You are viewing a plain text version of this content. The canonical link for it is here.
Posted to general@gump.apache.org by Antoine Lévy-Lambert <an...@antbuild.com> on 2004/03/15 14:37:23 UTC
Persian enigma
xerces1 builds with classical gump, but not with gumpy.
Anybody has an idea, why this Persian Monarch does not like Pythons ?
Antoine
[1] http://gump.covalent.net/log/xml-xerces1.html
[2] http://lsd.student.utwente.nl/gump/xml-xerces/xml-xerces1.html
[3] http://article.gmane.org/gmane.comp.jakarta.gump/4847
The command line on lsd is :
java -Djava.awt.headless=true -Dbuild.clonevm=true org.apache.tools.ant.Main
-debug -Dgump.merge=/data3/gump/gump-install/work/merge.xml
-Dbuild.sysclasspath=only jar
4 of this type of compilation errors happen there :
/data3/gump/xml-xerces/java/build/src/org/apache/xerces/utils/regex/
RegularExpression.java:139: illegal unicode escape
[javac] * <kbd>\u005cu</kbd><var>c</var>, <kbd>\L</kbd>, <kbd>\U</kbd>,
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: Persian enigma
Posted by Stefan Bodewig <bo...@apache.org>.
On Mon, 15 Mar 2004, Antoine Lévy-Lambert <an...@antbuild.com>
wrote:
> which is the locale setting in which you run the build ?
well, while checking the environment I'm using when running in cron I
realized that my nightly Gump run uses JDK 1.4.1_02 and not the
1.4.2_03 that I run interactively.
This is what you get for having ten JDKs plus Kaffe on your machine 8-)
So I now officially support the "it's a bug in JDK 1.4.2" explanation,
1.4.2_04 is on its way to my box.
Stefan
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: Persian enigma
Posted by Antoine Lévy-Lambert <an...@antbuild.com>.
Stefan Bodewig wrote:
>On Mon, 15 Mar 2004, Antoine Lévy-Lambert <an...@antbuild.com>
>wrote:
>
>
>
>>what does Xvfb mean ?
>>
>>
>
>a virtual frame buffer X server. An X server that writes into memory
>and doesn't display anything anywhere. It could also be a different
>locale setting since Gump seems to use a different locale when it gets
>run by cron on my machine.
>
>
>
>>when it fails on you, do you get the same compilation errors as
>>gumpy,
>>
>>
>
>Yes.
>
>Stefan
>
>
>
Stefan,
which is the locale setting in which you run the build ?
- when you do it by cron,
- when if fails on you ?
the X frame server does not seem like a good explanation, the locale is
more appealing.
can we reasonably ask the xml-xerces developers to change the source
file which is causing the problem (after all it is only in JavaDoc
comments that we have problems).
Since I think they want to display a backslash in html, maybe changing
\u005c into \ would give the same result and not be sensitive to
locale setting on the gump machine.
Antoine
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: Persian enigma
Posted by Stefan Bodewig <bo...@apache.org>.
On Mon, 15 Mar 2004, Antoine Lévy-Lambert <an...@antbuild.com>
wrote:
> what does Xvfb mean ?
a virtual frame buffer X server. An X server that writes into memory
and doesn't display anything anywhere. It could also be a different
locale setting since Gump seems to use a different locale when it gets
run by cron on my machine.
> when it fails on you, do you get the same compilation errors as
> gumpy,
Yes.
Stefan
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: Persian enigma
Posted by Antoine Lévy-Lambert <an...@antbuild.com>.
Stefan,
what does Xvfb mean ?
when it fails on you, do you get the same compilation errors as gumpy,
(a complaint about forbidden \u escapes).
I have had a look (using Excel, thanks M$) to find out what the
forbidden unicode escape actually is, I think it is a back slash (92 in
decimal).
Antoine
Stefan Bodewig wrote:
>On Mon, 15 Mar 2004, Stefan Bodewig <bo...@apache.org> wrote:
>
>
>
>>The only differences I see are the two system properties since I run
>>Xvfb and set DISPLAY accordingly.
>>
>>
>
>It just failed for me when I run ./build.sh xml-xerces1 jar in an
>xterm without using Xvfb. This looks really strange.
>
>Stefan
>
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: Persian enigma
Posted by Stefan Bodewig <bo...@apache.org>.
On Mon, 15 Mar 2004, Stefan Bodewig <bo...@apache.org> wrote:
> The only differences I see are the two system properties since I run
> Xvfb and set DISPLAY accordingly.
It just failed for me when I run ./build.sh xml-xerces1 jar in an
xterm without using Xvfb. This looks really strange.
Stefan
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: Persian enigma
Posted by Stefan Bodewig <bo...@apache.org>.
On Mon, 15 Mar 2004, Antoine Lévy-Lambert <an...@antbuild.com>
wrote:
> xerces1 builds with classical gump, but not with gumpy.
I've unsuccessfully tried to understand it myself.
The command line on my machine is
java org.apache.tools.ant.Main -Dbuild.sysclasspath=only jar
and it works, if I use Gumpy, it doesn't work.
The only differences I see are the two system properties since I run
Xvfb and set DISPLAY accordingly.
I'll try to run "traditional" Gump with the system properties set to
see whether there will be any difference.
Stefan
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: build of jakarta-pluto - gump or gumpy problem ?[was : Persian
enigma]
Posted by Antoine Lévy-Lambert <an...@antbuild.com>.
Antoine Lévy-Lambert wrote:
> The project descriptor for jakarta-pluto says this :
>
> <jar id="container" name="container/target/pluto-1.0.jar"/>
> <jar id="api" name="api/target/portlet-api-1.0.jar"/>
> <jar id="portal" name="portal/target/pluto-portal-impl-1.0.jar"/>
>
> and the jars exist at these locations, ie for instance :
> /data3/gump/jakarta-pluto/container/target/pluto-1.0.jar
> but gumpy looks for :
> /data3/gump/jakarta-pluto/temp/container/target/pluto-1.0.jar
> which does not exist.
>
> the jakarta-pluto.xml file also contains these lines :
> <mkdir dir="temp/api/classes/"/>
> <mkdir dir="temp/container/classes/"/>
> <home nested="temp"/>
>
>
>
I think I misqualified the problem.
I have just changed the project descriptor for jakarta-pluto, removing
these home related settings.
I wonder whether, to the contrary, traditional gump does not check the
presence of the jars listed in the descriptor.
And should check ?
Cheers,
Antoine
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: gumpy bug building jakarta-pluto ?
Posted by Stefan Bodewig <bo...@apache.org>.
On Mon, 15 Mar 2004, Adam R. B. Jack <aj...@trysybase.com> wrote:
> but was (is) there a mistake by Gumpy?
No, I don't think so.
> I think Gumpy was right, but are you seeing traditional acting
> differently & working?
"traditional" doesn't flag a build failure for projects that don't
deliver the output they have promised. It helps a lot that Gumpy does
this since you'd only find a wrong jar name by a missing pre-req state
of another build in "traditional". Gumpy is a lot more correct here.
On the other hand, if a project delivers the jars and fails after
that, "traditional" Gump will use the jars while Gumpy will flag a
missing pre-req. Look at jstl-jsp12 and the -13 cactus projects for
an example.
Stefan
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: gumpy bug building jakarta-pluto ? [was : Persian enigma]
Posted by "Adam R. B. Jack" <aj...@trysybase.com>.
Antoine
I see you removed the <home nested="temp" />
> but gumpy looks for :
> /data3/gump/jakarta-pluto/temp/container/target/pluto-1.0.jar
> which does not exist.
>
> the jakarta-pluto.xml file also contains these lines :
> <mkdir dir="temp/api/classes/"/>
> <mkdir dir="temp/container/classes/"/>
> <home nested="temp"/>
but was (is) there a mistake by Gumpy? Reading :
http://gump.apache.org/metadata/project.html#home
I think Gumpy was right, but are you seeing traditional acting differently &
working? I'd be shocked to find traditional not working as expected in this
area, so ???
[Maybe we'll have to wait for Stefan to explain what traditional was doing,
and if I am reading that #home wrong.]
regards,
Adam
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: gumpy bug building jakarta-pluto ? [was : Persian enigma]
Posted by Antoine Lévy-Lambert <an...@antbuild.com>.
The project descriptor for jakarta-pluto says this :
<jar id="container" name="container/target/pluto-1.0.jar"/>
<jar id="api" name="api/target/portlet-api-1.0.jar"/>
<jar id="portal" name="portal/target/pluto-portal-impl-1.0.jar"/>
and the jars exist at these locations, ie for instance :
/data3/gump/jakarta-pluto/container/target/pluto-1.0.jar
but gumpy looks for :
/data3/gump/jakarta-pluto/temp/container/target/pluto-1.0.jar
which does not exist.
the jakarta-pluto.xml file also contains these lines :
<mkdir dir="temp/api/classes/"/>
<mkdir dir="temp/container/classes/"/>
<home nested="temp"/>
Julie MacNaught wrote:
> There is a similar problem with jakarta-pluto. I.e. builds on
> covalent, not on lsd .
>
> Julie
>
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: Persian enigma
Posted by Julie MacNaught <jm...@apache.org>.
There is a similar problem with jakarta-pluto. I.e. builds on covalent,
not on lsd .
Julie
Antoine Lévy-Lambert wrote:
> xerces1 builds with classical gump, but not with gumpy.
>
> Anybody has an idea, why this Persian Monarch does not like Pythons ?
>
> Antoine
>
> [1] http://gump.covalent.net/log/xml-xerces1.html
>
> [2] http://lsd.student.utwente.nl/gump/xml-xerces/xml-xerces1.html
>
> [3] http://article.gmane.org/gmane.comp.jakarta.gump/4847
>
> The command line on lsd is :
>
> java -Djava.awt.headless=true -Dbuild.clonevm=true
> org.apache.tools.ant.Main
> -debug -Dgump.merge=/data3/gump/gump-install/work/merge.xml
> -Dbuild.sysclasspath=only jar
> 4 of this type of compilation errors happen there :
>
> /data3/gump/xml-xerces/java/build/src/org/apache/xerces/utils/regex/
> RegularExpression.java:139: illegal unicode escape
> [javac] * <kbd>\u005cu</kbd><var>c</var>, <kbd>\L</kbd>,
> <kbd>\U</kbd>,
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
> For additional commands, e-mail: general-help@gump.apache.org
>
>
--
Julie MacNaught
IBM Research
jmacna@apache.org
jmacna@us.ibm.com
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org
Re: Persian enigma
Posted by Michael Davey <Mi...@coderage.org>.
Antoine Lévy-Lambert wrote:
> xerces1 builds with classical gump, but not with gumpy.
>
> Anybody has an idea, why this Persian Monarch does not like Pythons ?
>
> Antoine
>
> [1] http://gump.covalent.net/log/xml-xerces1.html
>
> [2] http://lsd.student.utwente.nl/gump/xml-xerces/xml-xerces1.html
>
> [3] http://article.gmane.org/gmane.comp.jakarta.gump/4847
>
> The command line on lsd is :
>
> java -Djava.awt.headless=true -Dbuild.clonevm=true
> org.apache.tools.ant.Main
> -debug -Dgump.merge=/data3/gump/gump-install/work/merge.xml
> -Dbuild.sysclasspath=only jar
> 4 of this type of compilation errors happen there :
>
> /data3/gump/xml-xerces/java/build/src/org/apache/xerces/utils/regex/
> RegularExpression.java:139: illegal unicode escape
> [javac] * <kbd>\u005cu</kbd><var>c</var>, <kbd>\L</kbd>,
> <kbd>\U</kbd>,
search the xerces-j-dev list for messages with subject line containing
"illegal unicode escape".
summary
=======
starting from Java 1.5 (or 1.4.x when appropriate modifier flags are
used), norms for unicode
characters are made more stringent.
See also:
<http://developer.java.sun.com/developer/bugParade/bugs/4863451.html>
Looks like Neeraj fixed Xerces2 but not Xerces1:
http://cvs.apache.org/viewcvs.cgi/xml-xerces/java/src/org/apache/xerces/impl/xpath/regex/RegularExpression.java?r1=1.6&r2=1.7
suggested fix: change the javadoc comments for Xerces1, to be compliant:
Manually edit the source, replacing "\u005cu" with "\u005c u".
--
Michael
---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org