOBAL);
#else
void *handle;
if (Py_VerboseFlag)
printf("dlopen(\"%s\", %d);\n", pathname,
- RTLD_LAZY);
- handle = dlopen(pathname, RTLD_LAZY);
+ RTLD_LAZY | RTLD_GLOBAL);
+ handle = dlopen(pathname, RTLD_LAZY | RTLD_GLOBAL);
#endif /* RTLD_NOW */
if (handle == NULL) {
PyErr_SetString(PyExc_ImportError, dlerror());
Re: [pygtk] new pygtk for gtk 2.0 snapshot
On Tue, 17 Apr 2001, Skip Montanaro wrote: James This is a known problem or modify python to pass the James RTLD_GLOBAL flag when dlopen'ing extensions. James The second option is easier (Red Hat has a patch to do it in James their python rpms), Any pointers to RedHat's RTLD_GLOBAL patch would be appreciated. I couldn't find it via their search mechanisms. Don't have an exact reference (the patch is probably in their source rpms). The change is very simple. Just edit Python/dynload_shlib.c, and find the call to dlopen close to the bottom of the file. The second argument to dlopen() is a set of flags. You can just or RTLD_GLOBAL against the flags. That is, changing the line: handle = dlopen(pathname, RTLD_NOW); to: handle = dlopen(pathname, RTLD_NOW | RTLD_GLOBAL); (there is no reason to change the second dlopen call, as Linux systems have the RTLD_NOW flag). The effect of this is that loaded extension modules can see each other's symbols (plus the symbols of any other libraries pulled in by shared library dependencies). This could potentially cause symbol conflicts, but in practice doesn't cause problems very often (most extensions only export a single function -- the initmodulename() function. I have some patches that fix the dynamic dependencies in glib, pango and gtk+ (the gtk+ patch still isn't perfect -- the depdendencies for the gdk-pixbuf modules are not quite right yet). They make use of features in libtool-1.3d (a beta release of 1.4). You can find the current set of patches attached to this bugzilla bug: http://bugzilla.gnome.org/show_bug.cgi?id=50707 At the moment, it is probably easier to patch python rather than gtk. Owen says he doesn't want to switch the official gtk packages to the newer libtool until libtool-1.4 is released. James. -- Email: [EMAIL PROTECTED] WWW: http://www.daa.com.au/~james/ ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
On Tue, 3 Apr 2001, Skip Montanaro wrote: James I just put up a new snapshot of devel pygtk at: James http://www.gnome.org/~james/pygtk2-SNAP-20010330.tar.gz I managed to get all the prerequisite stuff installed, but can't build the pygtk snapshot. When compiling gtkobject-support.c I get (long lines wrapped): They went and made some changes after my last snapshot, which broke the pygtk build (as you noticed :) I put together a new snapshot while at GUADEC, but didn't get round to announcing it on any mailing lists. You can grab it from http://canvas.gnome.org/~james/pygtk2-SNAP-20010408.tar.gz It will work fine with the glib/gtk+ 1.3.3 releases (along with whatever version of pango was released at the time). James. -- Email: [EMAIL PROTECTED] WWW: http://www.daa.com.au/~james/ ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
James I put together a new snapshot while at GUADEC ...
Thanks very much. I found it and it installed just fine.
James It will work fine with the glib/gtk+ 1.3.3 releases (along with
James whatever version of pango was released at the time).
I'm having a bit of a problem with pango. The following simple script gets
a number of GRuntime-CRITICAL errors, then repeatedly tries to open
/usr/local/lib/pango/modules/pango-basic-x.so.la, which fails, finally
terminating with a floating point exception:
import gtk
win = gtk.GtkWindow()
t = gtk.GtkTextView()
win.add(t)
win.connect("destroy", gtk.mainquit)
win.show_all()
gtk.mainloop()
Does it run for you? If so, I have some config problems. If not, perhaps
there is something wrong with pango and/or glib and/or libtool.
Thx,
--
Skip Montanaro ([EMAIL PROTECTED])
(847)971-7098
___
pygtk mailing list [EMAIL PROTECTED]
http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
Brent Actually, they made it quite simple to *avoid* having to know Brent auto. Just run ./autogen.sh and it will run generate all the Brent necessary file for you. Thanks to Brent and others who pointed out the existence of the autogen.sh script. I managed to get glib, pango and pkg-config installed okay, however the autogen.sh script in the gtk+ directory didn't work. Executing autogen.sh there spit out a bunch of ominous messages and removed all non-directory files from the directory. Turns out the gettext-devel RPM wasn't installed on my Mandrake system. I don't know if I uninstalled it sometime in the dim dark past or if Mandrake's install process failed to install it. After installing it, autogen.sh execution worked fine. -- Skip Montanaro ([EMAIL PROTECTED]) (847)971-7098 ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
--- Skip Montanaro [EMAIL PROTECTED] escreveu: I realize the gnome folks are trying to do me a favor by making it harder for me to build glib, pango and gtk+ so that it is correspondingly harder to shoot myself in the foot, but I'd like to experiment with James's new release and find it frustrating that I should need to know the ins and outs of autoconf and/or automake. Would someone please tell me how to create ./configure? I tried "autoconf" in the glib dir and got a configure script, but also this error (and a bunch of warnings about AC_TRY_RUN): autoconf: Undefined macros: configure.in:201:AC_LIBTOOL_WIN32_DLL I get the same error when running autoconf on gtk+. No error when running in the pango dir, but in all three directories, when I run the generated configure script I get: ./configure: line 602: syntax error near unexpected token `AM_INIT_AUTOMAKE($PACKAGE,' ./configure: line 602: `AM_INIT_AUTOMAKE($PACKAGE, $VERSION, no-define)' Every other package I've ever pulled by CVS that uses GNU configure to generate config.h has come with a working configure script. Why can't these packages? If it makes any difference, I have autoconf 2.14.1 and automake 1.4. HI, -Try this: ~# ./autogen.sh ~# make ~# make install -Note that others dependencies could exist, requerided to build those cvs packages. s, roberto Roberto J. Teixeira Junior - krivilli '-' ' http://alsabrazil.sourceforge.net ' ' [EMAIL PROTECTED] ' '-' http://www.alsa-project.org O YAHOO! GEOCITIES CHEGOU AO BRASIL! Crie sua home page com tudo em portugus - http://br.geocities.com ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
On Fri, 30 Mar 2001, Skip Montanaro wrote: James I just put up a new snapshot of devel pygtk at: James http://www.gnome.org/~james/pygtk2-SNAP-20010330.tar.gz James It requires python = 2.0 and recent CVS checkouts of glib, pango James and gtk+ to compile. It is not fully up to date with all the API James additions/changes, but it compiles and runs. *sigh*... I realize the gnome folks are trying to do me a favor by making it harder for me to build glib, pango and gtk+ so that it is correspondingly harder to shoot myself in the foot, but I'd like to experiment with James's new release and find it frustrating that I should need to know the ins and outs of autoconf and/or automake. Would someone please tell me how to create ./configure? I tried "autoconf" in the glib dir and got a configure script, but also this error (and a bunch of warnings about AC_TRY_RUN): I have never tried building gtk on win32, so probably won't be able to help that much. From posts to gtk-devel-list, it sounds like you need CVS libtool in order to compile it on win32. I don't know about the other details. It probably has build problems on x11 too, because the gtk-2.0.m4 autoconf macro I was using is broken (it uses gtk-config-2.0 macro, which doesn't exist anymore -- there is a bug in bugzilla about this). As far as pygtk goes, you probably want to use the snapshot tarball, rather than CVS, as the tarball is built with the CVS version of automake (needed for python support). James. ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
[pygtk] new pygtk for gtk 2.0 snapshot
I just put up a new snapshot of devel pygtk at: http://www.gnome.org/~james/pygtk2-SNAP-20010330.tar.gz It requires python = 2.0 and recent CVS checkouts of glib, pango and gtk+ to compile. It is not fully up to date with all the API additions/changes, but it compiles and runs. This release also has support for creating new signals, complete with a class closure. See examples/gobject/signal.py for an example: http://cvs.gnome.org/lxr/source/gnome-python/pygtk/examples/gobject/signal.py Currently you can only use it to add signals to existing GTypes. Creating new GTypes is not currently supported due to some required glib APIs not having been implemented yet. James. -- Email: [EMAIL PROTECTED] WWW: http://www.daa.com.au/~james/ ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
Re: [pygtk] new pygtk for gtk 2.0 snapshot
James I just put up a new snapshot of devel pygtk at: James http://www.gnome.org/~james/pygtk2-SNAP-20010330.tar.gz James It requires python = 2.0 and recent CVS checkouts of glib, pango James and gtk+ to compile. It is not fully up to date with all the API James additions/changes, but it compiles and runs. *sigh*... I realize the gnome folks are trying to do me a favor by making it harder for me to build glib, pango and gtk+ so that it is correspondingly harder to shoot myself in the foot, but I'd like to experiment with James's new release and find it frustrating that I should need to know the ins and outs of autoconf and/or automake. Would someone please tell me how to create ./configure? I tried "autoconf" in the glib dir and got a configure script, but also this error (and a bunch of warnings about AC_TRY_RUN): autoconf: Undefined macros: configure.in:201:AC_LIBTOOL_WIN32_DLL I get the same error when running autoconf on gtk+. No error when running in the pango dir, but in all three directories, when I run the generated configure script I get: ./configure: line 602: syntax error near unexpected token `AM_INIT_AUTOMAKE($PACKAGE,' ./configure: line 602: `AM_INIT_AUTOMAKE($PACKAGE, $VERSION, no-define)' Every other package I've ever pulled by CVS that uses GNU configure to generate config.h has come with a working configure script. Why can't these packages? If it makes any difference, I have autoconf 2.14.1 and automake 1.4. Thanks, -- Skip Montanaro ([EMAIL PROTECTED]) (847)971-7098 ___ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk
