This patch reduces the number of calls of ::exit() and hands the return
value of main() trough the various exec functions instead. The difference
to the one I sent earlier is that it does not remove the QApplication hack
in qt4, because that caused a crash on windows.
This will go in this evening.
Georg Baum [EMAIL PROTECTED] writes:
| This patch reduces the number of calls of ::exit() and hands the return
| value of main() trough the various exec functions instead. The difference
| to the one I sent earlier is that it does not remove the QApplication hack
| in qt4, because that caused a
Andre == Andre Poenitz [EMAIL PROTECTED] writes:
Andre On Sun, Jul 02, 2006 at 10:48:24PM +0100, Jose' Matos wrote:
Torino (Turin) is a fine place, and I should tell you that they
were supporting the french team yesterday. :-)
Andre Supporting a French team in Torino?
Andre Looks like I lost
Andre == Andre Poenitz [EMAIL PROTECTED] writes:
Andre 'Using LyX 1.5 on a distribution shipped in July 2006' means
Andre 'installing LyX 1.5'. Installing Qt 4 in the same place is
Andre about the same (I'd even argue 'less') effort and requires the
Andre same set of priviledges.
Are we 100%
Angus == Angus Leeming [EMAIL PROTECTED] writes:
Author: bpeng Date: Tue Jul 4 15:58:52 2006 New Revision: 14337
URL: http://www.lyx.org/trac/changeset/14337 Log: lib/configure.py:
fix checkTeXAllowSpaces, by Enrico
Modified: lyx-devel/branches/BRANCH_1_4_X/lib/ChangeLog
Lars == Lars Gullik Bjønnes [EMAIL PROTECTED] writes:
Lars Georg Baum [EMAIL PROTECTED] writes: | This patch
Lars reduces the number of calls of ::exit() and hands the return |
Lars value of main() trough the various exec functions instead. The
Lars difference | to the one I sent earlier is that
On Wednesday 05 July 2006 10:59, Jean-Marc Lasgouttes wrote:
Andre I think it is sufficient to keep Qt 3 compilable, let GTK rot
Andre (in the sense to declare it 'tolerated, but not supported by
Andre the core team'), and put all effort into Qt 4.
Also, are we sure that at that time qt4 will
Jose' == Jose' Matos [EMAIL PROTECTED] writes:
Jose' Without too much data, I have faith that it will, if not
Jose' become faster than qt3. I have two reasons for this, there are
Jose' (will be) more lyx developers working in qt4, and qt4 as André
Jose' says is improving speedwise (but not
Lars Gullik Bjønnes wrote:
It looks like you change behaviour. Did we use to exit if it was
impossible to read lyxrc.dist? I think we do after your patch.
The behaviour is not changed: We exit if lyxrc.dist exists, but lyxrc.read()
was not successful. We don't exit if the file does not exist.
Georg Baum [EMAIL PROTECTED] writes:
| Lars Gullik Bjønnes wrote:
|
| It looks like you change behaviour. Did we use to exit if it was
| impossible to read lyxrc.dist? I think we do after your patch.
|
| The behaviour is not changed: We exit if lyxrc.dist exists, but lyxrc.read()
| was not
Lars Gullik Bjønnes wrote:
Georg Baum [EMAIL PROTECTED]
writes:
| Lars Gullik Bjønnes wrote:
|
| It looks like you change behaviour. Did we use to exit if it was
| impossible to read lyxrc.dist? I think we do after your patch.
|
| The behaviour is not changed: We exit if lyxrc.dist
Georg Baum [EMAIL PROTECTED] writes:
| Lars Gullik Bjønnes wrote:
|
| Georg Baum [EMAIL PROTECTED]
| writes:
|
| | Lars Gullik Bjønnes wrote:
| |
| | It looks like you change behaviour. Did we use to exit if it was
| | impossible to read lyxrc.dist? I think we do after your patch.
On Tue, Jul 04, 2006 at 09:46:48AM +0200, Jean-Marc Lasgouttes wrote:
Lars == Lars Gullik Bjønnes [EMAIL PROTECTED] writes:
Lars This patch speeds up the scroll-down test by and odd second or
Lars two for me. Do others see the same? I guess it cannot do any
Lars harm, so committing.
On Tue, Jul 04, 2006 at 02:05:12PM +0200, Enrico Forestieri wrote:
Leave it out. Letting vim read settings from files of unknown source is a
security risk, so it is disabled by default, and we should not encourage
people to switch this on. I guess it is similar with emacs.
I cannot resist
On Tue, Jul 04, 2006 at 12:01:30PM +, Angus Leeming wrote:
Now that Lars has installed Trac, commit mails sent out automatically
by lyx-cvs all have hyperlinks to the changes.
Which is, on a side-note, useless for people that are not connected to
the net most of the time (i.e. 'me').
Could
Andre Poenitz [EMAIL PROTECTED] writes:
| On Tue, Jul 04, 2006 at 12:01:30PM +, Angus Leeming wrote:
| Now that Lars has installed Trac, commit mails sent out automatically
| by lyx-cvs all have hyperlinks to the changes.
|
| Which is, on a side-note, useless for people that are not
Hi, all,
I just rewrote the boost handling of scons.
1. Stop building dependency tree of boost libraries when using local
boost. This reduces the null build time of lyx from 29s to 16s. A
significant improvement. (I did not know scons spends so much time on
boost header files).
2. Check for
Bo == Bo Peng [EMAIL PROTECTED] writes:
Bo 1. Stop building dependency tree of boost libraries when using
Bo local boost. This reduces the null build time of lyx from 29s to
Bo 16s. A significant improvement. (I did not know scons spends so
Bo much time on boost header files).
What happens when
On 7/5/06, Jean-Marc Lasgouttes [EMAIL PROTECTED] wrote:
What happens when a file from local boost is changed? This happens
from time to time (small tweaks to allow newer gcc or odd system).
Scons will not know that so a manual full rebuild is required (remove
build directory and rebuild). I
Bo Peng wrote:
On 7/5/06, Jean-Marc Lasgouttes [EMAIL PROTECTED]
wrote:
What happens when a file from local boost is changed? This happens
from time to time (small tweaks to allow newer gcc or odd system).
Scons will not know that so a manual full rebuild is required (remove
build
I am not sure if autotools can detect
such changes and do a rebuild automatically.
It can.
How? I only see boost directory is included, no other special
treatment. Do you mean make tracks boost dependency in each local .dep
folder, does it check time stamps of /usr/include?
autotools do that
Bo Peng wrote:
I am not sure if autotools can detect
such changes and do a rebuild automatically.
It can.
How? I only see boost directory is included, no other special
treatment. Do you mean make tracks boost dependency in each local .dep
folder, does it check time stamps of
You are right that boost is not touched very often, but I don't want to
think about that, I want the build system to do that for me. Adding 20
seconds to every build is far better than spending two hours to debug a
non-existing bug!
OK. Here is the deal. I have changed the logic of scons so
Am Mittwoch, 5. Juli 2006 21:03 schrieb Bo Peng:
OK. Here is the deal. I have changed the logic of scons so that boost
headers will be excluded from the dependency tree only when
boost=system. That is to say, developers can choose to install system
boost libraries and enjoy faster start (16s
Did something just change? I am getting automatically built RPMs which look
like this:
# ls -lt ~ftp/pub/lyx/devel/RHFC4/lyx-1.5.0svn-1_*
-rw-r--r-- 1 kayvan kayvan 4979312 Jul 5 15:35
/var/ftp/pub/lyx/devel/RHFC4/lyx-1.5.0svn-1_%{frontend}.x86_64.rpm
-rw-r--r-- 1 kayvan kayvan 9890110 Jul
D'uh!!! I just realized that I need to fix my build scripts to not
built the now non-existent xforms frontend.
Nevermind.
On Wed, Jul 05, 2006 at 06:45:59PM -0700, Kayvan A. Sylvan wrote:
Did something just change? I am getting automatically built RPMs which look
like this:
# ls -lt
Angus Leeming wrote:
Asger Ottar Alstrup [EMAIL PROTECTED] writes:
It's much better to make 10 commits during 3 hours. Then each diff is
manageable in a review, and it is technically easier to revert.
Lars has a point that it can improve your workflow when you need to
think small,
This patch reduces the number of calls of ::exit() and hands the return
value of main() trough the various exec functions instead. The difference
to the one I sent earlier is that it does not remove the QApplication hack
in qt4, because that caused a crash on windows.
This will go in this evening.
Georg Baum <[EMAIL PROTECTED]> writes:
| This patch reduces the number of calls of ::exit() and hands the return
| value of main() trough the various exec functions instead. The difference
| to the one I sent earlier is that it does not remove the QApplication hack
| in qt4, because that caused a
> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes:
Andre> On Sun, Jul 02, 2006 at 10:48:24PM +0100, Jose' Matos wrote:
>> Torino (Turin) is a fine place, and I should tell you that they
>> were supporting the french team yesterday. :-)
Andre> Supporting a French team in Torino?
Andre>
> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes:
Andre> 'Using LyX 1.5 on a distribution shipped in July 2006' means
Andre> 'installing LyX 1.5'. Installing Qt 4 in the same place is
Andre> about the same (I'd even argue 'less') effort and requires the
Andre> same set of priviledges.
> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:
>> Author: bpeng Date: Tue Jul 4 15:58:52 2006 New Revision: 14337
>>
>> URL: http://www.lyx.org/trac/changeset/14337 Log: lib/configure.py:
>> fix checkTeXAllowSpaces, by Enrico
>>
>> Modified:
> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
Lars> Georg Baum <[EMAIL PROTECTED]> writes: | This patch
Lars> reduces the number of calls of ::exit() and hands the return |
Lars> value of main() trough the various exec functions instead. The
Lars> difference | to the one I sent
On Wednesday 05 July 2006 10:59, Jean-Marc Lasgouttes wrote:
> Andre> I think it is sufficient to keep Qt 3 compilable, let GTK rot
> Andre> (in the sense to declare it 'tolerated, but not supported by
> Andre> the core team'), and put all effort into Qt 4.
>
> Also, are we sure that at that time
> "Jose'" == Jose' Matos <[EMAIL PROTECTED]> writes:
Jose'> Without too much data, I have faith that it will, if not
Jose'> become faster than qt3. I have two reasons for this, there are
Jose'> (will be) more lyx developers working in qt4, and qt4 as André
Jose'> says is improving speedwise
Lars Gullik Bjønnes wrote:
> It looks like you change behaviour. Did we use to exit if it was
> impossible to read lyxrc.dist? I think we do after your patch.
The behaviour is not changed: We exit if lyxrc.dist exists, but lyxrc.read()
was not successful. We don't exit if the file does not
Georg Baum <[EMAIL PROTECTED]> writes:
| Lars Gullik Bjønnes wrote:
|
| > It looks like you change behaviour. Did we use to exit if it was
| > impossible to read lyxrc.dist? I think we do after your patch.
|
| The behaviour is not changed: We exit if lyxrc.dist exists, but lyxrc.read()
| was
Lars Gullik Bjønnes wrote:
> Georg Baum <[EMAIL PROTECTED]>
> writes:
>
> | Lars Gullik Bjønnes wrote:
> |
> | > It looks like you change behaviour. Did we use to exit if it was
> | > impossible to read lyxrc.dist? I think we do after your patch.
> |
> | The behaviour is not changed: We exit
Georg Baum <[EMAIL PROTECTED]> writes:
| Lars Gullik Bjønnes wrote:
|
| > Georg Baum <[EMAIL PROTECTED]>
| > writes:
| >
| > | Lars Gullik Bjønnes wrote:
| > |
| > | > It looks like you change behaviour. Did we use to exit if it was
| > | > impossible to read lyxrc.dist? I think we do
On Tue, Jul 04, 2006 at 09:46:48AM +0200, Jean-Marc Lasgouttes wrote:
> > "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
>
> Lars> This patch speeds up the scroll-down test by and odd second or
> Lars> two for me. Do others see the same? I guess it cannot do any
> Lars> harm, so
On Tue, Jul 04, 2006 at 02:05:12PM +0200, Enrico Forestieri wrote:
> > Leave it out. Letting vim read settings from files of unknown source is a
> > security risk, so it is disabled by default, and we should not encourage
> > people to switch this on. I guess it is similar with emacs.
> > I cannot
On Tue, Jul 04, 2006 at 12:01:30PM +, Angus Leeming wrote:
> Now that Lars has installed Trac, commit mails sent out automatically
> by lyx-cvs all have hyperlinks to the changes.
Which is, on a side-note, useless for people that are not connected to
the net most of the time (i.e. 'me').
Andre Poenitz <[EMAIL PROTECTED]> writes:
| On Tue, Jul 04, 2006 at 12:01:30PM +, Angus Leeming wrote:
| > Now that Lars has installed Trac, commit mails sent out automatically
| > by lyx-cvs all have hyperlinks to the changes.
|
| Which is, on a side-note, useless for people that are not
Hi, all,
I just rewrote the boost handling of scons.
1. Stop building dependency tree of boost libraries when using local
boost. This reduces the null build time of lyx from 29s to 16s. A
significant improvement. (I did not know scons spends so much time on
boost header files).
2. Check for
> "Bo" == Bo Peng <[EMAIL PROTECTED]> writes:
Bo> 1. Stop building dependency tree of boost libraries when using
Bo> local boost. This reduces the null build time of lyx from 29s to
Bo> 16s. A significant improvement. (I did not know scons spends so
Bo> much time on boost header files).
What
On 7/5/06, Jean-Marc Lasgouttes <[EMAIL PROTECTED]> wrote:
What happens when a file from local boost is changed? This happens
from time to time (small tweaks to allow newer gcc or odd system).
Scons will not know that so a manual full rebuild is required (remove
build directory and rebuild). I
Bo Peng wrote:
> On 7/5/06, Jean-Marc Lasgouttes <[EMAIL PROTECTED]>
> wrote:
>> What happens when a file from local boost is changed? This happens
>> from time to time (small tweaks to allow newer gcc or odd system).
>
> Scons will not know that so a manual full rebuild is required (remove
>
I am not sure if autotools can detect
such changes and do a rebuild automatically.
It can.
How? I only see boost directory is included, no other special
treatment. Do you mean make tracks boost dependency in each local .dep
folder, does it check time stamps of /usr/include?
autotools do that
Bo Peng wrote:
>> I am not sure if autotools can detect
>> such changes and do a rebuild automatically.
>>
>> It can.
>
> How? I only see boost directory is included, no other special
> treatment. Do you mean make tracks boost dependency in each local .dep
> folder, does it check time stamps of
You are right that boost is not touched very often, but I don't want to
think about that, I want the build system to do that for me. Adding 20
seconds to every build is far better than spending two hours to debug a
non-existing bug!
OK. Here is the deal. I have changed the logic of scons so
Am Mittwoch, 5. Juli 2006 21:03 schrieb Bo Peng:
> OK. Here is the deal. I have changed the logic of scons so that boost
> headers will be excluded from the dependency tree only when
> boost=system. That is to say, developers can choose to install system
> boost libraries and enjoy faster start
Did something just change? I am getting automatically built RPMs which look
like this:
# ls -lt ~ftp/pub/lyx/devel/RHFC4/lyx-1.5.0svn-1_*
-rw-r--r-- 1 kayvan kayvan 4979312 Jul 5 15:35
/var/ftp/pub/lyx/devel/RHFC4/lyx-1.5.0svn-1_%{frontend}.x86_64.rpm
-rw-r--r-- 1 kayvan kayvan 9890110 Jul
D'uh!!! I just realized that I need to fix my build scripts to not
built the now non-existent xforms frontend.
Nevermind.
On Wed, Jul 05, 2006 at 06:45:59PM -0700, Kayvan A. Sylvan wrote:
> Did something just change? I am getting automatically built RPMs which look
> like this:
>
> # ls -lt
Angus Leeming wrote:
Asger Ottar Alstrup <[EMAIL PROTECTED]> writes:
It's much better to make 10 commits during 3 hours. Then each diff is
manageable in a review, and it is technically easier to revert.
Lars has a point that it can improve your workflow when you need to
think small,
54 matches
Mail list logo
