Hi Pavel,
On 10.07.2012 21:47, Pavel Janík wrote:
Andre,
this particular change is incompatible with my build system. I do [meta-shell
code]:
I should have explained this change better (or at all). It is only
intended as a temporary hack to find out the reason for the build
breaker of the 64bit Linux buildbot build. That hangs when compiling
Java code. The buildbot eventually terminates the build (after 12000s)
and without my change all output is lost. With the change the output is
available and I was able to locate the problem in the jurt module. But
I still have no clue why compiling a single Java file (after having
compiled many already) could lead to a hang.
Having said that I have no problem with you reverting the change. But
it may be necessary to turn it back on for short times in the future to
hunt down other buildbot problems.
cd instsetoo_util; build --all --html
The trunk now does this:
build --all works as expected:
dmake: Error code 2, while making 'all'
1 module(s):
framework
need(s) to be rebuilt
Reason(s):
ERROR: error 65280 occurred while making
/Users/pavel/BUILD/BuildDir/ooo_trunk_src/framework/prj
When you have fixed the errors in that module you can resume the build by
running:
build --all:framework
BUT:
build --all --html ignores build errors and packs instsets:
ERROR: File not found: smen-US.res
ERROR: File not found: svxen-US.res
ERROR: File not found: textconversiondlgsen-US.res
ERROR: File not found: services.rdb
... cleaning the output tree ...
... removing directory
/var/folders/Lm/LmD3R9N-G681RwIIfVAMbE+++TI/-Tmp-/ooopackaging/i_422831341949313
...
Tue Jul 10 21:41:55 2012 (00:02 min.)
dmake: Error code 255, while making 'openoffice_en-US.archive'
Pavel-Janiks-MacBook-Pro:util pavel$
I am not quite sure that I understand the problem. Without the --html
switch you have one error (in framework), with the --html switch you
have another (later in instsetoo_native).
Reverting this change made it work again as expected.
Does that mean that it breaks in framework or that it builds without error?
It is strange in any case that the change should have that effect. But
then again, in build.pl the processing of all the switches (I counted
33) is interwoven and hard to separate.
-Andre
On Jul 4, 2012, at 5:26 PM, a...@apache.org wrote:
Author: af
Date: Wed Jul 4 15:26:40 2012
New Revision: 1357306
URL: http://svn.apache.org/viewvc?rev=1357306&view=rev
Log:
Show output synchronously for build --html
Modified:
incubator/ooo/trunk/main/solenv/bin/build.pl
Modified: incubator/ooo/trunk/main/solenv/bin/build.pl
URL:
http://svn.apache.org/viewvc/incubator/ooo/trunk/main/solenv/bin/build.pl?rev=1357306&r1=1357305&r2=1357306&view=diff
==============================================================================
--- incubator/ooo/trunk/main/solenv/bin/build.pl (original)
+++ incubator/ooo/trunk/main/solenv/bin/build.pl Wed Jul 4 15:26:40 2012
@@ -2017,10 +2017,7 @@ sub run_job {
if (!-d $log_dir) {
system("$perl $mkout");
};
- $error_code = system ("$job_to_do > $log_file 2>&1");
- if (!$grab_output && -f $log_file) {
- system("cat $log_file");
- };
+ $error_code = system ("$job_to_do 2>&1 | tee $log_file");
} else {
$error_code = system ("$job_to_do");
};