Errors building 1.4.0 (?)

[Mike Meyer]

I'm trying to build the CVS version for the first time. I used the
page that google had cached from www.devel.lyx.org on getting LyX
sources from CVS to get started.

That page talks about building the development module but doesn't talk
about what that is. The difference is that you run the autogen.sh
script before doing the configure/make/make install dance. I did that.

The build platform is FreeBSD 5.4-STABLE. That uses gcc 3.4.2. The
configure args I used were --with-frontend=qt --with-qt-dir=/usr/X11R6
--with-aspell -with-extra-lib=/usr/opt/lib
-with-extra-inc=/usr/opt/include.

The FreeBSD ports tree builds version 1.3.5 with no patches. I used
the option it had to build with qt and aspell, and it adds the
-with-extra-* arguments to configure when it builds 1.3.5. Since I'm
using the same qt installation, I figured copying them couldn't hurt.

The compile exits with a slew of missig references to Qt objects. The
command that generated them and the first of them looks like this:

g++ -fno-exceptions -g -O -o lyx-qt main.o Bidi.o BufferView.o 
BufferView_pimpl.o Bullet.o Bran
chList.o Chktex.o CutAndPaste.o DepTable.o FloatList.o Floating.o 
FontIterator.o FuncStatus.o I
nsetList.o LColor.o LaTeX.o LaTeXFeatures.o LyXAction.o MenuBackend.o 
ParagraphParameters.o Pri
nterParams.o Spacing.o Thesaurus.o ToolbarBackend.o author.o boost.o box.o 
buffer.o buffer_func
s.o bufferlist.o bufferparams.o bufferview_funcs.o changes.o chset.o 
converter.o counters.o coo
rdcache.o cursor.o cursor_slice.o debug.o dimension.o dociterator.o encoding.o 
errorlist.o expo
rter.o gettext.o factory.o format.o funcrequest.o graph.o importer.o intl.o 
insetiterator.o kbm
ap.o kbsequence.o language.o lastfiles.o lengthcommon.o lyx_cb.o lyx_main.o 
lyx_sty.o lyxfont.o
 lyxfind.o lyxfunc.o lyxgluelength.o lyxlayout.o lyxlength.o lyxlex.o 
lyxlex_pimpl.o lyxrc.o ly
xrow.o lyxrow_funcs.o lyxserver.o lyxsocket.o lyxtextclass.o lyxtextclasslist.o 
lyxvc.o message
s.o metricsinfo.o mover.o output.o outputparams.o output_docbook.o 
output_latex.o output_linuxd
oc.o output_plaintext.o paragraph.o paragraph_funcs.o paragraph_pimpl.o 
pariterator.o ispell.o 
SpellBase.o rowpainter.o sgml.o tabular.o tex-accent.o tex-strings.o texrow.o 
text.o text2.o te
xt3.o toc.o trans.o trans_mgr.o undo.o vc-backend.o version.o vspace.o  
mathed/.libs/libmathed.
a insets/.libs/libinsets.a frontends/.libs/libfrontends.a 
frontends/qt2/.libs/libqt2.a -L/usr/X
11R6//lib frontends/controllers/.libs/libcontrollers.a 
graphics/.libs/libgraphics.a support/.li
bs/libsupport.a ../boost/libs/regex/src/.libs/libboost_regex.a 
../boost/libs/signals/src/.libs/
libboost_signals.a ../boost/libs/filesystem/src/.libs/libboost_filesystem.a 
../intl/libintl.a -
lSM -lICE -lm -L/usr/X11R6/lib -lX11 -lz
frontends/qt2/.libs/libqt2.a(Alert_pimpl.o)(.text+0x150): In function 
`prompt_pimpl(std::string
 const, std::string const, int, int, std::string const, std::string const, 
std::string cons
t)':
/usr/X11R6//include/qapplication.h:451: undefined reference to 
`QApplication::focus_widget'

Looking over this, it's a link command - except that I don't see a qt
library being linked in. Grovelling through config.status finds: 
s,@QT_LIB@,,;t t,
which sure looks like it didn't set the QT library properlty.

I could probably figure this out myself - except it's 3am, and I've
been fighting recalcitrant hardware all night. I'm going to sleep,
then to work, and with luck and help from you all, I'll find a
solution in my mailbox tomorrow evening.

Thanks,
mike
-- 
Mike Meyer [EMAIL PROTECTED]  http://www.mired.org/consulting.html
Independent Network/Unix/Perforce consultant, email for more information.

Re: Errors building 1.4.0 (?)

[Georg Baum]

Mike Meyer wrote:

 Looking over this, it's a link command - except that I don't see a qt
 library being linked in. Grovelling through config.status finds:
 s,@QT_LIB@,,;t t, which sure looks like it didn't set the QT library
 properlty.

It looks like configure does not find the qt libary (but it seems to find
the headers). This looks like a configure bug: It should stop with an error
message if the library cannot be found.
Could you please post the lines of config.log where it searches for the qt
library? (around the message checking for Qt library name).
Have you tried it without the --qt-dir switch?


Georg

Re: byte compilation of lyx2lyx files

[Lars Gullik Bjønnes]

Angus Leeming [EMAIL PROTECTED] writes:

| Lars Gullik Bjønnes wrote:
| 
|  Angus Leeming [EMAIL PROTECTED] writes:
|  
|  | Luis Rivera reports that make install is failing for him because he
|  | doesn't have python installed. Could someone sort this out please.
|  
|  Hmm... we kindo require python for building.
| 
| We do? Where? Since when?

It is in the kindo cathegory.

-- 
Lgb

Re: byte compilation of lyx2lyx files

[Lars Gullik Bjønnes]

[EMAIL PROTECTED] (Lars Gullik Bjønnes) writes:

| Angus Leeming [EMAIL PROTECTED] writes:
| 
| | Lars Gullik Bjønnes wrote:
| | 
| |  Angus Leeming [EMAIL PROTECTED] writes:
| |  
| |  | Luis Rivera reports that make install is failing for him because he
| |  | doesn't have python installed. Could someone sort this out please.
| |  
| |  Hmm... we kindo require python for building.
| | 
| | We do? Where? Since when?
| 
| It is in the kindo cathegory.

We should be able to handle the no-python case with some use of 

AM_CONDITIONAL([HAVE_PYTHON], [test $PYTHON != :])

just let lyx2lyx's Makefile.am not do anything unless HAVE_PYTHON is
defined.

-- 
Lgb

Re: large executable

[Jean-Marc Lasgouttes]

 Pol == Pol  [EMAIL PROTECTED] writes:

Pol Jean-Marc Lasgouttes wrote:
 that i am compiling lyx 1.4
Paolo The compiled lyx 1.4 executable from cvs, last april, was 4.2
Paolo megabytes; but in that case compilation had been accoplished on
Paolo ordinary intel desktop workstation. That is the reason for
Paolo suspecting something wrong,
 On what system are you compiling?
 

Pol it is debian unstable with gcc 4.0.1-3 (corresponding to gcc
Pol version 4.0.2 20050725 (prerelease), as resulting from ' gcc -v
Pol ') and libqt3-mt-dev 3.3.4-3

what does size ./lyx day? Here (gcc 3.4.1) I get:

fantomas: size src/lyx
   textdata bss dec hex filename
6083056   17680   35060 6135796  5d9ff4 src/lyx

JMarc

Re: LyX 140 configure problem with awk

[Jean-Marc Lasgouttes]

 Bennett == Bennett Helm [EMAIL PROTECTED] writes:

Bennett That seems to work for me.

The important question is whether the information in lyx.pot is
reasonable. You should see entries like:

msgid \\arabic{enumi}.
msgstr 

(what is important here is the double backslash).

JMarc

Re: byte compilation of lyx2lyx files

[Jean-Marc Lasgouttes]

 Angus == Angus Leeming [EMAIL PROTECTED] writes:

Angus Luis Rivera reports that make install is failing for him
Angus because he doesn't have python installed. Could someone sort
Angus this out please.

I need more details, could you point me to the relevant thread?
Basically, this is supposed to work provided automake is recent
enough.

So, if I do not see the error message, there is nothing I can do.

JMarc

Re: [PATCH] bug 1941: toolbars disabled after dialog action

[Jean-Marc Lasgouttes]

 Georg == Georg Baum [EMAIL PROTECTED] writes:

Georg I tried to understand where in the code that happens, but
Georg failed. Could you please tell me the place where the main
Georg window without focus prevents the icons from being enabled?

In LyXFunc::getStatus, you will find:

/* In LyX/Mac, when a dialog is open, the menus of the
   application can still be accessed without giving focus to
   the main window. In this case, we want to disable the menu
   entries that are buffer-related.
*/
Buffer * buf;
if (cmd.origin == FuncRequest::UI  !owner-hasFocus())
buf = 0;
else
buf = owner-buffer();

We do as if there was no buffer. This means that all not
buffer-related functions are still active. 

 My fix is to avoid updating toolbar/menubar for INTERNAL
 dispatches.

Georg I don't think that this is the right fix. The problem is
Georg qt-only, so it should be fixed in the qt frontend. 

It is not really qt-only. The same code would trigger for a Cocoa
port, or a Gtk/osx. 

Georg I don't know if that currently happens, but I could imagine the
Georg case where applying a dialog makes an lfun invalid. In this
Georg case we certainly want that icon to be disabled, but that does
Georg not work anymore with your fix.

Indeed.

Georg The right fix is IMHO to fix the focus problem. Why is it not
Georg possible to give the focus to the main window after a dialog is
Georg closed?

This is what I tried first, actually: updating the tool/menu bars after
the window has been hidden in Dialog::hide. It did not work, and I do
not understand why.

JMarc

Re: byte compilation of lyx2lyx files

[Jean-Marc Lasgouttes]

 Lars == Lars Gullik Bjønnes [EMAIL PROTECTED] writes:

Lars [EMAIL PROTECTED] (Lars Gullik Bjønnes) writes: | Angus Leeming
Lars [EMAIL PROTECTED] writes:
Lars | 
Lars | | Lars Gullik Bjønnes wrote:
Lars | | 
Lars | |  Angus Leeming [EMAIL PROTECTED] writes:
Lars | |  
Lars | |  | Luis Rivera reports that make install is failing for
Lars him because he | |  | doesn't have python installed. Could
Lars someone sort this out please.
Lars | |  
Lars | |  Hmm... we kindo require python for building.
Lars | | 
Lars | | We do? Where? Since when?
Lars | 
Lars | It is in the kindo cathegory.

Lars We should be able to handle the no-python case with some use of

Actually, I do not see why it should not work when PYTHON=:. That's
why I'd like to see the error first.

JMarc

Re: byte compilation of lyx2lyx files

[Jose' Matos]

On Tuesday 13 September 2005 10:44, Jean-Marc Lasgouttes wrote:
 Actually, I do not see why it should not work when PYTHON=:. That's
 why I'd like to see the error first.

  I am puzzled as well. We made the support for python conditional, it should 
not fail if python is not present.

 JMarc

-- 
José Abílio

Re: byte compilation of lyx2lyx files

[Angus Leeming]

Jean-Marc Lasgouttes wrote:
 Angus Luis Rivera reports that make install is failing for him
 Angus because he doesn't have python installed. Could someone sort
 Angus this out please.

 I need more details, could you point me to the relevant thread?
 Basically, this is supposed to work provided automake is recent
 enough.

 So, if I do not see the error message, there is nothing I can do.
 JMarc

Luis didn't provide much information. Just the second paragraph here, IIRC:
http://article.gmane.org/gmane.editors.lyx.devel:47931

I'll drop him a line.

-- 
Angus

Re: byte compilation of lyx2lyx files

[Jean-Marc Lasgouttes]

 Angus == Angus Leeming [EMAIL PROTECTED] writes:

Angus Luis didn't provide much information. Just the second paragraph
Angus here, IIRC:
Angus http://article.gmane.org/gmane.editors.lyx.devel:47931

Yes, I saw that.

Angus I'll drop him a line.

Thanks.

JMarc

Safer Win32 get_temp_path, os::internal_path

[Angus Leeming]

Jean-Marc, all,

following our earlier discussion, I attach a little test code to show what 
I think we need to produce robust code on Windows. The attached code should 
compile on all flavours of Windows = Win95 and should behave gracefully on 
older versions for which GetLongPathName is just a stub.


I propose to post this sample off to Luis Rivera and get him to try it out 
on Win95/Win98 but I'd of course welcome any sanity checks and comments 
from you first.


Angus (the one of questionable moral character :-P)

$ g++ -I$HOME/lyx/13x/boost -o trial trial.cpp
$ ./trial
Temp dir is C:/Documents and Settings/Angus/Local Settings/Temp/
#include boost/assert.hpp

#include iostream
#include string
#include vector

/*
 * MinGW's version of winver.h contains this comment:
 *
 * If you need Win32 API features newer the Win95 and WinNT then you must
 * define WINVER before including windows.h or any other method of including
 * the windef.h header.
 *
 * GetLongPathNameA requires WINVER == 0x0500.
 *
 * It doesn't matter if the Windows version is older than this because the
 * function will compile but will fail at run time. See
 * 
http://msdn.microsoft.com/library/en-us/mslu/winprog/microsoft_layer_for_unicode_apis_with_limited_support.asp
 */
# if defined(__MINGW32__)
#  define WINVER 0x0500
# endif

#include windows.h

using std::string;
using std::vector;


namespace {

string const subst(string const  a,
   string const  oldstr, string const  newstr)
{
BOOST_ASSERT(!oldstr.empty());

string lstr(a);
string::size_type i = 0;
string::size_type const olen = oldstr.length();
while ((i = lstr.find(oldstr, i)) != string::npos) {
lstr.replace(i, olen, newstr);
i += newstr.length(); // We need to be sure that we dont
// use the same i over and over again.
}
return lstr;
}


namespace os {

string const internal_path(string const p)
{
string const short_path = subst(p, \\, /);

vectorchar long_path(PATH_MAX);
DWORD result = GetLongPathName(short_path.c_str(),
   long_path[0], long_path.size());

if (result  long_path.size()) {
long_path.resize(result);
result = GetLongPathName(short_path.c_str(),
 long_path[0], long_path.size());
BOOST_ASSERT(result = long_path.size());
}

return (result == 0) ? short_path : long_path[0];
}

} // namespace os


string const get_temp_dir()
{
// Typical example: C:/TEMP/.
char path[PATH_MAX];
DWORD const result = GetTempPath(PATH_MAX, path);
BOOST_ASSERT(result  0  result = PATH_MAX);
return os::internal_path(path);
}

} // namespace anon


int main()
{
string const temp_dir = get_temp_dir();
std::cout  Temp dir is   temp_dir  std::endl;
return 0;
}

Re: large executable

[Pol]

Jean-Marc Lasgouttes wrote:


what does size ./lyx day? Here (gcc 3.4.1) I get:

fantomas: size src/lyx
   textdata bss dec hex filename
6083056   17680   35060 6135796  5d9ff4 src/lyx



% size /usr/local/bin/lyx


  textdata bss  dec hex   filename
  11134899 394540   35332 11564771b076e3 /usr/local/bin/lyx


--
Paolo

Re: gcc4 prevents some languages from working - still stuck :-(

[Helge Hafting]

Martin Vermeer wrote:


On Wed, Aug 17, 2005 at 04:31:07PM +0100, Angus Leeming wrote:
 


Martin Vermeer wrote:

   


On Tue, 2005-08-16 at 12:59, Helge Hafting wrote:
 


I noticed something very strange today.
I could not get lyx-1.4cvs to display norwegian text at all.
no_NO, no_NO.iso8859-1, no_NO.utf-8 all gave me english.
This on a machine that uses gcc 4.01, another machine
using gcc-3.3.6 works.

Other languages, like nn_NO works.

So, it is not a case of complete locale breakdown,
and not a case of a completely broken no.po either.

Any ideas?  I was going to check if recent fixes had added
strings I could translate, but this is driving me nuts.
The no_NO locales works for other programs, such as ls.  So
I belive the distribution is correctly set up.

Helge Hafting
   


Does nb_NO work?
 


Debugging this would be much, much easier if we changed gettext_init
(lyx 1.3.x) in src/gettext.C to something like:

void gettext_init(string const  localedir)
{
   char const * const btd = bindtextdomain(PACKAGE, localedir.c_str());
   char const * const td = textdomain(PACKAGE);

   if (!btd || !td) {
   lyxerr  Failed to initialize localedir   localedir
successfully\nbindtextdomain   btd
   \ntextdomain   td  std::endl;
   }
}

An equivalent change to lyx 1.4 is also needed, IMO.
   



This *is* a 1.4 issue. In 1.4, the relevant code is in src/messages.C.
It's not pretty. This stuff was touched in Paris.

117 char const * works = setlocale(LC_MESSAGES, lang_.c_str());

add a printout here:

if (!works)
lyxerr  Locale   lang_   could not be set  
std::endl;

I'll commit this as it is clearly needed.
 


I am still stuck on this.  I can't get no_NO to work for
the no.po in cvs.  There are no error messages (or any message different
from when trying other working languages, like nn_NO).  I simply
get english when LANG=no_NO (or no_NO.iso8859-1).

This seems to imply that something is wrong in the no.po
file, but I have no idea what that could be.  make no.gmo
gives no error messages, the file is created, but Norwegian
language just don't work. 


Helge Hafting

Re: gcc4 prevents some languages from working - still stuck :-(

[Martin Vermeer]

On Tue, 2005-09-13 at 13:26 +0200, Helge Hafting wrote:

...

 I am still stuck on this.  I can't get no_NO to work for
 the no.po in cvs.  There are no error messages (or any message different
 from when trying other working languages, like nn_NO).  I simply
 get english when LANG=no_NO (or no_NO.iso8859-1).
 
 This seems to imply that something is wrong in the no.po
 file, but I have no idea what that could be.  make no.gmo
 gives no error messages, the file is created, but Norwegian
 language just don't work. 
 
 Helge Hafting

This is utterly weird. Did you create a bug in bugzilla for this?

- Martin



signature.asc
Description: This is a digitally signed message part

Re: Safer Win32 get_temp_path, os::internal_path

[Jean-Marc Lasgouttes]

 Angus == Angus Leeming [EMAIL PROTECTED] writes:

Angus Jean-Marc, all, following our earlier discussion, I attach a
Angus little test code to show what I think we need to produce robust
Angus code on Windows. The attached code should compile on all
Angus flavours of Windows = Win95 and should behave gracefully on
Angus older versions for which GetLongPathName is just a stub.

This looks good, if it works.

Angus Angus (the one of questionable moral character :-P)

To tell, the truth, I suspected it already, and was rather relieved
not to have you in my house last July. I do not know what my
neighbours would have thought.

JMarc

Re: Errors building 1.4.0 (?)

[Mike Meyer]

In [EMAIL PROTECTED], Georg Baum [EMAIL PROTECTED] typed:
 Mike Meyer wrote:
  Looking over this, it's a link command - except that I don't see a qt
  library being linked in. Grovelling through config.status finds:
  s,@QT_LIB@,,;t t, which sure looks like it didn't set the QT library
  properlty.
 It looks like configure does not find the qt libary (but it seems to find
 the headers). This looks like a configure bug: It should stop with an error
 message if the library cannot be found.
 Could you please post the lines of config.log where it searches for the qt
 library? (around the message checking for Qt library name).

Sure:

configure:24202: checking for Qt library name
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt-mt 5
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_cleanup_pop'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_attr_destroy'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_attr_init'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_exit'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_cancel'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_testcancel'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_cleanup_push'
/usr/X11R6//lib/libqt-mt.so: undefined reference to 
`pthread_attr_getschedpolicy'
/usr/X11R6//lib/libqt-mt.so: undefined reference to 
`pthread_attr_setinheritsched'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_attr_setstacksize'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_attr_setschedparam'
/usr/X11R6//lib/libqt-mt.so: undefined reference to 
`pthread_attr_setdetachstate'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_cond_timedwait'
configure:24253: $? = 1
configure: failed program was:
[listing of failing program elided]
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt-mt3 5
/usr/bin/ld: cannot find -lqt-mt3
configure:24253: $? = 1
configure: failed program was:
[listing of failed program elded]
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt3 5
/usr/bin/ld: cannot find -lqt3
configure:24253: $? = 1
configure: failed program was:
[listing of failed program elded]
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt2 5
/usr/bin/ld: cannot find -lqt2
configure:24253: $? = 1
configure: failed program was:
[listing of failed program elded]
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt 5
/usr/bin/ld: cannot find -lqt
configure:24253: $? = 1
configure: failed program was:
configure:24289: result: failed
configure:24306: checking whether the Qt library is multi-threaded
configure:24345: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11   5
/var/tmp//cc4P8UEz.o(.text+0x118): In function `main':
/home/mwm/external-src/lyx-devel/conftest.cc:51: undefined reference to 
`QApplication::QApplica
tion(_XDisplay*, unsigned long, unsigned long)'
/var/tmp//cc4P8UEz.o(.text+0x123):/home/mwm/external-src/lyx-devel/conftest.cc:52:
 undefined re
ference to `QApplication::unlock(bool)'
/var/tmp//cc4P8UEz.o(.text+0x136):/home/mwm/external-src/lyx-devel/conftest.cc:55:
 undefined re
ference to `QApplication::~QApplication()'
/var/tmp//cc4P8UEz.o(.text+0x14c):/home/mwm/external-src/lyx-devel/conftest.cc:55:
 undefined re
ference to `QApplication::~QApplication()'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0xc): undefined reference to 
`QGList::clear()'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0x10):/usr/include/c++/3.4/bits/basic_string.h
:257: undefined reference to `QGList::~QGList()'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0x14):/usr/include/c++/3.4/bits/basic_string.h
:532: undefined reference to `QGList::~QGList()'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0x18):/usr/include/c++/3.4/bits/stl_algobase.h
:151: undefined reference to `QPtrCollection::newItem(void*)'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0x20):/usr/include/c++/3.4/bits/stl_algobase.h
:156: undefined reference to `QGList::compareItems(void*, void*)'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTI6QGList+0x8):/usr/X11R6//include/qglist.h:152:
 undefin
ed reference to 

Re: Safer Win32 get_temp_path, os::internal_path

[Angus Leeming]

Jean-Marc Lasgouttes wrote:

Angus Jean-Marc, all, following our earlier discussion, I attach a
Angus little test code to show what I think we need to produce robust
Angus code on Windows. The attached code should compile on all
Angus flavours of Windows = Win95 and should behave gracefully on
Angus older versions for which GetLongPathName is just a stub.



This looks good, if it works.


Well, it works for me. Time will tell if it works for Luis.


Angus Angus (the one of questionable moral character :-P)



To tell, the truth, I suspected it already, and was rather relieved
not to have you in my house last July. I do not know what my
neighbours would have thought.


Funny. I had you down as bohémien rather than a member of the bourgeoisie. 
Anyway, wasn't it your parents' neighbours rather than your own?


Angus (scandalous individual)

Re: Errors building 1.4.0 (?)

[Georg Baum]

Mike Meyer wrote:

 I can tell you what went wrong with the first command: it needs the
 FreeBSDism -pthread to link with the threaded version of libc.

That means we'll need to fix the configure script. Thanks for debugging! Can
you test the attached (untested) patch? Don't forget to run autogen.sh.

I also think that we should set FATAL=1 in qt.m4. That would give a sensible
error message. It is nonsense to compile for qt when the resulting
executable cannot be linked.

 Again, this worked without patching in 1.3.5.

Strange.


GeorgIndex: config/qt.m4
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/config/qt.m4,v
retrieving revision 1.15
diff -u -p -r1.15 qt.m4
--- config/qt.m4	17 Jul 2005 23:40:08 -	1.15
+++ config/qt.m4	13 Sep 2005 12:31:19 -
@@ -114,6 +114,12 @@ AC_DEFUN([QT_CHECK_COMPILE],
 			QT_TRY_LINK($libname)
 			if test -n $qt_cv_libname; then
 break;
+			else
+dnl on FreeBSD we need -pthread for threaded libs
+QT_TRY_LINK($libname -pthread)
+if test -n $qt_cv_libname; then
+	break;
+fi
 			fi
 		done
 

Replacing SHGetFolderPath

[Angus Leeming]

As older flavours of Windows don't support SHGetFolderPath out of the box, 
Jean-Marc has suggested that we use the more cumbersome but also more 
portable SHGetSpecialFolderLocation.


I propose the attached code as a replacement for the existing 
win32_folder_path in src/support/package.C.


$ g++ -I$HOME/lyx/13x/boost -W -Wall -o get_folder get_folder.cpp
$ ./get_folder.exe
Document dir: C:\Documents and Settings\Angus\My Documents
App data dir: C:\Documents and Settings\Angus\Application Data

I'll post this off to Luis Rivera to check that it compiles/runs on his 
older machines.


One thing that is side-stepped by the code at the moment is what should we 
do if string() is returned for either document dir or app data dir? Should 
we assert?


Angus

#include boost/assert.hpp

#include iostream
#include string

# include windows.h
# include shlobj.h  // SHGetFolderPath

using std::string;


// Given a folder ID, returns the folder name (in unix-style format).
// Eg CSIDL_PERSONAL - C:/Documents and Settings/USERNAME/My Documents
string const win32_folder_path(int folder_id)
{
// Microsoft recommend the use of SHGetFolderPath to do
// what we're attempting here, but SHGetFolderPath won't work
// out of the box on older (Win98) flavours of Windows.
// Either we distribute shfolder.dll (fully-redistributable part
// of IE 4) and ensure that we link against it before we link
// shell32.dll, or we use SHGetSpecialFolderLocation instead.
LPITEMIDLIST id_list;
if (SHGetSpecialFolderLocation(0, folder_id, id_list) != S_OK)
return string();

char folder_path[PATH_MAX];
DWORD const result = SHGetPathFromIDList(id_list, folder_path);

LPMALLOC allocator;
if (SHGetMalloc(allocator) == S_OK) {
allocator-Free(id_list);
allocator-Release();
}

return (result == TRUE) ? folder_path : string();
}


int main()
{
std::cout  Document dir:   win32_folder_path(CSIDL_PERSONAL)
   \nApp data dir:   win32_folder_path(CSIDL_APPDATA)
   std::endl;

return 0;
}

Re: Safer Win32 get_temp_path, os::internal_path

[Jean-Marc Lasgouttes]

 Angus == Angus Leeming [EMAIL PROTECTED] writes:

 To tell, the truth, I suspected it already, and was rather relieved
 not to have you in my house last July. I do not know what my
 neighbours would have thought.

Angus Funny. I had you down as bohémien rather than a member of the
Angus bourgeoisie. 

Well, I have some funny bourgeois in my mother's family. And I went
mainly to private catholic schools.

Angus Anyway, wasn't it your parents' neighbours rather than your
Angus own?

God forbid! I would not even have thought about inviting strangers in
this case! We are now in our own appartment, so I only have to care
about my reputation, which is nevertheless a lot, as you will
undoubtlessly agree.

JMarc

Re: LyX 140 configure problem with awk

[Bennett Helm]

On Sep 13, 2005, at 5:16 AM, Jean-Marc Lasgouttes wrote:


The important question is whether the information in lyx.pot is
reasonable. You should see entries like:

msgid \\arabic{enumi}.
msgstr 

(what is important here is the double backslash).


I don't get \\arabic{enumi} in particular. I do get double  
backslashes in the following cases, all of which look reasonable:


\\begin_header
\\begin_document
\\,
\\:
\\;
\\quad
\\qquad
... etc. in QMathDialog.C
\\input
\\include
\\verbatiminput
\\usepackage
\\documentclass
\\selectlanguage

There aren't cases of single backslashes where it looks like doubles  
should be used instead.


Is that what you need to know?

Bennett

Re: Replacing SHGetFolderPath

[Jean-Marc Lasgouttes]

 Angus == Angus Leeming [EMAIL PROTECTED] writes:

Angus As older flavours of Windows don't support SHGetFolderPath out
Angus of the box, Jean-Marc has suggested that we use the more
Angus cumbersome but also more portable SHGetSpecialFolderLocation.

Angus I propose the attached code as a replacement for the existing
Angus win32_folder_path in src/support/package.C.

Looks good.

Angus One thing that is side-stepped by the code at the moment is
Angus what should we do if string() is returned for either document
Angus dir or app data dir? Should we assert?

If the app data dir is empty, we will eventually exit somewhere, I
guess. As far as the document data dir is concerned, I am not sure
this really matters.

JMarc

Re: Safer Win32 get_temp_path, os::internal_path

[Angus Leeming]

Jean-Marc Lasgouttes wrote:
 Angus Anyway, wasn't it your parents' neighbours rather than your
 Angus own?

 God forbid! I would not even have thought about inviting strangers in
 this case! We are now in our own appartment, so I only have to care
 about my reputation, which is nevertheless a lot, as you will
 undoubtlessly agree.

Oh, I don't know. I'm quite careless about the reputation of others,
apparently, which is what got me into trouble on the lyx-users' list in
the first place :-P

-- 
Angus

Re: LyX 140 configure problem with awk

[Jean-Marc Lasgouttes]

 Bennett == Bennett Helm [EMAIL PROTECTED] writes:

Bennett On Sep 13, 2005, at 5:16 AM, Jean-Marc Lasgouttes wrote:
 The important question is whether the information in lyx.pot is
 reasonable. You should see entries like:
 
 msgid \\arabic{enumi}. msgstr 
 
 (what is important here is the double backslash).

Bennett I don't get \\arabic{enumi} in particular. I do get double
Bennett backslashes in the following cases, all of which look
Bennett reasonable:

Bennett There aren't cases of single backslashes where it looks like
Bennett doubles should be used instead.

Bennett Is that what you need to know?

What I am interested in are entries related to LabelString in layout
files, so my enumi example was not very well chosen.

For example, numarticle.inc says

Style Section
LabelType Counter
LabelCounter  section
LabelString   \arabic{section}
LabelStringAppendix   \Alph{section}
TocLevel  1
End

so you should find the strings \\arabic{section} and
\\Alph{section} in the .pot file.

JMarc

Re: Replacing SHGetFolderPath

[Angus Leeming]

Jean-Marc Lasgouttes wrote:
 Angus As older flavours of Windows don't support SHGetFolderPath out
 Angus of the box, Jean-Marc has suggested that we use the more
 Angus cumbersome but also more portable SHGetSpecialFolderLocation.
 
 Angus I propose the attached code as a replacement for the existing
 Angus win32_folder_path in src/support/package.C.
 
 Looks good.

One thing I note from the MSDN website is:

=
http://msdn.microsoft.com/library/en-us/shellcc/platform/shell/reference/functions/shgetmalloc.asp
SHGetMalloc Function

Not currently supported.

Remarks
This function should no longer be used. Use the CoTaskMemFree and
CoTaskMemAlloc functions in its place.
=

I've just ignored this advice, but this blog article
http://blogs.msdn.com/oldnewthing/archive/2004/07/05/173226.aspx

suggests that all should be fine anyway (since I think that we link against
ole32.dll)

Ain't life on Windoze complicated!
-- 
Angus

Re: LyX 140 configure problem with awk

[Bennett Helm]

On Sep 13, 2005, at 8:56 AM, Jean-Marc Lasgouttes wrote:


What I am interested in are entries related to LabelString in layout
files, so my enumi example was not very well chosen.


Actually, it was my fault: I was stupidly looking at lyx.pot from the  
wrong build.



For example, numarticle.inc says

Style Section
LabelType Counter
LabelCounter  section
LabelString   \arabic{section}
LabelStringAppendix   \Alph{section}
TocLevel  1
End

so you should find the strings \\arabic{section} and
\\Alph{section} in the .pot file.


Yes -- they're all there (at least every one I checked!).

Bennett

Re: Replacing SHGetFolderPath

[Jean-Marc Lasgouttes]

 Angus == Angus Leeming [EMAIL PROTECTED] writes:

Angus One thing I note from the MSDN website is:

Angus =
Angus 
http://msdn.microsoft.com/library/en-us/shellcc/platform/shell/reference/functions/shgetmalloc.asp
Angus SHGetMalloc Function

Angus Not currently supported.

Angus Remarks This function should no longer be used. Use the
Angus CoTaskMemFree and CoTaskMemAlloc functions in its place.
Angus =

Angus I've just ignored this advice, but this blog article
Angus http://blogs.msdn.com/oldnewthing/archive/2004/07/05/173226.aspx

Angus suggests that all should be fine anyway (since I think that we
Angus link against ole32.dll)

Indeed...

JMarc

[PATCH] bug 2010: Entering/leaving text inset requires two cursor left/right keystrokes

[Jean-Marc Lasgouttes]

The following patch fixes bug 2010:
http://bugzilla.lyx.org/show_bug.cgi?id=2010

This was an unintended consequence of the work Juergen V. did to allow
placing the cursor in front of a displayed inset.

Please test.

JMarc

Index: src/ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/ChangeLog,v
retrieving revision 1.2277
diff -u -p -r1.2277 ChangeLog
--- src/ChangeLog	10 Sep 2005 13:29:43 -	1.2277
+++ src/ChangeLog	13 Sep 2005 13:44:39 -
@@ -1,3 +1,8 @@
+2005-09-13  Jean-Marc Lasgouttes  [EMAIL PROTECTED]
+
+	* text2.C (cursorEnd, cursorRight): do not set boundary to true at
+	end of paragraph. (bug 2010)
+
 2005-09-10  Georg Baum  [EMAIL PROTECTED]
 
 	* text.C (leftMargin): check for inInset == 0
Index: src/text2.C
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/text2.C,v
retrieving revision 1.627
diff -u -p -r1.627 text2.C
--- src/text2.C	5 Sep 2005 06:50:42 -	1.627
+++ src/text2.C	13 Sep 2005 13:44:40 -
@@ -498,12 +498,13 @@ void LyXText::cursorEnd(LCursor  cur)
 		// empty text, end-1 is no valid position
 		return;
 	bool boundary = false;
-	if (!cur.paragraph().isLineSeparator(end-1) 
-	!cur.paragraph().isNewline(end-1))
-	{
-		boundary = true;
-	} else if (end != cur.lastpos())
-		--end;
+	if (end != cur.lastpos()) {
+		if (!cur.paragraph().isLineSeparator(end-1) 
+			 !cur.paragraph().isNewline(end-1))
+			boundary = true;
+		else
+			--end;
+	}
 	setCursor(cur, cur.pit(), end, true, boundary);
 }
 
@@ -1021,6 +1022,7 @@ bool LyXText::cursorRight(LCursor  cur)
 		bool updateNeeded = false;
 		if (!checkAndActivateInset(cur, true)) {
 			if (cur.textRow().endpos() == (cur.pos() + 1) 
+			!cur.lastpos() == (cur.pos() + 1) 
 			!cur.paragraph().isLineSeparator(cur.pos()) 
 			!cur.paragraph().isNewline(cur.pos()))
 			{

Re: LyX 140 configure problem with awk

[Jean-Marc Lasgouttes]

 Bennett == Bennett Helm [EMAIL PROTECTED] writes:

 so you should find the strings \\arabic{section} and
 \\Alph{section} in the .pot file.

Bennett Yes -- they're all there (at least every one I checked!).

Very good. I'll apply the patch.

JMarc

Re: [PATCH] bug 1941: toolbars disabled after dialog action

[Georg Baum]

Jean-Marc Lasgouttes wrote:

 Georg == Georg Baum
 [EMAIL PROTECTED]
 writes:
 
 Georg I tried to understand where in the code that happens, but
 Georg failed. Could you please tell me the place where the main
 Georg window without focus prevents the icons from being enabled?
 
 In LyXFunc::getStatus, you will find:

Thanks, now I see it (I was searching in frontends/qt2)

 It is not really qt-only. The same code would trigger for a Cocoa
 port, or a Gtk/osx.

I see. Because XFormsView::hasFocus() and GView::hasFocus() are too stupid
we don't see the bug there.

 Georg The right fix is IMHO to fix the focus problem. Why is it not
 Georg possible to give the focus to the main window after a dialog is
 Georg closed?
 
 This is what I tried first, actually: updating the tool/menu bars after
 the window has been hidden in Dialog::hide. It did not work, and I do
 not understand why.

The problem is that the main window still has no focus when Dialog::hide is
called. I tried to activate it with QMainWindow::setFocus() and
QMainWindow::show(), but neither worked. I'll investigate further.


Georg

Re: [PATCH] bug 1941: toolbars disabled after dialog action

[Jean-Marc Lasgouttes]

 Georg == Georg Baum [EMAIL PROTECTED] writes:

Georg The problem is that the main window still has no focus when
Georg Dialog::hide is called. I tried to activate it with
Georg QMainWindow::setFocus() and QMainWindow::show(), but neither
Georg worked. I'll investigate further.

Thanks.

JMarc

Re: Errors building 1.4.0 (?)

[Georg Baum]

Am Dienstag, 13. September 2005 14:30 schrieb John Levon:
 On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:
  It looks like configure does not find the qt libary (but it seems to find
  the headers). This looks like a configure bug: It should stop with an
  error message if the library cannot be found.

 IIRC Lars changed this.

I now remember that, too. I also remember that I added error messages for moc 
and uic, so here comes the logical extension.

Mike, could you please test whether configure aborts with this patch (and 
without the other one) if it cannot find the qt library?


Georg
Index: ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/ChangeLog,v
retrieving revision 1.1023
diff -u -p -r1.1023 ChangeLog
--- ChangeLog	18 Jul 2005 15:49:56 -	1.1023
+++ ChangeLog	13 Sep 2005 14:23:16 -
@@ -1,3 +1,8 @@
+2005-09-13  Georg Baum  [EMAIL PROTECTED]
+
+	* configure.ac: error out if qt frontend is choosen but no qt library
+	was found
+
 2005-07-18  Lars Gullik Bjønnes  [EMAIL PROTECTED]
 
 	* configure.ac: version back to 1.4.0cvs 
Index: configure.ac
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/configure.ac,v
retrieving revision 1.58
diff -u -p -r1.58 configure.ac
--- configure.ac	18 Jul 2005 15:49:57 -	1.58
+++ configure.ac	13 Sep 2005 14:23:16 -
@@ -216,6 +219,9 @@ dnl qt build will fail without moc or ui
 	  fi
 	  if test -z $UIC; then
 	LYX_ERROR([uic binary not found !])
+	  fi
+	  if test -z $QT_LIB; then
+	LYX_ERROR([qt library not found !])
 	  fi
   ;;
 *)

Re: Errors building 1.4.0 (?)

[Jean-Marc Lasgouttes]

 Georg == Georg Baum [EMAIL PROTECTED] writes:

Georg Am Dienstag, 13. September 2005 14:30 schrieb John Levon:
 On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:  It
 looks like configure does not find the qt libary (but it seems to
 find  the headers). This looks like a configure bug: It should
 stop with an  error message if the library cannot be found.
 
 IIRC Lars changed this.

Georg I now remember that, too. I also remember that I added error
Georg messages for moc and uic, so here comes the logical extension.

Georg Mike, could you please test whether configure aborts with this
Georg patch (and without the other one) if it cannot find the qt
Georg library?

The patch looks good. Concerning the bug itself, could it be related
to
http://marc.theaimsgroup.com/?l=lyx-develm=112490112909207w=2

JMarc

Re: [PATCH] bug 2010: Entering/leaving text inset requires two cursor left/right keystrokes

[Angus Leeming]

Jean-Marc Lasgouttes wrote:

 The following patch fixes bug 2010:
 http://bugzilla.lyx.org/show_bug.cgi?id=2010
 
 This was an unintended consequence of the work Juergen V. did to allow
 placing the cursor in front of a displayed inset.
 
 Please test.
 
 JMarc

why not
+   cur.lastpos() != (cur.pos() + 1) 
?

@@ -1021,6 +1022,7 @@ bool LyXText::cursorRight(LCursor  cur)
bool updateNeeded = false;
if (!checkAndActivateInset(cur, true)) {
if (cur.textRow().endpos() == (cur.pos() + 1) 
+   !cur.lastpos() == (cur.pos() + 1) 
!cur.paragraph().isLineSeparator(cur.pos()) 
!cur.paragraph().isNewline(cur.pos()))

-- 
Angus

Re: [PATCH] bug 2010: Entering/leaving text inset requires two cursor left/right keystrokes

[Jean-Marc Lasgouttes]

 Angus == Angus Leeming [EMAIL PROTECTED] writes:

Angus why not + cur.lastpos() != (cur.pos() + 1)  ?

Oops. I am not even sure that the test meant what I thought it meant.

What about the following?

JMarc

Index: src/ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/ChangeLog,v
retrieving revision 1.2277
diff -u -p -r1.2277 ChangeLog
--- src/ChangeLog	10 Sep 2005 13:29:43 -	1.2277
+++ src/ChangeLog	13 Sep 2005 14:56:01 -
@@ -1,3 +1,8 @@
+2005-09-13  Jean-Marc Lasgouttes  [EMAIL PROTECTED]
+
+	* text2.C (cursorEnd, cursorRight): do not set boundary to true at
+	end of paragraph. (bug 2010)
+
 2005-09-10  Georg Baum  [EMAIL PROTECTED]
 
 	* text.C (leftMargin): check for inInset == 0
Index: src/text2.C
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/text2.C,v
retrieving revision 1.627
diff -u -p -r1.627 text2.C
--- src/text2.C	5 Sep 2005 06:50:42 -	1.627
+++ src/text2.C	13 Sep 2005 14:56:01 -
@@ -498,12 +498,13 @@ void LyXText::cursorEnd(LCursor  cur)
 		// empty text, end-1 is no valid position
 		return;
 	bool boundary = false;
-	if (!cur.paragraph().isLineSeparator(end-1) 
-	!cur.paragraph().isNewline(end-1))
-	{
-		boundary = true;
-	} else if (end != cur.lastpos())
-		--end;
+	if (end != cur.lastpos()) {
+		if (!cur.paragraph().isLineSeparator(end-1) 
+			 !cur.paragraph().isNewline(end-1))
+			boundary = true;
+		else
+			--end;
+	}
 	setCursor(cur, cur.pit(), end, true, boundary);
 }
 
@@ -1020,7 +1021,8 @@ bool LyXText::cursorRight(LCursor  cur)
 	if (cur.pos() != cur.lastpos()) {
 		bool updateNeeded = false;
 		if (!checkAndActivateInset(cur, true)) {
-			if (cur.textRow().endpos() == (cur.pos() + 1) 
+			if (cur.textRow().endpos() == cur.pos() + 1 
+			cur.lastpos() != cur.pos() + 1 
 			!cur.paragraph().isLineSeparator(cur.pos()) 
 			!cur.paragraph().isNewline(cur.pos()))
 			{

[PATCH] bug 1926: Paragraph DocIterator::paragraph() by failing check inTexted() in file dociterator.C:144

[Jean-Marc Lasgouttes]

This patch fixes the crash described in 
http://bugzilla.lyx.org/show_bug.cgi?id=1926

  ``Start a new doc, and start a math formula. Enter it, and leaving
  the cursor inside the formula, start another new doc. Move back to
  the old doc via the View menu and:

  Assertion triggered in Paragraph DocIterator::paragraph() by
  failing check inTexted() in file dociterator.C:144''

Testing appreciated, although it seems trivially correct (I tried to
use LFUN_GETLAYOUT but it (1) leaves the current math inset and (2)
displays the result of the lfun in the minibuffer).

JMarc

Index: src/BufferView_pimpl.C
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/BufferView_pimpl.C,v
retrieving revision 1.594
diff -u -p -r1.594 BufferView_pimpl.C
--- src/BufferView_pimpl.C	7 Sep 2005 10:00:17 -	1.594
+++ src/BufferView_pimpl.C	13 Sep 2005 15:21:25 -
@@ -386,8 +386,20 @@ void BufferView::Pimpl::setBuffer(Buffer
 	owner_-updateWindowTitle();
 
 	// This is done after the layout combox has been populated
-	if (buffer_)
-		owner_-setLayout(cursor_.paragraph().layout()-name());
+	if (buffer_) {
+		size_t i = cursor_.depth() - 1;
+		// we know we'll eventually find a paragraph
+		while (true) {
+			CursorSlice const  slice = cursor_[i];
+			if (!slice.inset().inMathed()) {
+LyXLayout_ptr const layout = slice.paragraph().layout();
+owner_-setLayout(layout-name());
+break;
+			}
+			BOOST_ASSERT(i0);
+			--i;
+		}
+	}	
 
 	if (buffer_  lyx::graphics::Previews::status() != LyXRC::PREVIEW_OFF)
 		lyx::graphics::Previews::get().generateBufferPreviews(*buffer_);
Index: src/ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/ChangeLog,v
retrieving revision 1.2277
diff -u -p -r1.2277 ChangeLog
--- src/ChangeLog	10 Sep 2005 13:29:43 -	1.2277
+++ src/ChangeLog	13 Sep 2005 15:21:26 -
@@ -1,3 +1,9 @@
+2005-09-13  Jean-Marc Lasgouttes  [EMAIL PROTECTED]
+
+	* BufferView_pimpl.C (setBuffer): when one wants to set the layout
+	combox, it is necessary to search for the topmost paragraph in the
+	cursor stack (bug 1926)
+
 2005-09-10  Georg Baum  [EMAIL PROTECTED]
 
 	* text.C (leftMargin): check for inInset == 0

Re: large executable

[Lars Gullik Bjønnes]

Pol [EMAIL PROTECTED] writes:

| Jean-Marc Lasgouttes wrote:
|  what does size ./lyx day? Here (gcc 3.4.1) I get:
|  fantomas: size src/lyx
| textdata bss dec hex filename
|  6083056   17680   35060 6135796  5d9ff4 src/lyx
| 
| 
| % size /usr/local/bin/lyx
|   textdata bss  dec hex   filename
|11134899 394540   35332 11564771b076e3 /usr/local/bin/lyx

Remind me on the platform again please.

Re: Errors building 1.4.0 (?)

[Lars Gullik Bjønnes]

John Levon [EMAIL PROTECTED] writes:

| On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:
| 
|  It looks like configure does not find the qt libary (but it seems to find
|  the headers). This looks like a configure bug: It should stop with an error
|  message if the library cannot be found.
| 
| IIRC Lars changed this.

firm beliver in running to completion.

Re: bug in tex2lyx windows

[Lars Gullik Bjønnes]

Luis Rivera [EMAIL PROTECTED] writes:

| Hello,
| 
| I tried to run tex2lyx on win95/98, without success.  The programs
| crashed crying as follows:
| 
| On win95
| 
| The TEX2LYX.EXE file is linked to missing export
| KERNEL32.DLL:GetFileAttributesExA

Can't we just declare that we do not support Win95?

How old is XP now? 4 years?

Re: Errors building 1.4.0 (?)

[Lars Gullik Bjønnes]

Georg Baum [EMAIL PROTECTED] writes:

| Am Dienstag, 13. September 2005 14:30 schrieb John Levon:
|  On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:
|   It looks like configure does not find the qt libary (but it seems to find
|   the headers). This looks like a configure bug: It should stop with an
|   error message if the library cannot be found.
| 
|  IIRC Lars changed this.
| 
| I now remember that, too. I also remember that I added error messages for moc 
| and uic, so here comes the logical extension.
| 
| Mike, could you please test whether configure aborts with this patch (and 
| without the other one) if it cannot find the qt library?

I hate configures that aborts. IMHO it should run to completion and
tell me about all important stuff that went wrong.

Re: [PATCH] bug 2010: Entering/leaving text inset requires two cursor left/right keystrokes

[Angus Leeming]

Jean-Marc Lasgouttes wrote:

 Angus why not + cur.lastpos() != (cur.pos() + 1)  ?
 
 Oops. I am not even sure that the test meant what I thought it meant.
 
 What about the following?

Isn't that semantically the same as:

@@ -1020,7 +1021,8 @@ bool LyXText::cursorRight(LCursor  cur)
if (cur.pos() != cur.lastpos()) {
bool updateNeeded = false;
if (!checkAndActivateInset(cur, true)) {
-   if (cur.textRow().endpos() == (cur.pos() + 1) 
+   if (cur.textRow().endpos() == cur.pos() + 1 
+   cur.textRow().endpos() != cur.lastpos()
!cur.paragraph().isLineSeparator(cur.pos()) 
!cur.paragraph().isNewline(cur.pos()))

Is that what you mean? Is it clearer?

-- 
Angus

Re: request for help with CVS access to LyX sources

[Lars Gullik Bjønnes]

Jean-Marc Lasgouttes [EMAIL PROTECTED] writes:

|  Lars == Lars Gullik Bjønnes [EMAIL PROTECTED] writes:
| 
| Lars We should take the oppurtunito to move the devel www-site into
| Lars the user one. So that we only have one...
| 
| Do you want to keep the web sites separates from the user POV? Merging
| them will be a lot of work.

We can keep them separate from the users pov, but have them in the
same cvs module, and in the same virtual host on the web server.

Re: Errors building 1.4.0 (?)

[Angus Leeming]

Lars Gullik Bjønnes wrote:

 John Levon writes:
 | On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:
 |
 |  It looks like configure does not find the qt libary (but it seems to
 |  find the headers). This looks like a configure bug: It should stop
 |  with an error message if the library cannot be found.
 |
 | IIRC Lars changed this.

 firm beliver in running to completion.

Can you run to completion but then not generate either the Makefiles or
config.h? That would appear to be the best of both worlds.

Ie LYX_ERROR sets a boolean flag that is queried before config.h, Makefiles
are generated and the configure script exits then.

-- 
Angus

Re: bug in tex2lyx windows

[Angus Leeming]

Lars Gullik Bjønnes wrote:
 Luis Rivera writes:
 | Hello,
 |
 | I tried to run tex2lyx on win95/98, without success.  The programs
 | crashed crying as follows:
 |
 | On win95
 |
 | The TEX2LYX.EXE file is linked to missing export
 | KERNEL32.DLL:GetFileAttributesExA

 Can't we just declare that we do not support Win95?

Sure, but what do we gain by such a declaration. AFAICT, the only piece of
code we need to change to support Win98 is the call to SHFetFolderPath.
Thereafter, the only piece of code we need to change to support Win95 is
the call to GetFileAttributesEx, GetLongPathName. Reviewing the existing
code shows that it's unsafe anyway (no handling of errors), so a rewrite
is needed anyway.

 How old is XP now? 4 years?

I'm more interested in the intellectual challenge really :)

-- 
Angus

Re: request for help with CVS access to LyX sources

[Jean-Marc Lasgouttes]

 Lars == Lars Gullik Bjønnes [EMAIL PROTECTED] writes:

Lars We can keep them separate from the users pov, but have them in
Lars the same cvs module, and in the same virtual host on the web
Lars server.

That would be OK.

JMarc

Re: Errors building 1.4.0 (?)

[Jean-Marc Lasgouttes]

 Lars == Lars Gullik Bjønnes [EMAIL PROTECTED] writes:

Lars I hate configures that aborts. IMHO it should run to completion
Lars and tell me about all important stuff that went wrong.

Georg's patch calls LYX_ERROR, so it should be OK.

JMarc

Re: [PATCH] bug 2010: Entering/leaving text inset requires two cursor left/right keystrokes

[Jean-Marc Lasgouttes]

 Angus == Angus Leeming [EMAIL PROTECTED] writes:

Angus Jean-Marc Lasgouttes wrote: why not + cur.lastpos() !=
Angus (cur.pos() + 1)  ?
  Oops. I am not even sure that the test meant what I thought it
 meant.
 
 What about the following?

Angus Isn't that semantically the same as:

Yes it is.

Angus Is that what you mean? Is it clearer?

Yes, slightly.

JMarc

Index: src/ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/ChangeLog,v
retrieving revision 1.2277
diff -u -p -r1.2277 ChangeLog
--- src/ChangeLog	10 Sep 2005 13:29:43 -	1.2277
+++ src/ChangeLog	13 Sep 2005 16:32:14 -
@@ -1,3 +1,8 @@
+2005-09-13  Jean-Marc Lasgouttes  [EMAIL PROTECTED]
+
+	* text2.C (cursorEnd, cursorRight): do not set boundary to true at
+	end of paragraph. (bug 2010)
+
 2005-09-10  Georg Baum  [EMAIL PROTECTED]
 
 	* text.C (leftMargin): check for inInset == 0
Index: src/text2.C
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/text2.C,v
retrieving revision 1.627
diff -u -p -r1.627 text2.C
--- src/text2.C	5 Sep 2005 06:50:42 -	1.627
+++ src/text2.C	13 Sep 2005 16:32:14 -
@@ -498,12 +498,13 @@ void LyXText::cursorEnd(LCursor  cur)
 		// empty text, end-1 is no valid position
 		return;
 	bool boundary = false;
-	if (!cur.paragraph().isLineSeparator(end-1) 
-	!cur.paragraph().isNewline(end-1))
-	{
-		boundary = true;
-	} else if (end != cur.lastpos())
-		--end;
+	if (end != cur.lastpos()) {
+		if (!cur.paragraph().isLineSeparator(end-1) 
+			 !cur.paragraph().isNewline(end-1))
+			boundary = true;
+		else
+			--end;
+	}
 	setCursor(cur, cur.pit(), end, true, boundary);
 }
 
@@ -1020,7 +1021,8 @@ bool LyXText::cursorRight(LCursor  cur)
 	if (cur.pos() != cur.lastpos()) {
 		bool updateNeeded = false;
 		if (!checkAndActivateInset(cur, true)) {
-			if (cur.textRow().endpos() == (cur.pos() + 1) 
+			if (cur.textRow().endpos() == cur.pos() + 1 
+			cur.textRow().endpos() != cur.lastpos() 
 			!cur.paragraph().isLineSeparator(cur.pos()) 
 			!cur.paragraph().isNewline(cur.pos()))
 			{

Re: large executable

[Pol]

Lars Gullik Bjønnes wrote:
 |
 | % size /usr/local/bin/lyx
 |   textdata bss  dec hex   filename
 |11134899 394540   35332 11564771b076e3 
/usr/local/bin/lyx

 Remind me on the platform again please.


It is debian/unstable + tetex ver. 3.0 packages from the 
debian/experimental section,


--
pol

Re: large executable

[Pol]

Lars Gullik Bjønnes wrote:
| 
| % size /usr/local/bin/lyx

|   textdata bss  dec hex   filename
|11134899 394540   35332 11564771b076e3 /usr/local/bin/lyx
Remind me on the platform again please.



It is debian/unstable + tetex ver. 3.0 packages from the 
debian/experimental section,


--
pol

Re: large executable

[Michael Gerz]

Pol wrote:


--
pol

Pol, could you please change the default settings of your email client? 
Everytime I receive an email from you, Thunderbird asks me to confirm 
the reception :-(


Michael

Re: Errors building 1.4.0 (?)

[Mike Meyer]

[Apologies to George, who's seeing this twice.]

In [EMAIL PROTECTED], Bennett Helm [EMAIL PROTECTED] typed:
 On Sep 13, 2005, at 10:03 AM, Mike Meyer wrote:
  the configure script walks your $PATH looking for sgmltools and/or
  db2dvi to decide whether or not to configure docbook support. Fink
  probably put those in /sw/bin.
  If that's the case, you need to add /sw/bin the path used by LyX. On
  tiger (sorry, I'm a Mac newbie and haven't dealt with anything else)
  you edit ~/.MacOSX/environment.plist. There's a GUI plist editor
  available, but the file is just XML text, so you can use your favorite
  text editor on it as well.
 Actually, if all you want is to add /sw/bin to the path that LyX  
 uses, you should do it from within LyX itself: LyX  Preferences   
 PATH  PATH Prefix. (For LyX/Mac-1.3.6, the default settings already  
 have /sw/bin there.) If you add something to the PATH setting in  
 ~/.MacOSX/environment.plist, that will apply to *every* OS X  
 application, which is probably overkill and potentially a security risk.

Whether or not it's overkill depends on how many applications you have
that want to use programs from fink (or, in my case, darwinports). I
have a couple, so doing things this way is easier for me.

I found this method while trying to figure out why LyX wasn't finding
my darwinports software. That I didn't find the LyX  Preferences 
PATH  PATH Prefix methods suggests a doc bug. This appears to be a
Mac issue - I don't find the PATH entries in preferences in 1.3.5 on
my Unix box. Might just be my not having taken time to read the docs
properly, though.

mike

-- 
Mike Meyer [EMAIL PROTECTED]  http://www.mired.org/consulting.html
Independent Network/Unix/Perforce consultant, email for more information.

Re: byte compilation of lyx2lyx files

[Luis Rivera]

Jean-Marc Lasgouttes [EMAIL PROTECTED] writes:

 
  Angus == Angus Leeming [EMAIL PROTECTED] writes:
 
 Angus Luis didn't provide much information. Just the second
paragraph
 Angus here, IIRC:
 Angus http://article.gmane.org/gmane.editors.lyx.devel:47931
 
 Yes, I saw that.
 
 Angus I'll drop him a line.
 
 Thanks.
 
 JMarc
 

JMarc,

You'll find the log below:



This file contains any messages produced by compilers while
running configure, to aid debugging if configure makes a mistake.

It was created by lyx configure 1.4.0pre1, which was
generated by GNU Autoconf 2.56.  Invocation command line was

  $ ../configure --disable-debug --enable-optimization
--prefix=/home/unknown/lyx-1.4.0pre1/build-tex2lyx/LyX-1.4
--with-frontend=foo

## - ##
## Platform. ##
## - ##

hostname = PC-0808
uname -m = i686
uname -r = 1.0.10(0.46/3/2)
uname -s = MINGW32_98-4.10
uname -v = 2004-03-15 07:17

/usr/bin/uname -p = unknown
/bin/uname -X = unknown

/bin/arch  = unknown
/usr/bin/arch -k   = unknown
/usr/convex/getsysinfo = unknown
hostinfo   = unknown
/bin/machine   = unknown
/usr/bin/oslevel   = unknown
/bin/universe  = unknown

PATH: .
PATH: /usr/local/bin
PATH: /mingw/bin
PATH: /bin
PATH: /c/CAR/TEXMF/MIKTEX/BIN
PATH: /c/PERL/BIN/
PATH: /c/TEXMF/MIKTEX/BIN
PATH: /c/WINDOWS
PATH: /c/WINDOWS/COMMAND
PATH: /c/EMTEX/BIN
PATH: /c/TINYAPPS/UFO
PATH: /c/PP/BIN/OS2


## --- ##
## Core tests. ##
## --- ##

configure:1584: checking whether to enable maintainer-specific portions
of Makefiles
configure:1593: result: no
configure:1614: checking build system type
configure:1632: result: i686-pc-mingw32
configure:1640: checking host system type
configure:1654: result: i686-pc-mingw32
configure:1662: checking target system type
configure:1676: result: i686-pc-mingw32
configure:1692: checking for install target ... 
configure:1711: result: lyx
configure:1728: checking for a BSD-compatible install
configure:1782: result: /bin/install -c
configure:1793: checking whether build environment is sane
configure:1836: result: yes
configure:1869: checking for gawk
configure:1885: found /bin/gawk
configure:1895: result: gawk
configure:1905: checking whether make sets $(MAKE)
configure:1925: result: yes
configure:2087: checking whether make sets $(MAKE)
configure:2107: result: yes
configure:2128: checking for a BSD-compatible install
configure:2182: result: /bin/install -c
configure:2200: checking for kpsewhich
configure:2216: found /c/CAR/TEXMF/MIKTEX/BIN/kpsewhich
configure:2227: result: kpsewhich
configure:2245: checking for gm4
configure:2274: result: no
configure:2245: checking for gnum4
configure:2274: result: no
configure:2245: checking for m4
configure:2261: found /bin/m4
configure:2271: result: m4
configure:2315: checking for a Python interpreter with version = 1.5.2
configure:2334: python -c import sys, string # split strings by '.' and
convert
to numeric. Append some zeros # because we need at least 4 digits for
the hex
conversion. minver = map(int, string.split('1.5.2', '.')) + [0, 0, 0]
minverhex
= 0 for i in xrange(0, 4): minverhex = (minverhex  8) + minver[i]
sys.exit(sys.hexversion  minverhex)
../configure: python: command not found
configure:2337: $? = 127
configure:2334: python2 -c import sys, string # split strings by '.'
and convert
to numeric. Append some zeros # because we need at least 4 digits for
the hex
conversion. minver = map(int, string.split('1.5.2', '.')) + [0, 0, 0]
minverhex
= 0 for i in xrange(0, 4): minverhex = (minverhex  8) + minver[i]
sys.exit(sys.hexversion  minverhex)
../configure: python2: command not found
configure:2337: $? = 127
configure:2334: python2.2 -c import sys, string # split strings by '.'
and
convert to numeric. Append some zeros # because we need at least 4
digits for
the hex conversion. minver = map(int, string.split('1.5.2', '.')) + [0,
0, 0]
minverhex = 0 for i in xrange(0, 4): minverhex = (minverhex  8) +
minver[i]
sys.exit(sys.hexversion  minverhex)
../configure: python2.2: command not found
configure:2337: $? = 127
configure:2334: python2.1 -c import sys, string # split strings by '.'
and
convert to numeric. Append some zeros # because we need at least 4
digits for
the hex conversion. minver = map(int, string.split('1.5.2', '.')) + [0,
0, 0]
minverhex = 0 for i in xrange(0, 4): minverhex = (minverhex  8) +
minver[i]
sys.exit(sys.hexversion  minverhex)
../configure: python2.1: command not found
configure:2337: $? = 127
configure:2334: python2.0 -c import sys, string # split strings by '.'
and
convert to numeric. Append some zeros # because we need at least 4
digits for
the hex conversion. minver = map(int, string.split('1.5.2', '.')) + [0,
0, 0]
minverhex = 0 for i in xrange(0, 4): minverhex = (minverhex  8) +
minver[i]
sys.exit(sys.hexversion  minverhex)
../configure: python2.0: command not found
configure:2337: $? = 127
configure:2334: python1.6 -c import 

Daily lyx builds are now set up on ftp.sylvan.com

[Kayvan A. Sylvan]

Hi folks,

I fixed up the FTP site now.

The latest builds for Redhat Fedora Core 4 (x86_64) are at:

ftp://ftp.sylvan.com/pub/lyx/devel/RHFC4

Best regards,

---Kayvan
-- 
Kayvan A. Sylvan  | Proud husband of   | Father to my kids:
Sylvan Associates, Inc.   | Laura Isabella Sylvan  | Katherine Yelena (8/8/89)
http://sylvan.com/~kayvan | crown of her husband | Robin Gregory (2/28/92)

Re: Errors building 1.4.0 (?)

[Mike Meyer]

In [EMAIL PROTECTED], Georg Baum [EMAIL PROTECTED] typed:
 Am Dienstag, 13. September 2005 14:30 schrieb John Levon:
  On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:
   It looks like configure does not find the qt libary (but it seems to find
   the headers). This looks like a configure bug: It should stop with an
   error message if the library cannot be found.
 
  IIRC Lars changed this.
 
 I now remember that, too. I also remember that I added error messages for moc 
 and uic, so here comes the logical extension.
 
 Mike, could you please test whether configure aborts with this patch (and 
 without the other one) if it cannot find the qt library?

Ok, I installed just this patch. After doing autogen.sh and configure,
it ends with the line qt library not found. So you've fixed that bug.

Now, I'll keep trying to build the thing.

Thanks,
mike

 
 Georg
 Index: ChangeLog
 ===
 RCS file: /usr/local/lyx/cvsroot/lyx-devel/ChangeLog,v
 retrieving revision 1.1023
 diff -u -p -r1.1023 ChangeLog
 --- ChangeLog 18 Jul 2005 15:49:56 -  1.1023
 +++ ChangeLog 13 Sep 2005 14:23:16 -
 @@ -1,3 +1,8 @@
 +2005-09-13  Georg Baum  [EMAIL PROTECTED]
 +
 + * configure.ac: error out if qt frontend is choosen but no qt library
 + was found
 +
  2005-07-18  Lars Gullik Bjønnes  [EMAIL PROTECTED]
  
   * configure.ac: version back to 1.4.0cvs 
 Index: configure.ac
 ===
 RCS file: /usr/local/lyx/cvsroot/lyx-devel/configure.ac,v
 retrieving revision 1.58
 diff -u -p -r1.58 configure.ac
 --- configure.ac  18 Jul 2005 15:49:57 -  1.58
 +++ configure.ac  13 Sep 2005 14:23:16 -
 @@ -216,6 +219,9 @@ dnl qt build will fail without moc or ui
 fi
 if test -z $UIC; then
   LYX_ERROR([uic binary not found !])
 +   fi
 +   if test -z $QT_LIB; then
 + LYX_ERROR([qt library not found !])
 fi
;;
  *)

-- 
Mike Meyer [EMAIL PROTECTED]  http://www.mired.org/consulting.html
Independent Network/Unix/Perforce consultant, email for more information.

Re: Errors building 1.4.0 (?)

[Mike Meyer]

In [EMAIL PROTECTED], Jean-Marc Lasgouttes [EMAIL PROTECTED] typed:
  Georg == Georg Baum [EMAIL PROTECTED] writes:
 
 Georg Am Dienstag, 13. September 2005 14:30 schrieb John Levon:
  On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:  It
  looks like configure does not find the qt libary (but it seems to
  find  the headers). This looks like a configure bug: It should
  stop with an  error message if the library cannot be found.
  
  IIRC Lars changed this.
 
 Georg I now remember that, too. I also remember that I added error
 Georg messages for moc and uic, so here comes the logical extension.
 
 Georg Mike, could you please test whether configure aborts with this
 Georg patch (and without the other one) if it cannot find the qt
 Georg library?
 
 The patch looks good. Concerning the bug itself, could it be related
 to
 http://marc.theaimsgroup.com/?l=lyx-develm=112490112909207w=2

Looks to be the same kind of problem. Both systems require magic flags
at compile time to get to the threads versions of libraries, and the
configure system isn't providing them. They are different flags, though.

Since LyX checks for the non-threaded libraries, maybe the configure
script should grow a --with-thread-flags option (or family of
options), and only try the threads libraries if that (or one of them)
is set?

mike
-- 
Mike Meyer [EMAIL PROTECTED]  http://www.mired.org/consulting.html
Independent Network/Unix/Perforce consultant, email for more information.

Re: large executable

[Pol]

Michael Gerz wrote:
Pol, could you please change the default settings of your email client? 
Everytime I receive an email from you, Thunderbird asks me to confirm 
the reception :-(


sorry, i have been trying to disable such feature, but it appears it is 
already off.

On my account settings (mozilla 1.7), option when sending messages,
always request a return receipt is  unchecked.   Any hints to fix that?
thank you
--
pol

Errors building 1.4.0 (?)

[Mike Meyer]

I'm trying to build the CVS version for the first time. I used the
page that google had cached from www.devel.lyx.org on getting LyX
sources from CVS to get started.

That page talks about building "the development module" but doesn't talk
about what that is. The difference is that you run the autogen.sh
script before doing the configure/make/make install dance. I did that.

The build platform is FreeBSD 5.4-STABLE. That uses gcc 3.4.2. The
configure args I used were "--with-frontend=qt --with-qt-dir=/usr/X11R6
--with-aspell -with-extra-lib=/usr/opt/lib
-with-extra-inc=/usr/opt/include".

The FreeBSD ports tree builds version 1.3.5 with no patches. I used
the option it had to build with qt and aspell, and it adds the
-with-extra-* arguments to configure when it builds 1.3.5. Since I'm
using the same qt installation, I figured copying them couldn't hurt.

The compile exits with a slew of missig references to Qt objects. The
command that generated them and the first of them looks like this:

g++ -fno-exceptions -g -O -o lyx-qt main.o Bidi.o BufferView.o 
BufferView_pimpl.o Bullet.o Bran
chList.o Chktex.o CutAndPaste.o DepTable.o FloatList.o Floating.o 
FontIterator.o FuncStatus.o I
nsetList.o LColor.o LaTeX.o LaTeXFeatures.o LyXAction.o MenuBackend.o 
ParagraphParameters.o Pri
nterParams.o Spacing.o Thesaurus.o ToolbarBackend.o author.o boost.o box.o 
buffer.o buffer_func
s.o bufferlist.o bufferparams.o bufferview_funcs.o changes.o chset.o 
converter.o counters.o coo
rdcache.o cursor.o cursor_slice.o debug.o dimension.o dociterator.o encoding.o 
errorlist.o expo
rter.o gettext.o factory.o format.o funcrequest.o graph.o importer.o intl.o 
insetiterator.o kbm
ap.o kbsequence.o language.o lastfiles.o lengthcommon.o lyx_cb.o lyx_main.o 
lyx_sty.o lyxfont.o
 lyxfind.o lyxfunc.o lyxgluelength.o lyxlayout.o lyxlength.o lyxlex.o 
lyxlex_pimpl.o lyxrc.o ly
xrow.o lyxrow_funcs.o lyxserver.o lyxsocket.o lyxtextclass.o lyxtextclasslist.o 
lyxvc.o message
s.o metricsinfo.o mover.o output.o outputparams.o output_docbook.o 
output_latex.o output_linuxd
oc.o output_plaintext.o paragraph.o paragraph_funcs.o paragraph_pimpl.o 
pariterator.o ispell.o 
SpellBase.o rowpainter.o sgml.o tabular.o tex-accent.o tex-strings.o texrow.o 
text.o text2.o te
xt3.o toc.o trans.o trans_mgr.o undo.o vc-backend.o version.o vspace.o  
mathed/.libs/libmathed.
a insets/.libs/libinsets.a frontends/.libs/libfrontends.a 
frontends/qt2/.libs/libqt2.a -L/usr/X
11R6//lib frontends/controllers/.libs/libcontrollers.a 
graphics/.libs/libgraphics.a support/.li
bs/libsupport.a ../boost/libs/regex/src/.libs/libboost_regex.a 
../boost/libs/signals/src/.libs/
libboost_signals.a ../boost/libs/filesystem/src/.libs/libboost_filesystem.a 
../intl/libintl.a -
lSM -lICE -lm -L/usr/X11R6/lib -lX11 -lz
frontends/qt2/.libs/libqt2.a(Alert_pimpl.o)(.text+0x150): In function 
`prompt_pimpl(std::string
 const&, std::string const&, int, int, std::string const&, std::string const&, 
std::string cons
t&)':
/usr/X11R6//include/qapplication.h:451: undefined reference to 
`QApplication::focus_widget'

Looking over this, it's a link command - except that I don't see a qt
library being linked in. Grovelling through config.status finds: 
"s,@QT_LIB@,,;t t",
which sure looks like it didn't set the QT library properlty.

I could probably figure this out myself - except it's 3am, and I've
been fighting recalcitrant hardware all night. I'm going to sleep,
then to work, and with luck and help from you all, I'll find a
solution in my mailbox tomorrow evening.

Thanks,
  http://www.mired.org/consulting.html
Independent Network/Unix/Perforce consultant, email for more information.

Re: Errors building 1.4.0 (?)

[Georg Baum]

Mike Meyer wrote:

> Looking over this, it's a link command - except that I don't see a qt
> library being linked in. Grovelling through config.status finds:
> "s,@QT_LIB@,,;t t", which sure looks like it didn't set the QT library
> properlty.

It looks like configure does not find the qt libary (but it seems to find
the headers). This looks like a configure bug: It should stop with an error
message if the library cannot be found.
Could you please post the lines of config.log where it searches for the qt
library? (around the message "checking for Qt library name").
Have you tried it without the --qt-dir switch?


Georg

Re: byte compilation of lyx2lyx files

[Lars Gullik Bjønnes]

Angus Leeming <[EMAIL PROTECTED]> writes:

| Lars Gullik Bjønnes wrote:
| 
| > Angus Leeming <[EMAIL PROTECTED]> writes:
| > 
| > | Luis Rivera reports that "make install" is failing for him because he
| > | doesn't have python installed. Could someone sort this out please.
| > 
| > Hmm... we kindo require python for building.
| 
| We do? Where? Since when?

It is in the "kindo" cathegory.

-- 
Lgb

Re: byte compilation of lyx2lyx files

[Lars Gullik Bjønnes]

[EMAIL PROTECTED] (Lars Gullik Bjønnes) writes:

| Angus Leeming <[EMAIL PROTECTED]> writes:
| 
| | Lars Gullik Bjønnes wrote:
| | 
| | > Angus Leeming <[EMAIL PROTECTED]> writes:
| | > 
| | > | Luis Rivera reports that "make install" is failing for him because he
| | > | doesn't have python installed. Could someone sort this out please.
| | > 
| | > Hmm... we kindo require python for building.
| | 
| | We do? Where? Since when?
| 
| It is in the "kindo" cathegory.

We should be able to handle the no-python case with some use of 

AM_CONDITIONAL([HAVE_PYTHON], [test "$PYTHON" != :])

just let lyx2lyx's Makefile.am not do anything unless HAVE_PYTHON is
defined.

-- 
Lgb

Re: large executable

[Jean-Marc Lasgouttes]

> "Pol" == Pol  <[EMAIL PROTECTED]> writes:

Pol> Jean-Marc Lasgouttes wrote:
>> that i am compiling lyx 1.4
Paolo> The compiled lyx 1.4 executable from cvs, last april, was 4.2
Paolo> megabytes; but in that case compilation had been accoplished on
Paolo> ordinary intel desktop workstation. That is the reason for
Paolo> suspecting something wrong,
>> On what system are you compiling?
>> 

Pol> it is debian unstable with gcc 4.0.1-3 (corresponding to gcc
Pol> version 4.0.2 20050725 (prerelease), as resulting from ' gcc -v
Pol> ') and libqt3-mt-dev 3.3.4-3

what does "size ./lyx" day? Here (gcc 3.4.1) I get:

fantomas: size src/lyx
   textdata bss dec hex filename
6083056   17680   35060 6135796  5d9ff4 src/lyx

JMarc

Re: LyX 140 configure problem with awk

[Jean-Marc Lasgouttes]

> "Bennett" == Bennett Helm <[EMAIL PROTECTED]> writes:

Bennett> That seems to work for me.

The important question is whether the information in lyx.pot is
reasonable. You should see entries like:

msgid "\\arabic{enumi}."
msgstr ""

(what is important here is the double backslash).

JMarc

Re: byte compilation of lyx2lyx files

[Jean-Marc Lasgouttes]

> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:

Angus> Luis Rivera reports that "make install" is failing for him
Angus> because he doesn't have python installed. Could someone sort
Angus> this out please.

I need more details, could you point me to the relevant thread?
Basically, this is supposed to work provided automake is recent
enough.

So, if I do not see the error message, there is nothing I can do.

JMarc

Re: [PATCH] bug 1941: toolbars disabled after dialog action

[Jean-Marc Lasgouttes]

> "Georg" == Georg Baum <[EMAIL PROTECTED]> writes:

Georg> I tried to understand where in the code that happens, but
Georg> failed. Could you please tell me the place where the main
Georg> window without focus prevents the icons from being enabled?

In LyXFunc::getStatus, you will find:

/* In LyX/Mac, when a dialog is open, the menus of the
   application can still be accessed without giving focus to
   the main window. In this case, we want to disable the menu
   entries that are buffer-related.
*/
Buffer * buf;
if (cmd.origin == FuncRequest::UI && !owner->hasFocus())
buf = 0;
else
buf = owner->buffer();

We do as if there was no buffer. This means that all not
buffer-related functions are still active. 

>> My fix is to avoid updating toolbar/menubar for INTERNAL
>> dispatches.

Georg> I don't think that this is the right fix. The problem is
Georg> qt-only, so it should be fixed in the qt frontend. 

It is not really qt-only. The same code would trigger for a Cocoa
port, or a Gtk/osx. 

Georg> I don't know if that currently happens, but I could imagine the
Georg> case where applying a dialog makes an lfun invalid. In this
Georg> case we certainly want that icon to be disabled, but that does
Georg> not work anymore with your fix.

Indeed.

Georg> The right fix is IMHO to fix the focus problem. Why is it not
Georg> possible to give the focus to the main window after a dialog is
Georg> closed?

This is what I tried first, actually: updating the tool/menu bars after
the window has been hidden in Dialog::hide. It did not work, and I do
not understand why.

JMarc

Re: byte compilation of lyx2lyx files

[Jean-Marc Lasgouttes]

> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:

Lars> [EMAIL PROTECTED] (Lars Gullik Bjønnes) writes: | Angus Leeming
Lars> <[EMAIL PROTECTED]> writes:
Lars> | 
Lars> | | Lars Gullik Bjønnes wrote:
Lars> | | 
Lars> | | > Angus Leeming <[EMAIL PROTECTED]> writes:
Lars> | | > 
Lars> | | > | Luis Rivera reports that "make install" is failing for
Lars> him because he | | > | doesn't have python installed. Could
Lars> someone sort this out please.
Lars> | | > 
Lars> | | > Hmm... we kindo require python for building.
Lars> | | 
Lars> | | We do? Where? Since when?
Lars> | 
Lars> | It is in the "kindo" cathegory.

Lars> We should be able to handle the no-python case with some use of

Actually, I do not see why it should not work when PYTHON=:. That's
why I'd like to see the error first.

JMarc

Re: byte compilation of lyx2lyx files

[Jose' Matos]

On Tuesday 13 September 2005 10:44, Jean-Marc Lasgouttes wrote:
> Actually, I do not see why it should not work when PYTHON=:. That's
> why I'd like to see the error first.

  I am puzzled as well. We made the support for python conditional, it should 
not fail if python is not present.

> JMarc

-- 
José Abílio

Re: byte compilation of lyx2lyx files

[Angus Leeming]

Jean-Marc Lasgouttes wrote:
> Angus> Luis Rivera reports that "make install" is failing for him
> Angus> because he doesn't have python installed. Could someone sort
> Angus> this out please.

> I need more details, could you point me to the relevant thread?
> Basically, this is supposed to work provided automake is recent
> enough.

> So, if I do not see the error message, there is nothing I can do.
> JMarc

Luis didn't provide much information. Just the second paragraph here, IIRC:
http://article.gmane.org/gmane.editors.lyx.devel:47931

I'll drop him a line.

-- 
Angus

Re: byte compilation of lyx2lyx files

[Jean-Marc Lasgouttes]

> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:

Angus> Luis didn't provide much information. Just the second paragraph
Angus> here, IIRC:
Angus> http://article.gmane.org/gmane.editors.lyx.devel:47931

Yes, I saw that.

Angus> I'll drop him a line.

Thanks.

JMarc

Safer Win32 get_temp_path, os::internal_path

[Angus Leeming]

Jean-Marc, all,

following our earlier discussion, I attach a little test code to show what 
I think we need to produce robust code on Windows. The attached code should 
compile on all flavours of Windows >= Win95 and should behave gracefully on 
older versions for which GetLongPathName is just a stub.


I propose to post this sample off to Luis Rivera and get him to try it out 
on Win95/Win98 but I'd of course welcome any sanity checks and comments 
from you first.


Angus (the one of questionable moral character :-P)

$ g++ -I$HOME/lyx/13x/boost -o trial trial.cpp
$ ./trial
Temp dir is C:/Documents and Settings/Angus/Local Settings/Temp/
#include 

#include 
#include 
#include 

/*
 * MinGW's version of winver.h contains this comment:
 *
 * If you need Win32 API features newer the Win95 and WinNT then you must
 * define WINVER before including windows.h or any other method of including
 * the windef.h header.
 *
 * GetLongPathNameA requires WINVER == 0x0500.
 *
 * It doesn't matter if the Windows version is older than this because the
 * function will compile but will fail at run time. See
 * 
http://msdn.microsoft.com/library/en-us/mslu/winprog/microsoft_layer_for_unicode_apis_with_limited_support.asp
 */
# if defined(__MINGW32__)
#  define WINVER 0x0500
# endif

#include 

using std::string;
using std::vector;


namespace {

string const subst(string const & a,
   string const & oldstr, string const & newstr)
{
BOOST_ASSERT(!oldstr.empty());

string lstr(a);
string::size_type i = 0;
string::size_type const olen = oldstr.length();
while ((i = lstr.find(oldstr, i)) != string::npos) {
lstr.replace(i, olen, newstr);
i += newstr.length(); // We need to be sure that we dont
// use the same i over and over again.
}
return lstr;
}


namespace os {

string const internal_path(string const )
{
string const short_path = subst(p, "\\", "/");

vector long_path(PATH_MAX);
DWORD result = GetLongPathName(short_path.c_str(),
   _path[0], long_path.size());

if (result > long_path.size()) {
long_path.resize(result);
result = GetLongPathName(short_path.c_str(),
 _path[0], long_path.size());
BOOST_ASSERT(result <= long_path.size());
}

return (result == 0) ? short_path : _path[0];
}

} // namespace os


string const get_temp_dir()
{
// Typical example: C:/TEMP/.
char path[PATH_MAX];
DWORD const result = GetTempPath(PATH_MAX, path);
BOOST_ASSERT(result > 0 && result <= PATH_MAX);
return os::internal_path(path);
}

} // namespace anon


int main()
{
string const temp_dir = get_temp_dir();
std::cout << "Temp dir is " << temp_dir << std::endl;
return 0;
}

Re: large executable

[Pol]

Jean-Marc Lasgouttes wrote:


what does "size ./lyx" day? Here (gcc 3.4.1) I get:

fantomas: size src/lyx
   textdata bss dec hex filename
6083056   17680   35060 6135796  5d9ff4 src/lyx



% size /usr/local/bin/lyx


  textdata bss  dec hex   filename
  11134899 394540   35332 11564771b076e3 /usr/local/bin/lyx


--
Paolo

Re: gcc4 prevents some languages from working - still stuck :-(

[Helge Hafting]

Martin Vermeer wrote:


On Wed, Aug 17, 2005 at 04:31:07PM +0100, Angus Leeming wrote:
 


Martin Vermeer wrote:

   


On Tue, 2005-08-16 at 12:59, Helge Hafting wrote:
 


I noticed something very strange today.
I could not get lyx-1.4cvs to display norwegian text at all.
no_NO, no_NO.iso8859-1, no_NO.utf-8 all gave me english.
This on a machine that uses gcc 4.01, another machine
using gcc-3.3.6 works.

Other languages, like nn_NO works.

So, it is not a case of complete locale breakdown,
and not a case of a completely broken no.po either.

Any ideas?  I was going to check if recent fixes had added
strings I could translate, but this is driving me nuts.
The no_NO locales works for other programs, such as "ls".  So
I belive the distribution is correctly set up.

Helge Hafting
   


Does nb_NO work?
 


Debugging this would be much, much easier if we changed gettext_init
(lyx 1.3.x) in src/gettext.C to something like:

void gettext_init(string const & localedir)
{
   char const * const btd = bindtextdomain(PACKAGE, localedir.c_str());
   char const * const td = textdomain(PACKAGE);

   if (!btd || !td) {
   lyxerr << "Failed to initialize localedir " << localedir
  << " successfully\nbindtextdomain " << btd
  << "\ntextdomain " << td << std::endl;
   }
}

An equivalent change to lyx 1.4 is also needed, IMO.
   



This *is* a 1.4 issue. In 1.4, the relevant code is in src/messages.C.
It's not pretty. This stuff was touched in Paris.

117 char const * works = setlocale(LC_MESSAGES, lang_.c_str());

add a printout here:

if (!works)
lyxerr << "Locale " << lang_ " << could not be set" << 
std::endl;

I'll commit this as it is clearly needed.
 


I am still stuck on this.  I can't get no_NO to work for
the no.po in cvs.  There are no error messages (or any message different
from when trying other working languages, like nn_NO).  I simply
get english when LANG=no_NO (or no_NO.iso8859-1).

This seems to imply that something is wrong in the no.po
file, but I have no idea what that could be.  "make no.gmo"
gives no error messages, the file is created, but Norwegian
language just don't work. 


Helge Hafting

Re: gcc4 prevents some languages from working - still stuck :-(

[Martin Vermeer]

On Tue, 2005-09-13 at 13:26 +0200, Helge Hafting wrote:

...

> I am still stuck on this.  I can't get no_NO to work for
> the no.po in cvs.  There are no error messages (or any message different
> from when trying other working languages, like nn_NO).  I simply
> get english when LANG=no_NO (or no_NO.iso8859-1).
> 
> This seems to imply that something is wrong in the no.po
> file, but I have no idea what that could be.  "make no.gmo"
> gives no error messages, the file is created, but Norwegian
> language just don't work. 
> 
> Helge Hafting

This is utterly weird. Did you create a bug in bugzilla for this?

- Martin



signature.asc
Description: This is a digitally signed message part

Re: Safer Win32 get_temp_path, os::internal_path

[Jean-Marc Lasgouttes]

> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:

Angus> Jean-Marc, all, following our earlier discussion, I attach a
Angus> little test code to show what I think we need to produce robust
Angus> code on Windows. The attached code should compile on all
Angus> flavours of Windows >= Win95 and should behave gracefully on
Angus> older versions for which GetLongPathName is just a stub.

This looks good, if it works.

Angus> Angus (the one of questionable moral character :-P)

To tell, the truth, I suspected it already, and was rather relieved
not to have you in my house last July. I do not know what my
neighbours would have thought.

JMarc

Re: Errors building 1.4.0 (?)

[Mike Meyer]

In <[EMAIL PROTECTED]>, Georg Baum <[EMAIL PROTECTED]> typed:
> Mike Meyer wrote:
> > Looking over this, it's a link command - except that I don't see a qt
> > library being linked in. Grovelling through config.status finds:
> > "s,@QT_LIB@,,;t t", which sure looks like it didn't set the QT library
> > properlty.
> It looks like configure does not find the qt libary (but it seems to find
> the headers). This looks like a configure bug: It should stop with an error
> message if the library cannot be found.
> Could you please post the lines of config.log where it searches for the qt
> library? (around the message "checking for Qt library name").

Sure:

configure:24202: checking for Qt library name
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt-mt >&5
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_cleanup_pop'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_attr_destroy'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_attr_init'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_exit'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_cancel'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_testcancel'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_cleanup_push'
/usr/X11R6//lib/libqt-mt.so: undefined reference to 
`pthread_attr_getschedpolicy'
/usr/X11R6//lib/libqt-mt.so: undefined reference to 
`pthread_attr_setinheritsched'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_attr_setstacksize'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_attr_setschedparam'
/usr/X11R6//lib/libqt-mt.so: undefined reference to 
`pthread_attr_setdetachstate'
/usr/X11R6//lib/libqt-mt.so: undefined reference to `pthread_cond_timedwait'
configure:24253: $? = 1
configure: failed program was:
[listing of failing program elided]
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt-mt3 >&5
/usr/bin/ld: cannot find -lqt-mt3
configure:24253: $? = 1
configure: failed program was:
[listing of failed program elded]
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt3 >&5
/usr/bin/ld: cannot find -lqt3
configure:24253: $? = 1
configure: failed program was:
[listing of failed program elded]
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt2 >&5
/usr/bin/ld: cannot find -lqt2
configure:24253: $? = 1
configure: failed program was:
[listing of failed program elded]
configure:24247: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11  -lqt >&5
/usr/bin/ld: cannot find -lqt
configure:24253: $? = 1
configure: failed program was:
configure:24289: result: failed
configure:24306: checking whether the Qt library is multi-threaded
configure:24345: g++ -o conftest -g -O -I/usr/X11R6//include -L/usr/X11R6//lib  
-I/usr/opt/include  -Wextra -Wall-I/usr/X11R6/include  conftest.cc  -lSM 
-lICE -lc -lm   -L/usr/X11R6/lib -lX11   >&5
/var/tmp//cc4P8UEz.o(.text+0x118): In function `main':
/home/mwm/external-src/lyx-devel/conftest.cc:51: undefined reference to 
`QApplication::QApplica
tion(_XDisplay*, unsigned long, unsigned long)'
/var/tmp//cc4P8UEz.o(.text+0x123):/home/mwm/external-src/lyx-devel/conftest.cc:52:
 undefined re
ference to `QApplication::unlock(bool)'
/var/tmp//cc4P8UEz.o(.text+0x136):/home/mwm/external-src/lyx-devel/conftest.cc:55:
 undefined re
ference to `QApplication::~QApplication()'
/var/tmp//cc4P8UEz.o(.text+0x14c):/home/mwm/external-src/lyx-devel/conftest.cc:55:
 undefined re
ference to `QApplication::~QApplication()'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0xc): undefined reference to 
`QGList::clear()'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0x10):/usr/include/c++/3.4/bits/basic_string.h
:257: undefined reference to `QGList::~QGList()'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0x14):/usr/include/c++/3.4/bits/basic_string.h
:532: undefined reference to `QGList::~QGList()'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0x18):/usr/include/c++/3.4/bits/stl_algobase.h
:151: undefined reference to `QPtrCollection::newItem(void*)'
/var/tmp//cc4P8UEz.o(.gnu.linkonce.r._ZTV6QGList+0x20):/usr/include/c++/3.4/bits/stl_algobase.h
:156: undefined reference to `QGList::compareItems(void*, void*)'

Re: Safer Win32 get_temp_path, os::internal_path

[Angus Leeming]

Jean-Marc Lasgouttes wrote:

Angus> Jean-Marc, all, following our earlier discussion, I attach a
Angus> little test code to show what I think we need to produce robust
Angus> code on Windows. The attached code should compile on all
Angus> flavours of Windows >= Win95 and should behave gracefully on
Angus> older versions for which GetLongPathName is just a stub.



This looks good, if it works.


Well, it works for me. Time will tell if it works for Luis.


Angus> Angus (the one of questionable moral character :-P)



To tell, the truth, I suspected it already, and was rather relieved
not to have you in my house last July. I do not know what my
neighbours would have thought.


Funny. I had you down as bohémien rather than a member of the bourgeoisie. 
Anyway, wasn't it your parents' neighbours rather than your own?


Angus (scandalous individual)

Re: Errors building 1.4.0 (?)

[Georg Baum]

Mike Meyer wrote:

> I can tell you what went wrong with the first command: it needs the
> FreeBSDism "-pthread" to link with the threaded version of libc.

That means we'll need to fix the configure script. Thanks for debugging! Can
you test the attached (untested) patch? Don't forget to run autogen.sh.

I also think that we should set FATAL=1 in qt.m4. That would give a sensible
error message. It is nonsense to compile for qt when the resulting
executable cannot be linked.

> Again, this worked without patching in 1.3.5.

Strange.


GeorgIndex: config/qt.m4
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/config/qt.m4,v
retrieving revision 1.15
diff -u -p -r1.15 qt.m4
--- config/qt.m4	17 Jul 2005 23:40:08 -	1.15
+++ config/qt.m4	13 Sep 2005 12:31:19 -
@@ -114,6 +114,12 @@ AC_DEFUN([QT_CHECK_COMPILE],
 			QT_TRY_LINK($libname)
 			if test -n "$qt_cv_libname"; then
 break;
+			else
+dnl on FreeBSD we need -pthread for threaded libs
+QT_TRY_LINK("$libname -pthread")
+if test -n "$qt_cv_libname"; then
+	break;
+fi
 			fi
 		done
 

Replacing SHGetFolderPath

[Angus Leeming]

As older flavours of Windows don't support SHGetFolderPath out of the box, 
Jean-Marc has suggested that we use the more cumbersome but also more 
portable SHGetSpecialFolderLocation.


I propose the attached code as a replacement for the existing 
win32_folder_path in src/support/package.C.


$ g++ -I$HOME/lyx/13x/boost -W -Wall -o get_folder get_folder.cpp
$ ./get_folder.exe
Document dir: C:\Documents and Settings\Angus\My Documents
App data dir: C:\Documents and Settings\Angus\Application Data

I'll post this off to Luis Rivera to check that it compiles/runs on his 
older machines.


One thing that is side-stepped by the code at the moment is what should we 
do if string() is returned for either document dir or app data dir? Should 
we assert?


Angus

#include 

#include 
#include 

# include 
# include   // SHGetFolderPath

using std::string;


// Given a folder ID, returns the folder name (in unix-style format).
// Eg CSIDL_PERSONAL -> "C:/Documents and Settings/USERNAME/My Documents"
string const win32_folder_path(int folder_id)
{
// Microsoft recommend the use of SHGetFolderPath to do
// what we're attempting here, but SHGetFolderPath won't work
// out of the box on older (Win98) flavours of Windows.
// Either we distribute shfolder.dll (fully-redistributable part
// of IE 4) and ensure that we link against it before we link
// shell32.dll, or we use SHGetSpecialFolderLocation instead.
LPITEMIDLIST id_list;
if (SHGetSpecialFolderLocation(0, folder_id, _list) != S_OK)
return string();

char folder_path[PATH_MAX];
DWORD const result = SHGetPathFromIDList(id_list, folder_path);

LPMALLOC allocator;
if (SHGetMalloc() == S_OK) {
allocator->Free(id_list);
allocator->Release();
}

return (result == TRUE) ? folder_path : string();
}


int main()
{
std::cout << "Document dir: " << win32_folder_path(CSIDL_PERSONAL)
  << "\nApp data dir: " << win32_folder_path(CSIDL_APPDATA)
  << std::endl;

return 0;
}

Re: Safer Win32 get_temp_path, os::internal_path

[Jean-Marc Lasgouttes]

> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:

>> To tell, the truth, I suspected it already, and was rather relieved
>> not to have you in my house last July. I do not know what my
>> neighbours would have thought.

Angus> Funny. I had you down as bohémien rather than a member of the
Angus> bourgeoisie. 

Well, I have some funny bourgeois in my mother's family. And I went
mainly to private catholic schools.

Angus> Anyway, wasn't it your parents' neighbours rather than your
Angus> own?

God forbid! I would not even have thought about inviting strangers in
this case! We are now in our own appartment, so I only have to care
about my reputation, which is nevertheless a lot, as you will
undoubtlessly agree.

JMarc

Re: LyX 140 configure problem with awk

[Bennett Helm]

On Sep 13, 2005, at 5:16 AM, Jean-Marc Lasgouttes wrote:


The important question is whether the information in lyx.pot is
reasonable. You should see entries like:

msgid "\\arabic{enumi}."
msgstr ""

(what is important here is the double backslash).


I don't get \\arabic{enumi} in particular. I do get double  
backslashes in the following cases, all of which look reasonable:


\\begin_header
\\begin_document
\\,
\\:
\\;
\\quad
\\qquad
... etc. in QMathDialog.C
\\input
\\include
\\verbatiminput
\\usepackage
\\documentclass
\\selectlanguage

There aren't cases of single backslashes where it looks like doubles  
should be used instead.


Is that what you need to know?

Bennett

Re: Replacing SHGetFolderPath

[Jean-Marc Lasgouttes]

> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:

Angus> As older flavours of Windows don't support SHGetFolderPath out
Angus> of the box, Jean-Marc has suggested that we use the more
Angus> cumbersome but also more portable SHGetSpecialFolderLocation.

Angus> I propose the attached code as a replacement for the existing
Angus> win32_folder_path in src/support/package.C.

Looks good.

Angus> One thing that is side-stepped by the code at the moment is
Angus> what should we do if string() is returned for either document
Angus> dir or app data dir? Should we assert?

If the app data dir is empty, we will eventually exit somewhere, I
guess. As far as the document data dir is concerned, I am not sure
this really matters.

JMarc

Re: Safer Win32 get_temp_path, os::internal_path

[Angus Leeming]

Jean-Marc Lasgouttes wrote:
> Angus> Anyway, wasn't it your parents' neighbours rather than your
> Angus> own?

> God forbid! I would not even have thought about inviting strangers in
> this case! We are now in our own appartment, so I only have to care
> about my reputation, which is nevertheless a lot, as you will
> undoubtlessly agree.

Oh, I don't know. I'm quite careless about the reputation of others,
apparently, which is what got me into trouble on the lyx-users' list in
the first place :-P

-- 
Angus

Re: LyX 140 configure problem with awk

[Jean-Marc Lasgouttes]

> "Bennett" == Bennett Helm <[EMAIL PROTECTED]> writes:

Bennett> On Sep 13, 2005, at 5:16 AM, Jean-Marc Lasgouttes wrote:
>> The important question is whether the information in lyx.pot is
>> reasonable. You should see entries like:
>> 
>> msgid "\\arabic{enumi}." msgstr ""
>> 
>> (what is important here is the double backslash).

Bennett> I don't get \\arabic{enumi} in particular. I do get double
Bennett> backslashes in the following cases, all of which look
Bennett> reasonable:

Bennett> There aren't cases of single backslashes where it looks like
Bennett> doubles should be used instead.

Bennett> Is that what you need to know?

What I am interested in are entries related to LabelString in layout
files, so my enumi example was not very well chosen.

For example, numarticle.inc says

Style Section
LabelType Counter
LabelCounter  section
LabelString   "\arabic{section}"
LabelStringAppendix   "\Alph{section}"
TocLevel  1
End

so you should find the strings "\\arabic{section}" and
"\\Alph{section}" in the .pot file.

JMarc

Re: Replacing SHGetFolderPath

[Angus Leeming]

Jean-Marc Lasgouttes wrote:
> Angus> As older flavours of Windows don't support SHGetFolderPath out
> Angus> of the box, Jean-Marc has suggested that we use the more
> Angus> cumbersome but also more portable SHGetSpecialFolderLocation.
> 
> Angus> I propose the attached code as a replacement for the existing
> Angus> win32_folder_path in src/support/package.C.
> 
> Looks good.

One thing I note from the MSDN website is:

=
http://msdn.microsoft.com/library/en-us/shellcc/platform/shell/reference/functions/shgetmalloc.asp
SHGetMalloc Function

Not currently supported.

Remarks
This function should no longer be used. Use the CoTaskMemFree and
CoTaskMemAlloc functions in its place.
=

I've just ignored this advice, but this blog article
http://blogs.msdn.com/oldnewthing/archive/2004/07/05/173226.aspx

suggests that all should be fine anyway (since I think that we link against
ole32.dll)

Ain't life on Windoze complicated!
-- 
Angus

Re: LyX 140 configure problem with awk

[Bennett Helm]

On Sep 13, 2005, at 8:56 AM, Jean-Marc Lasgouttes wrote:


What I am interested in are entries related to LabelString in layout
files, so my enumi example was not very well chosen.


Actually, it was my fault: I was stupidly looking at lyx.pot from the  
wrong build.



For example, numarticle.inc says

Style Section
LabelType Counter
LabelCounter  section
LabelString   "\arabic{section}"
LabelStringAppendix   "\Alph{section}"
TocLevel  1
End

so you should find the strings "\\arabic{section}" and
"\\Alph{section}" in the .pot file.


Yes -- they're all there (at least every one I checked!).

Bennett

Re: Replacing SHGetFolderPath

[Jean-Marc Lasgouttes]

> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:

Angus> One thing I note from the MSDN website is:

Angus> =
Angus> 
http://msdn.microsoft.com/library/en-us/shellcc/platform/shell/reference/functions/shgetmalloc.asp
Angus> SHGetMalloc Function

Angus> Not currently supported.

Angus> Remarks This function should no longer be used. Use the
Angus> CoTaskMemFree and CoTaskMemAlloc functions in its place.
Angus> =

Angus> I've just ignored this advice, but this blog article
Angus> http://blogs.msdn.com/oldnewthing/archive/2004/07/05/173226.aspx

Angus> suggests that all should be fine anyway (since I think that we
Angus> link against ole32.dll)

Indeed...

JMarc

[PATCH] bug 2010: Entering/leaving text inset requires two cursor left/right keystrokes

[Jean-Marc Lasgouttes]

The following patch fixes bug 2010:
http://bugzilla.lyx.org/show_bug.cgi?id=2010

This was an unintended consequence of the work Juergen V. did to allow
placing the cursor in front of a displayed inset.

Please test.

JMarc

Index: src/ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/ChangeLog,v
retrieving revision 1.2277
diff -u -p -r1.2277 ChangeLog
--- src/ChangeLog	10 Sep 2005 13:29:43 -	1.2277
+++ src/ChangeLog	13 Sep 2005 13:44:39 -
@@ -1,3 +1,8 @@
+2005-09-13  Jean-Marc Lasgouttes  <[EMAIL PROTECTED]>
+
+	* text2.C (cursorEnd, cursorRight): do not set boundary to true at
+	end of paragraph. (bug 2010)
+
 2005-09-10  Georg Baum  <[EMAIL PROTECTED]>
 
 	* text.C (leftMargin): check for inInset == 0
Index: src/text2.C
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/text2.C,v
retrieving revision 1.627
diff -u -p -r1.627 text2.C
--- src/text2.C	5 Sep 2005 06:50:42 -	1.627
+++ src/text2.C	13 Sep 2005 13:44:40 -
@@ -498,12 +498,13 @@ void LyXText::cursorEnd(LCursor & cur)
 		// empty text, end-1 is no valid position
 		return;
 	bool boundary = false;
-	if (!cur.paragraph().isLineSeparator(end-1) &&
-	!cur.paragraph().isNewline(end-1))
-	{
-		boundary = true;
-	} else if (end != cur.lastpos())
-		--end;
+	if (end != cur.lastpos()) {
+		if (!cur.paragraph().isLineSeparator(end-1) 
+			&& !cur.paragraph().isNewline(end-1))
+			boundary = true;
+		else
+			--end;
+	}
 	setCursor(cur, cur.pit(), end, true, boundary);
 }
 
@@ -1021,6 +1022,7 @@ bool LyXText::cursorRight(LCursor & cur)
 		bool updateNeeded = false;
 		if (!checkAndActivateInset(cur, true)) {
 			if (cur.textRow().endpos() == (cur.pos() + 1) &&
+			!cur.lastpos() == (cur.pos() + 1) &&
 			!cur.paragraph().isLineSeparator(cur.pos()) &&
 			!cur.paragraph().isNewline(cur.pos()))
 			{

Re: LyX 140 configure problem with awk

[Jean-Marc Lasgouttes]

> "Bennett" == Bennett Helm <[EMAIL PROTECTED]> writes:

>> so you should find the strings "\\arabic{section}" and
>> "\\Alph{section}" in the .pot file.

Bennett> Yes -- they're all there (at least every one I checked!).

Very good. I'll apply the patch.

JMarc

Re: [PATCH] bug 1941: toolbars disabled after dialog action

[Georg Baum]

Jean-Marc Lasgouttes wrote:

>> "Georg" == Georg Baum
>> <[EMAIL PROTECTED]>
>> writes:
> 
> Georg> I tried to understand where in the code that happens, but
> Georg> failed. Could you please tell me the place where the main
> Georg> window without focus prevents the icons from being enabled?
> 
> In LyXFunc::getStatus, you will find:

Thanks, now I see it (I was searching in frontends/qt2)

> It is not really qt-only. The same code would trigger for a Cocoa
> port, or a Gtk/osx.

I see. Because XFormsView::hasFocus() and GView::hasFocus() are too stupid
we don't see the bug there.

> Georg> The right fix is IMHO to fix the focus problem. Why is it not
> Georg> possible to give the focus to the main window after a dialog is
> Georg> closed?
> 
> This is what I tried first, actually: updating the tool/menu bars after
> the window has been hidden in Dialog::hide. It did not work, and I do
> not understand why.

The problem is that the main window still has no focus when Dialog::hide is
called. I tried to activate it with QMainWindow::setFocus() and
QMainWindow::show(), but neither worked. I'll investigate further.


Georg

Re: [PATCH] bug 1941: toolbars disabled after dialog action

[Jean-Marc Lasgouttes]

> "Georg" == Georg Baum <[EMAIL PROTECTED]> writes:

Georg> The problem is that the main window still has no focus when
Georg> Dialog::hide is called. I tried to activate it with
Georg> QMainWindow::setFocus() and QMainWindow::show(), but neither
Georg> worked. I'll investigate further.

Thanks.

JMarc

Re: Errors building 1.4.0 (?)

[Georg Baum]

Am Dienstag, 13. September 2005 14:30 schrieb John Levon:
> On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:
> > It looks like configure does not find the qt libary (but it seems to find
> > the headers). This looks like a configure bug: It should stop with an
> > error message if the library cannot be found.
>
> IIRC Lars changed this.

I now remember that, too. I also remember that I added error messages for moc 
and uic, so here comes the logical extension.

Mike, could you please test whether configure aborts with this patch (and 
without the other one) if it cannot find the qt library?


Georg
Index: ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/ChangeLog,v
retrieving revision 1.1023
diff -u -p -r1.1023 ChangeLog
--- ChangeLog	18 Jul 2005 15:49:56 -	1.1023
+++ ChangeLog	13 Sep 2005 14:23:16 -
@@ -1,3 +1,8 @@
+2005-09-13  Georg Baum  <[EMAIL PROTECTED]>
+
+	* configure.ac: error out if qt frontend is choosen but no qt library
+	was found
+
 2005-07-18  Lars Gullik Bjønnes  <[EMAIL PROTECTED]>
 
 	* configure.ac: version back to 1.4.0cvs 
Index: configure.ac
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/configure.ac,v
retrieving revision 1.58
diff -u -p -r1.58 configure.ac
--- configure.ac	18 Jul 2005 15:49:57 -	1.58
+++ configure.ac	13 Sep 2005 14:23:16 -
@@ -216,6 +219,9 @@ dnl qt build will fail without moc or ui
 	  fi
 	  if test -z "$UIC"; then
 	LYX_ERROR([uic binary not found !])
+	  fi
+	  if test -z "$QT_LIB"; then
+	LYX_ERROR([qt library not found !])
 	  fi
   ;;
 *)

Re: Errors building 1.4.0 (?)

[Jean-Marc Lasgouttes]

> "Georg" == Georg Baum <[EMAIL PROTECTED]> writes:

Georg> Am Dienstag, 13. September 2005 14:30 schrieb John Levon:
>> On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote: > It
>> looks like configure does not find the qt libary (but it seems to
>> find > the headers). This looks like a configure bug: It should
>> stop with an > error message if the library cannot be found.
>> 
>> IIRC Lars changed this.

Georg> I now remember that, too. I also remember that I added error
Georg> messages for moc and uic, so here comes the logical extension.

Georg> Mike, could you please test whether configure aborts with this
Georg> patch (and without the other one) if it cannot find the qt
Georg> library?

The patch looks good. Concerning the bug itself, could it be related
to
http://marc.theaimsgroup.com/?l=lyx-devel=112490112909207=2

JMarc

Re: [PATCH] bug 2010: Entering/leaving text inset requires two cursor left/right keystrokes

[Angus Leeming]

Jean-Marc Lasgouttes wrote:

> The following patch fixes bug 2010:
> http://bugzilla.lyx.org/show_bug.cgi?id=2010
> 
> This was an unintended consequence of the work Juergen V. did to allow
> placing the cursor in front of a displayed inset.
> 
> Please test.
> 
> JMarc

why not
+   cur.lastpos() != (cur.pos() + 1) &&
?

@@ -1021,6 +1022,7 @@ bool LyXText::cursorRight(LCursor & cur)
bool updateNeeded = false;
if (!checkAndActivateInset(cur, true)) {
if (cur.textRow().endpos() == (cur.pos() + 1) &&
+   !cur.lastpos() == (cur.pos() + 1) &&
!cur.paragraph().isLineSeparator(cur.pos()) &&
!cur.paragraph().isNewline(cur.pos()))

-- 
Angus

Re: [PATCH] bug 2010: Entering/leaving text inset requires two cursor left/right keystrokes

[Jean-Marc Lasgouttes]

> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:

Angus> why not + cur.lastpos() != (cur.pos() + 1) && ?

Oops. I am not even sure that the test meant what I thought it meant.

What about the following?

JMarc

Index: src/ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/ChangeLog,v
retrieving revision 1.2277
diff -u -p -r1.2277 ChangeLog
--- src/ChangeLog	10 Sep 2005 13:29:43 -	1.2277
+++ src/ChangeLog	13 Sep 2005 14:56:01 -
@@ -1,3 +1,8 @@
+2005-09-13  Jean-Marc Lasgouttes  <[EMAIL PROTECTED]>
+
+	* text2.C (cursorEnd, cursorRight): do not set boundary to true at
+	end of paragraph. (bug 2010)
+
 2005-09-10  Georg Baum  <[EMAIL PROTECTED]>
 
 	* text.C (leftMargin): check for inInset == 0
Index: src/text2.C
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/text2.C,v
retrieving revision 1.627
diff -u -p -r1.627 text2.C
--- src/text2.C	5 Sep 2005 06:50:42 -	1.627
+++ src/text2.C	13 Sep 2005 14:56:01 -
@@ -498,12 +498,13 @@ void LyXText::cursorEnd(LCursor & cur)
 		// empty text, end-1 is no valid position
 		return;
 	bool boundary = false;
-	if (!cur.paragraph().isLineSeparator(end-1) &&
-	!cur.paragraph().isNewline(end-1))
-	{
-		boundary = true;
-	} else if (end != cur.lastpos())
-		--end;
+	if (end != cur.lastpos()) {
+		if (!cur.paragraph().isLineSeparator(end-1) 
+			&& !cur.paragraph().isNewline(end-1))
+			boundary = true;
+		else
+			--end;
+	}
 	setCursor(cur, cur.pit(), end, true, boundary);
 }
 
@@ -1020,7 +1021,8 @@ bool LyXText::cursorRight(LCursor & cur)
 	if (cur.pos() != cur.lastpos()) {
 		bool updateNeeded = false;
 		if (!checkAndActivateInset(cur, true)) {
-			if (cur.textRow().endpos() == (cur.pos() + 1) &&
+			if (cur.textRow().endpos() == cur.pos() + 1 &&
+			cur.lastpos() != cur.pos() + 1 &&
 			!cur.paragraph().isLineSeparator(cur.pos()) &&
 			!cur.paragraph().isNewline(cur.pos()))
 			{

[PATCH] bug 1926: Paragraph& DocIterator::paragraph() by failing check "inTexted()" in file dociterator.C:144

[Jean-Marc Lasgouttes]

This patch fixes the crash described in 
http://bugzilla.lyx.org/show_bug.cgi?id=1926

  ``Start a new doc, and start a math formula. Enter it, and leaving
  the cursor inside the formula, start another new doc. Move back to
  the old doc via the View menu and:

  Assertion triggered in Paragraph& DocIterator::paragraph() by
  failing check "inTexted()" in file dociterator.C:144''

Testing appreciated, although it seems trivially correct (I tried to
use LFUN_GETLAYOUT but it (1) leaves the current math inset and (2)
displays the result of the lfun in the minibuffer).

JMarc

Index: src/BufferView_pimpl.C
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/BufferView_pimpl.C,v
retrieving revision 1.594
diff -u -p -r1.594 BufferView_pimpl.C
--- src/BufferView_pimpl.C	7 Sep 2005 10:00:17 -	1.594
+++ src/BufferView_pimpl.C	13 Sep 2005 15:21:25 -
@@ -386,8 +386,20 @@ void BufferView::Pimpl::setBuffer(Buffer
 	owner_->updateWindowTitle();
 
 	// This is done after the layout combox has been populated
-	if (buffer_)
-		owner_->setLayout(cursor_.paragraph().layout()->name());
+	if (buffer_) {
+		size_t i = cursor_.depth() - 1;
+		// we know we'll eventually find a paragraph
+		while (true) {
+			CursorSlice const & slice = cursor_[i];
+			if (!slice.inset().inMathed()) {
+LyXLayout_ptr const layout = slice.paragraph().layout();
+owner_->setLayout(layout->name());
+break;
+			}
+			BOOST_ASSERT(i>0);
+			--i;
+		}
+	}	
 
 	if (buffer_ && lyx::graphics::Previews::status() != LyXRC::PREVIEW_OFF)
 		lyx::graphics::Previews::get().generateBufferPreviews(*buffer_);
Index: src/ChangeLog
===
RCS file: /usr/local/lyx/cvsroot/lyx-devel/src/ChangeLog,v
retrieving revision 1.2277
diff -u -p -r1.2277 ChangeLog
--- src/ChangeLog	10 Sep 2005 13:29:43 -	1.2277
+++ src/ChangeLog	13 Sep 2005 15:21:26 -
@@ -1,3 +1,9 @@
+2005-09-13  Jean-Marc Lasgouttes  <[EMAIL PROTECTED]>
+
+	* BufferView_pimpl.C (setBuffer): when one wants to set the layout
+	combox, it is necessary to search for the topmost paragraph in the
+	cursor stack (bug 1926)
+
 2005-09-10  Georg Baum  <[EMAIL PROTECTED]>
 
 	* text.C (leftMargin): check for inInset == 0

Re: large executable

[Lars Gullik Bjønnes]

Pol <[EMAIL PROTECTED]> writes:

| Jean-Marc Lasgouttes wrote:
| > what does "size ./lyx" day? Here (gcc 3.4.1) I get:
| > fantomas: size src/lyx
| >textdata bss dec hex filename
| > 6083056   17680   35060 6135796  5d9ff4 src/lyx
| >
| 
| % size /usr/local/bin/lyx
|   textdata bss  dec hex   filename
|11134899 394540   35332 11564771b076e3 /usr/local/bin/lyx

Remind me on the platform again please.

Re: Errors building 1.4.0 (?)

[Lars Gullik Bjønnes]

John Levon <[EMAIL PROTECTED]> writes:

| On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:
| 
| > It looks like configure does not find the qt libary (but it seems to find
| > the headers). This looks like a configure bug: It should stop with an error
| > message if the library cannot be found.
| 
| IIRC Lars changed this.

firm beliver in running to completion.

Re: bug in tex2lyx windows

[Lars Gullik Bjønnes]

Luis Rivera <[EMAIL PROTECTED]> writes:

| Hello,
| 
| I tried to run tex2lyx on win95/98, without success.  The programs
| crashed crying as follows:
| 
| On win95
| 
| The TEX2LYX.EXE file is linked to missing export
| KERNEL32.DLL:GetFileAttributesExA

Can't we just declare that we do not support Win95?

How old is XP now? 4 years?

Re: Errors building 1.4.0 (?)

[Lars Gullik Bjønnes]

Georg Baum <[EMAIL PROTECTED]> writes:

| Am Dienstag, 13. September 2005 14:30 schrieb John Levon:
| > On Tue, Sep 13, 2005 at 09:24:07AM +0200, Georg Baum wrote:
| > > It looks like configure does not find the qt libary (but it seems to find
| > > the headers). This looks like a configure bug: It should stop with an
| > > error message if the library cannot be found.
| >
| > IIRC Lars changed this.
| 
| I now remember that, too. I also remember that I added error messages for moc 
| and uic, so here comes the logical extension.
| 
| Mike, could you please test whether configure aborts with this patch (and 
| without the other one) if it cannot find the qt library?

I hate configures that aborts. IMHO it should run to completion and
tell me about all important stuff that went wrong.