Enrico Forestieri a écrit :
On Sun, May 21, 2006 at 10:35:30AM +0200, Abdelrazak Younes wrote:
Everything compiled with "-03 -w" in 35 minutes. That's twice as fast as
autotools (with -j3) :-)
Do you really mean "-03" or is it "-O3"? ;-)
"-O3" of course ;-). In the Qwerty keyboard, O and 0
Yes, I noticed that I have always to use "fast_start=no", too.
I spent days on this option! :-)
After scons is stabilized (no frequent env change), this feature is
supposed to be useful and is non-intrusive. The first time when scons
is used, a file called env.cache is created and cache all tes
1) a very minor glitch when compiling qt4: Both
-ID:\program\Qt\4.1\include\QtCore4 and
-ID:\program\Qt\4.1\include\QtCore are passed (same for QtGui). The
first one is not useful (and would never exists), please remove. The
trailing '4' is only possible for libraries.
This is a qt4.py bug and I
On Sun, May 21, 2006 at 10:35:30AM +0200, Abdelrazak Younes wrote:
> Everything compiled with "-03 -w" in 35 minutes. That's twice as fast as
> autotools (with -j3) :-)
Do you really mean "-03" or is it "-O3"? ;-)
> Then "scons install" restarted the compilation from the beginning. The
> full
On Sat, May 20, 2006 at 10:03:38PM -0500, Bo Peng wrote:
> > > > > 2) nls=yes gettext=included is correctly taken into account but -lintl
> > > > > is passed. So the mingw intl dll is used as opposed as libs/libintl.a,
> > > >
>
> Does the order of -L matter? I notice that -Ldebug/libs is the fir
Bo Peng a écrit :
> Please test the attached patch. I changed the names of included
> boost_* and intl to included_boost_* and included_intl, just to be
> safe. Other changes are trivial.
All my problems are solved Bo. While you are maybe right about -L
ordering I think this renaming is a safe b
Please test the attached patch. I changed the names of included
boost_* and intl to included_boost_* and included_intl, just to be
safe. Other changes are trivial.
Bo
Index: development/scons/SConscript
===
--- development/scons/SCon
> > > 2) nls=yes gettext=included is correctly taken into account but -lintl
> > > is passed. So the mingw intl dll is used as opposed as libs/libintl.a,
> >
Does the order of -L matter? I notice that -Ldebug/libs is the first
in the link command so shouldn't libintl.a in this directory be used
On Sat, May 20, 2006 at 04:41:26PM -0500, Bo Peng wrote:
> > > 2) nls=yes gettext=included is correctly taken into account but -lintl
> > > is passed. So the mingw intl dll is used as opposed as libs/libintl.a,
> >
> > Hmm, using -lsomelib a dynamic library takes precedence. When both
> > dynamic
> Sorry for not testing before Bo but there's still some problem with mingw:
> 1) CFLAGS=-O3 works correctly but not CFLAGS="-O3 -g". Scons gives "-O3
> -g" to gcc including the double-quotes.
Maybe scons tries to be too smart when seeing the space. Should be easy
to fix.
Should be easy to fix.
On Sat, May 20, 2006 at 06:57:51PM +0200, Abdelrazak Younes wrote:
> Sorry for not testing before Bo but there's still some problem with mingw:
> 1) CFLAGS=-O3 works correctly but not CFLAGS="-O3 -g". Scons gives "-O3
> -g" to gcc including the double-quotes.
Maybe scons tries to be too smart wh
Bo Peng a écrit :
A minor glitch is that it installs two copies of the lyx executable,
one named lyx-1.5.0svn.exe and the other lyx-qt3-1.5.0svn.exe.
If you think this is necessary, a symlink maybe in order here.
That is right now intentional, since I would like to keep both lyx-qt3
and lyx-qt4
A minor glitch is that it installs two copies of the lyx executable,
one named lyx-1.5.0svn.exe and the other lyx-qt3-1.5.0svn.exe.
If you think this is necessary, a symlink maybe in order here.
That is right now intentional, since I would like to keep both lyx-qt3
and lyx-qt4 executables for co
On Fri, May 19, 2006 at 11:17:28AM -0500, Bo Peng wrote:
> > It now fails when linking lyxclient. I think it is due to missing
> > -lintl and -lshlwapi. Then there is a problem with libraries ordering
> > (I think -lsupports should come before -lboost_*).
>
> Updated patch is attached, this time
It now fails when linking lyxclient. I think it is due to missing
-lintl and -lshlwapi. Then there is a problem with libraries ordering
(I think -lsupports should come before -lboost_*).
Updated patch is attached, this time tested under cygwin.
Bo
Index: development/scons/SConscript
===
On Fri, May 19, 2006 at 12:08:06AM -0500, Bo Peng wrote:
> Updated patch attached.
It now fails when linking lyxclient. I think it is due to missing
-lintl and -lshlwapi. Then there is a problem with libraries ordering
(I think -lsupports should come before -lboost_*).
g++ -o release/common/clie
Updated patch attached.
I am too lazy to separate my attempt of including _moc.cpp into .C
(qt3). The scons part should work but the autotool part is incomplete.
It would be nice if someone can finish it.
Bo
Index: development/scons/SConscript
Using nls=yes and gettext=included, it fails as follows:
OK. The lession is that I should not mix scons and autotools builds...
An updated patch will be supplied soon.
Bo
On Thu, May 18, 2006 at 08:49:41PM +0200, Enrico Forestieri wrote:
> On Thu, May 18, 2006 at 08:51:26AM -0500, Bo Peng wrote:
>
> > Please test it. I am not in a hurry to apply.
>
> I will test it on cygwin tonight.
Using nls=yes and gettext=included, it fails as follows:
g++ -o release/common
On Thu, May 18, 2006 at 08:51:26AM -0500, Bo Peng wrote:
> Please test it. I am not in a hurry to apply.
I will test it on cygwin tonight.
--
Enrico
Abdelrazak Younes wrote:
> OK, thanks to you both for the information. It's maybe a good time to
> try this now.
> So, the first step is to try to include these header instead of the
> stripped down ones. If that works, the second step would be to use the
> object already compiled for LyX. At this
Georg Baum a écrit :
Abdelrazak Younes wrote:
Hello Text2Lyx expert, could you tell us please if it is safe to assume
that using the general headers instead of the tex2lyx one is OK?
No. You would get linking errors (at least I got them when I tried to use
them). The general rule is that you
Abdelrazak Younes wrote:
> Hello Text2Lyx expert, could you tell us please if it is safe to assume
> that using the general headers instead of the tex2lyx one is OK?
No. You would get linking errors (at least I got them when I tried to use
them). The general rule is that you can (and should) use
> "Abdelrazak" == Abdelrazak Younes <[EMAIL PROTECTED]> writes:
Abdelrazak> Hello Text2Lyx expert, could you tell us please if it is
Abdelrazak> safe to assume that using the general headers instead of
Abdelrazak> the tex2lyx one is OK?
At the time this was done, this meant linking tex2lyx ag
Bo Peng a écrit :
Hum... This is *bad*. Which headers exactly?
You can compare lyxfont.h under tex2lyx and src.
OK, for this one at least using the one in src should be harmless. The
tex2lyx version is just a cut down version of the one in src/. This
comment in "context.h" make me think I a
Hum... This is *bad*. Which headers exactly?
You can compare lyxfont.h under tex2lyx and src.
Bo
Bo Peng a écrit :
Dear Abdel and Enrico,
Attached patch is supposed to fix:
Excellent!
QUESTION:
6. create multiple "config.h" in build directory not in the source (as
is done with autotools):
2.a. release/config.h for all general macro including
BOOST_USER_CONFIG
2.a. r
Dear Abdel and Enrico,
Attached patch is supposed to fix:
1. remove unnecessay rebuild, since I found that adding fast_start=yes
can sometimes trigger rebuild.
tested under linux
2. fix lyx2lyx_version.py (converted from lyx2lyx_version.py.in)
3. version_suffix (version-suffix=-svn et
28 matches
Mail list logo
