Re: [Libreoffice] Can I compile and build only libreoffice writer component instead of whole LO suite ?
On Sat, Aug 13, 2011 at 12:20 AM, Partha Chowdhury wrote: > Hello all.This is my first post to the list.So if I have not provided > necessary information or overlooked something obvious, please point me in > the right direction. > > I have build libreoffice 3.3.3.1 from source using the > "./configure;make;bin/ooinstall ;" .However, I need office > only to create and modify word documents.Thus i want to only compile the > libreoffice-writer component and not the draw, impress,calc etc,if that is > possible. short answer: no. longer answer... everything is 'possible' but that would require significant hacking, and you may not save as much as you'd imagined. Norbert ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] [PATCH] Series of fixes and migration to One Git for contrib/dev-tools/ubuntu-jenkins
On Thursday, 11 August 2011, 0:56 Norbert Thiebaud wrote: >On Wed, Aug 10, 2011 at 4:56 PM, Henrik Jensen >wrote: >> Patches licensed under LGPLv3+/MPL 1.1 ( or what ever Bjoern Michaelsen >> prefers :-) ) >> TODO: >> - Add the ccache lines to './autogen.sh' command line instead of >> concatenating to 'XEnv.Set.sh/Env.Host.sh' > >ccache is now (on master) automatically used if present in PATH, so >there should not be any need for special ccache lines to autogen.sh >anymore > >Norbert Ok, well I guess the ccache lines could be removed from the libreoffice-master job in jenkins, but I imagined the ubuntu-jenkins as an easy "one click" method of getting a working build (with added administrative bonuses) for relative newbee/wouldbe developers of lo, so the target users probably does not have ccache set in the enviroment in the first place. ( Though I don't know if that's the way Bjoern Michaelsen visioned it, so I'll wait for his opinions on the subject.) --Henrik Jensen (HenrikJ on #libreoffice) ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] [PATCH][PARTIAL PUSH] More perl cleanup (fdo#39747)
On Fri, Aug 12, 2011 at 11:14 PM, Jordan Ayers wrote:
> A few more patches to clean up parts of the solenv perl scripts.
>
> The first is just removing pieces that became useless after prior
> patches, or that were added as part of a larger patch, even though
> they were never used. (I'm basing this on the master branch git log
> entries.)
Applied and pushed, thanks.
>
> The second has a minor impact on build.pl calling behavior, but seems
> reasonable. The --dontgraboutput switch hasn't done anything for some
> time, except cause build.pl to abort if this switch was specified
> before the --html switch. If specified now, it will pass through to
> the dmake options (where it shouldn't do anything); it can no longer
> prevent build.pl from running.
Applied and push, thanks.
>
> The last just shortens the debuginfo / debugsave calls; since every
> time these were used, they were wrapped in checks against the debug
> flag, I pushed the debug flag check into those routines. Now they can
> just be called, and if debugging is off they do nothing.
Actually this 'shorten' their appearance, but lengthen their execution.
Indeed before the cost of call to debuginfo/debugsave in non debug case, was
if(variable = true)
with this patch it is:
Evaluate the argument of the call, make the function call,
if(variable=true) return to the caller
This can be very bad (see example below), and can never be faster.
So I have not pushed this one.
Norbert
-
example:
#! /usr/bin/perl -w
use strict;
my $mode = $ARGV[0];
my $debug = $ARGV[2];
my %foo =();
$foo{'s'} = "bar";
$foo{'f'} = "foo";
if(!defined($debug))
{
$debug = 0;
}
sub sdebug
{
my ($message) = @_;
if($debug)
{
print $message;
}
}
sub fdebug
{
my ($message) = @_;
};
if($mode eq "f")
{
for(my $i = 0; $i < 10; $i++)
{
if($debug)
{
fdebug("message $i $foo{$mode} $#ARGV\n");
}
}
}
else
{
for(my $i = 0; $i < 10; $i++)
{
sdebug("message $i $foo{$mode} $#ARGV\n");
}
}
n_th@tpamac ~/bin $time tperfperl.pl s
real0m0.180s
user0m0.173s
sys 0m0.006s
n_th@tpamac ~/bin $time tperfperl.pl s
real0m0.178s
user0m0.171s
sys 0m0.006s
n_th@tpamac ~/bin $time tperfperl.pl s
real0m0.177s
user0m0.171s
sys 0m0.006s
n_th@tpamac ~/bin $time tperfperl.pl f
real0m0.017s
user0m0.010s
sys 0m0.006s
n_th@tpamac ~/bin $time tperfperl.pl f
real0m0.018s
user0m0.010s
sys 0m0.006s
n_th@tpamac ~/bin $time tperfperl.pl f
real0m0.017s
user0m0.010s
sys 0m0.006s
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] [PATCH] Series of fixes and migration to One Git for contrib/dev-tools/ubuntu-jenkins
> I guess Bjoern knows most about this, adding him to cc list, think he is on > vacation now but surely will look when he gets back > >Noel Ok, I've put Bjoern Michaelsen on CC for some further small patches. Hope his is alright with that. --Henrik Jensen (HenrikJ on #libreoffice-dev) ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] Can I compile and build only libreoffice writer component instead of whole LO suite ?
Hello all.This is my first post to the list.So if I have not provided necessary information or overlooked something obvious, please point me in the right direction. I have build libreoffice 3.3.3.1 from source using the "./configure;make;bin/ooinstall ;" .However, I need office only to create and modify word documents.Thus i want to only compile the libreoffice-writer component and not the draw, impress,calc etc,if that is possible. At first I asked the question on the user's list, there I was advised to post here. I have tried abiword, but 1> it is lightweight, so lacks many features compared to LO writer 2>It cannot render the word documents created by Ms office 2003 properly.The file looks gibberish. I am using Linux From scratch, which builds a fast and minimal GNU/Linux system entirely from source code. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] [PATCH] Series of fixes and migration to One Git for contrib/dev-tools/ubuntu-jenkins
A few more suggested patches to 'setup-ubuntu-jenkins.sh'
( committed to local clone after
0007-Install-Jenkins-in-current-dir-instead-of-.jenkins.patch
so I guess they are dependent )
Patch descriptions:
---
0001-Add-install-path-to-cmdln.-args.patch:
Add install path to cmdln. args.
Per default installs in ~/.jenkins or the path added to the command line
of setup-ubuntu-jenkins.sh
---
0002-Add-simple-instructions-for-h-help.patch:
Add simple instructions for -h/--help
---
--Henrik Jensen
(HenrikJ on #libreoffice-dev)
On Wed, 2011-08-10 at 22:56 +0100, Henrik Jensen wrote:
Patches licensed under LGPLv3+/MPL 1.1 ( or what ever Bjoern Michaelsen prefers
:-) )
>
>
> A series of 7 suggested patches to fix some bugs and migrate to One Git in
> Bjoern Michaelsen
> Jenkins Continuous Integration Serversetup found in
> contrib/dev-tools/ubuntu-jenkins
>
> Patch descriptions:
> ---
> 0001-Bugfix-Prevent-redundant-tarfile-downloads.patch
>
> Bugfix: Prevent redundant tarfile downloads.
>
> './download' sources 'Env.Host.sh', not 'LinuxX86-64Env.Set.sh',so we must add
> 'set_tarfile_location.sh' to 'Env.Host.sh' after first './autogen.sh' run.
> I speculate if this is an artifact of the cloning from the master and
> "down-branching"
> to 3-4 !!!? - Maybe ./download should be called after the second autogen.sh
> though I
> can't yet see through the consequences of that?
>
> ---
> 0002-Bugfix-x86-processor-architecture-agnostic.patch
>
> Bugfix: x86 processor architecture agnostic.
>
> Checks if 2nd. run of 'autogen.sh' has generated 'LinuxX86Env.Set.sh' or
> 'LinuxX86-64Env.Set.sh' and use the appropriate one.
>
> ---
> 0003-Bugfix-Using-the-new-Env.Host.sh-in-libreoffice-mast.patch
>
> Bugfix: Using the new 'Env.Host.sh' in libreoffice-master instead of
> hardcoding 'LinuxX86-64Env.Set.sh'
>
> ---
> 0004-Bugfix-libreoffice-master-job-needs-a-make-before-ma.patch
>
> Bugfix: 'libreoffice-master' job needs a 'make' before 'make dev-install'
>
> 'dev-install' has dependencies to the 'all' target but it's not reflected in
> the
> lo-root makefile.
>
> ---
> 0005-Migrate-to-One-Git-keep-possibility-for-3-4-build.patch
>
> Migrate to One Git, keep possibility for 3-4 build.
>
> - Cloning from 'repo-mirror/core.git' for 'libreoffice-master'.
> - Cloning from 'repo-mirror-pre-one-git/bootstrap.git' for 'libreoffice-3-4'.
> - Adding a 'repo-mirror-pre-one-git' to still support the 'libreoffice-3-4'
> job
> - Disable cron schedule for the new (old) 'repo-mirror-pre-one-git'.
>
> Adding 2 repo-mirrors to support both the new master and the 'libreoffice-3-4'
> seems a bit unnecessary, but as I understand from asking on #libreoffice-dev
> the new one-git master can't be used to checkout libreoffice.3.4.
> Also tried a './g checkout libreoffice-3-4' on the new one-git repo
> Output: "error: pathspec 'libreoffice-3-4' did not match any file(s) known to
> git."
>
> A solution might be to only mirror the old locked remote master gits and
> locally use
> the onegit.sh conversion script located in
> 'anongit.freedesktop.org/libreoffice/contrib/dev-tools/onegit' to create a new
> copy and then 'git fetch --all --tags' to update it to the newest?
>
> ---
> 0006-Using-the-JENKINS_HOME-variable.patch
>
> Using the ${JENKINS_HOME} variable
>
> Using the ${JENKINS_HOME} variable instead of hardcoding to the '~/.jenkins'
> path
> This commit sets up an easier transition to a more generalized install
> procedure
> see: https://wiki.jenkins-ci.org/display/JENKINS/Winstone
> and https://wiki.jenkins-ci.org/display/JENKINS/Administering+Jenkins
> for the ${JENKINS_HOME} variable
>
> ---
> 0007-Install-Jenkins-in-current-dir-instead-of-.jenkins.patch
>
> Install Jenkins in current dir in stead of ~/.jenkins
>
> Let Jenkins be installed in current dir instead of force to '~/.jenkins'
> 'setup-ubuntu-jenkins.sh' now creates a 'start-lo-jenkins.sh' with the
> appropriate startup arguments
>
> -
Re: [Libreoffice] [PATCH] Improve the "alien save" dialog (part one)
Fantastic, thanks mate. I'll have another one next week with more complicated changes to make it even better so that should be good. Josh ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] [PUSHED] dbaccess/xmlfilter.cxx
On Mon, 2011-08-08 at 20:16 +0200, Jenei Gábor wrote: > Hello, > > I found out that on make -sr debug=true dbglevel=2 dbaccess did not > compile successfully, the problem is with xmlfilter.cxx, it uses > OStringBuffer, but there is no include to use it, so cxx says: > "OStringBuffer is not a member of rtl namespace" I did the include, and > the build works fine Thanks for this, just marking as pushed by Noel C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] Class Time and its funny ideas about time
On Thursday 11 of August 2011, Thorsten Behrens wrote: > Lubos Lunak wrote: > > As much as I don't like it, I can possibly see at least some reason for > > time having 25 hours, but 10 seconds without 20 seconds being 10 seconds > > (and there is actually explicit code to ensure that)? > > > > Does somebody know why the Time class does either of these and how much > > would break if I fixed these two to be sane? > > Reasons for that buried in the past I guess, and maybe not too > useful to try to answer - for the second question, GetHour() should > be easy to grep for, and operator- has at least client code in > editeng/source/misc/svxacorr.cxx, sfx2/source/doc/objcont.cxx, and > automation/source/server/statemnt.cxx > > It's unclear to me, how valuable it is to fix-up tools/time.hxx & > all client code, vs. to use e.g. > http://www.boost.org/doc/libs/1_47_0/doc/html/chrono.html when you > need a sane implementation in new code ... Makes sense, except that none of the Boost classes listed there is a suitable replacement for wall clock time (time_point does not seem to have any minutes(), or I'm again confused by Boost's posh APIs). > At the very least, documenting your findings in tools.hxx of course > would be great. Will do. -- Lubos Lunak l.lu...@suse.cz ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] Class Time and its funny ideas about time
On Thursday 11 of August 2011, Eike Rathke wrote: > > 2) Why should GetSec() of a negative delta return a negative amount of >seconds? I didn't say it should be negative, it just should be sane. Which in this case should be ... uhm ... oh well, another botched API. > The probability that something would break if you just "fixed" #2 is > quite high. The only way to make sure is to temporarily make the methods > private, build the entire office and inspect each and every place the > compiler complains about. Hint: It's simpler to check usage of something (no compilation errors) by marking it as deprecated, building with that and grepping the output for deprecation warnings. -- Lubos Lunak l.lu...@suse.cz ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] [patch] unused variable in Windows build
Hello, a small patch to suppress an unused variable (within a WNT define) according cppchecker Since I can not build on Windows, if someone could review / compile it for me before pushing ? Thanks and regards Pierre-André 0001-unused-variable-in-Windows-build.patch Description: application/mbox ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] oox fun in bringing it into tail_build...
On Fri, 12 Aug 2011 15:27:11 +0200 Bjoern Michaelsen wrote: > It would remove the touch-target of the animations header (the empty > file that signals to the build system 'everything for the header > target is done'), yes -- but it would not remove the headers > themselves: It doesnt even know about them as they are declared in > animations. So apart from a few possible superfluous relinkings of > libs using the animations headers, everything is fine. > > gb_Library_add_package_headers is the way to go. OTOH, the packages are registered in the module anyway, so a "make clean" in sd will remove the sd packages anyway. So maybe the dependency of the LinkTarget clean target on the package clean target can/should be removed anyway (thereby preventing the superfluous relinking)? Best, Bjoern -- https://launchpad.net/~bjoern-michaelsen ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] oox fun in bringing it into tail_build...
On Fri, 12 Aug 2011 01:59:37 -0500 Norbert Thiebaud wrote: > that would break the per-module build-ability, as you noticed below. No it shouldnt. > > headers because I found: > > > > define gb_LinkTarget_add_package_headers > > $(foreach package,$(2),$(call > > gb_LinkTarget__add_internal_headers,$(1),$(call > > gb_Package_get_target,$(package > > $(call gb_LinkTarget_get_clean_target,$(1)) : $(foreach > > package,$(2),$(call gb_Package_get_clean_target,$(package))) > > > > endef > > > > what probably means that when you run make clean in sd you would > > also remove animations' headers ? It would remove the touch-target of the animations header (the empty file that signals to the build system 'everything for the header target is done'), yes -- but it would not remove the headers themselves: It doesnt even know about them as they are declared in animations. So apart from a few possible superfluous relinkings of libs using the animations headers, everything is fine. gb_Library_add_package_headers is the way to go. Best, Bjoern -- https://launchpad.net/~bjoern-michaelsen ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] [Bug 35673] LibreOffice 3.4 most annoying bugs
https://bugs.freedesktop.org/show_bug.cgi?id=35673 mariosv changed: What|Removed |Added Depends on|39919 | -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] patch for preserving the table alignment
On Fri, 2011-08-12 at 10:45 +0100, Noel Power wrote: > > composite -compose difference old.pdf new.pdf - | identify -format %k - > did that for the document at iz#1291 and it shows no changes ( e.g. > result of 1 ) on my system That's good enough for me really. I'd say make your changes unless someone knows to the contrary. What we need here is a repo of pre rendered to pdf documents of "what currently works" for these issues. I really want to see first if we can bump up the resolution of the text layout to get it equivalent to the pdf export of the native apps that generates these formats to make it possible to directly compare without fuzzing. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] http://dev-www.libreoffice.org/bundles/ updated to OneGit
Hi, following opengrok and other services, the tar bundles of the git repos at http://dev-www.libreoffice.org/bundles/ are now also using the new OneGit layout. http://dev-www.libreoffice.org/bundles/libreoffice-git.tar.bz2 contains core, dictionaries, and help http://dev-www.libreoffice.org/bundles/libreoffice-extra-git.tar.bz2 contains binfilter and translations (note: those are meant as fallbacks for people having trouble cloning from [anon]git.freedesktop.org directly, as outlined here: http://wiki.documentfoundation.org/Development/Native_Build#Getting_the_sources) Cheers, -- Thorsten pgpKyDXT3EHp6.pgp Description: PGP signature ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] [PUSHED] Improve the "alien save" dialog (part one)
On Fri, 2011-08-12 at 14:02 +0930, Josh Heidenreich wrote: > Hi all, > > There has been a heap of discussion about this in another thread, and > I posted a patch there, but I though I should post a [PATCH] one too. > > This is version 3 of the patch. LGPLv3/MIT. Pushed now I believe. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] [PUSHED] "Confirmation of save format" dialog
On Fri, 2011-08-12 at 15:35 +0930, Josh Heidenreich wrote: > > I have the third version of the patch. > Oops, forgot to attach Hmm, its a patch against the original state of affairs, rather than a patch against the current one. Tweaked it up to apply and pushed. Best to update your sources (ensuring you're on the latest-and-greatest first http://www.documentfoundation.org/develop/) and make patches against "master" C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] patch for preserving the table alignment
Hmm seems I really need to return to this, comparing on windows ( where we really have like for like fonts ) it seems the problem returns ( because the column width issue comes into play again ) On 18/07/11 10:41, Caolán McNamara wrote: On Fri, 2011-07-15 at 16:35 +0100, Noel Power wrote: Hi Caolán could you have a look at this patch http://cgit.freedesktop.org/libreoffice/writer/commit/?id=95351a519bec2833f5d936c20e3916a4e283b0f6 This should be a nice stroll down memory lane for you as it seems related to http://openoffice.org/bugzilla/show_bug.cgi?id=1291, yes issue #1291 !!! Hmm, well I don't remember the code, I do remember the document itself, cause its a nasty. My intent with the SetFmtAttr is apparently to get a floating frame which shrinks right down the the size of the table contained in it, reset the positioning of the table so the left of the table is 0 from the left of the frame and the right 0 from the right of the frame so that it completely fills it widthwise. If "FULL" isn't actually "full width of the container", but automatic in a full-width-of-the-container, but clip-to-the-page-width, sort of way, that's a pain :-) yes that is what seems to happen ( although I failed to find exactly where this is happening ) I can rule out ( afaics ) the table implementation, so I guess it is happening in the layout stuff ( which I have don't even have the teeniest clue about ). So the Frame in this case has shrunk to the correct width of the table ( which is greater that the page width in my case ) but the actual table width ( when you raise the table properties dialog ) is the page width. This is due to the "FULL" setting ( corresponding to 'Automatic' in the Table properties ) which I guess is doing what it says on the tin, e.g. aligning with the page margins ( or I suppose really the container boundries or whichever is smaller it seems ) Bizarely: with either Left or 'From Left' alignment the a 'negative' right spacing is used to allow the table width to be larger than the container width. The right spacing value however will only influence the table width up to the container width ( or page width ) whichever is hit first I think I'd prefer an explicit horizontal setting which does the right thing, e.g. see if any of NONE/CENTER/LEFT do the right thing, rather than re-use the current setting which might be something weird like INSIDE, OUTSIDE or something and all we want is a table that completely fills the floating frame. ok, how about a limited hack here, e.g. if the horizontal setting is LEFT_AND_WIDTH ( which seems to be the converted table alignment setting in question for my test document and indeed the one from iz#1291 ) we set this at the Frame otherwise leave it as FULL. I suspect that LEFT_AND_WIDTH ( which I guess should try to fit the table into the container ( which is already sized ) aligned to the left with the right spacing adjusted to fit as appropriate ) probably would work in all cases but I wouldn't want to bet on it. FWIW, I suggest generally testing for changes in layout to existing docs with export to pdf before/after and with composite -compose difference old.pdf new.pdf - | identify -format %k - did that for the document at iz#1291 and it shows no changes ( e.g. result of 1 ) on my system, I'd be happy to check any other documents you could suggest Noel ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] [PATCH] Bug 38486
Hello,
Here is the patch that fixes the Query wizard, it seems that it works
fine now, however the problem mentioned in the patch was a minor
problem, wizard worked wrong for all queries where I wanted to sort by
an aggregate field. Please review it,and push it to git.
Thanks,
Gabor
>From 3210fcf363e242ecca3bfca686ef9b8ac91305ee Mon Sep 17 00:00:00 2001
From: Gabor Jenei
Date: Fri, 12 Aug 2011 10:15:24 +0200
Subject: [PATCH] Bug 38486 SQL fix
---
.../com/sun/star/wizards/db/SQLQueryComposer.java |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/wizards/com/sun/star/wizards/db/SQLQueryComposer.java b/wizards/com/sun/star/wizards/db/SQLQueryComposer.java
index a6d2c2b..3f595f2 100644
--- a/wizards/com/sun/star/wizards/db/SQLQueryComposer.java
+++ b/wizards/com/sun/star/wizards/db/SQLQueryComposer.java
@@ -205,7 +205,7 @@ public class SQLQueryComposer
{
sOrder += ", ";
}
-sOrder += CurDBMetaData.AggregateFieldNames[iAggregate][1] + "(" + CurDBMetaData.AggregateFieldNames[iAggregate][0] + ")";
+sOrder += CurDBMetaData.AggregateFieldNames[iAggregate][1] + "(" + getComposedAliasFieldName(CurDBMetaData.AggregateFieldNames[iAggregate][0]) + ")";
sOrder += " " + CurDBMetaData.getSortFieldNames()[i][1];
m_queryComposer.setOrder(sOrder);
}
--
1.7.2.5
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice
[Libreoffice] [PATCH] BUG 38486
Hello,
Here is a patch for the mentioned bug, actually the problem mentioned in
the bug info was just the minority, wizard didn't work for all the
queries where I wanted to sort by an aggregate field. I did several
tries,and it seems to be ok now, however without "group by" in an
aggregate query it seems to be silly to sort by anything, I must notice
Oracle even handles it as an error, it seems that for HSQL it's not a
problem. Please review the patch and push it to git.
Thanks,
Gabor
>From 3210fcf363e242ecca3bfca686ef9b8ac91305ee Mon Sep 17 00:00:00 2001
From: Gabor Jenei
Date: Fri, 12 Aug 2011 10:15:24 +0200
Subject: [PATCH] Bug 38486 SQL fix
---
.../com/sun/star/wizards/db/SQLQueryComposer.java |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/wizards/com/sun/star/wizards/db/SQLQueryComposer.java b/wizards/com/sun/star/wizards/db/SQLQueryComposer.java
index a6d2c2b..3f595f2 100644
--- a/wizards/com/sun/star/wizards/db/SQLQueryComposer.java
+++ b/wizards/com/sun/star/wizards/db/SQLQueryComposer.java
@@ -205,7 +205,7 @@ public class SQLQueryComposer
{
sOrder += ", ";
}
-sOrder += CurDBMetaData.AggregateFieldNames[iAggregate][1] + "(" + CurDBMetaData.AggregateFieldNames[iAggregate][0] + ")";
+sOrder += CurDBMetaData.AggregateFieldNames[iAggregate][1] + "(" + getComposedAliasFieldName(CurDBMetaData.AggregateFieldNames[iAggregate][0]) + ")";
sOrder += " " + CurDBMetaData.getSortFieldNames()[i][1];
m_queryComposer.setOrder(sOrder);
}
--
1.7.2.5
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice
