update on what is happening regarding the build.

The core problem is that when gump passes lib.dir in it passes it as a
value, and we want it bound to a location, which is what we do when we
declare it. So we could hack around the build files to do that (declare a
second lib.dir shadow property that we set it to force it into place.

<property name="fixedlib.dir" location="lib.dir"/>

But I'd still like to know what is going on.

----- Original Message -----
From: "Stefan Bodewig" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Wednesday, January 08, 2003 03:23
Subject: Re: Ant changes breaking Axis


> On Tue, 7 Jan 2003, Steve Loughran <[EMAIL PROTECTED]> wrote:
>
> > Whatever has happened to Ant recently has broken Axis
>
> At first glance, it seems to be a Gump problem.
>
> The exception happens when the reference to the <path> named
> "classpath" defined in axis/java/xmls/path_refs.xml gets expanded
> inside the <taskdef> from axis/java/xmls/taskdefs_post_compile.xml
> that is included in axis/java/samples/transport/build.xml, which in
> turn has been called by <ant> from axis/java/buildTest.xml.
>
> You sure know how to complicate builds ;-)
>
> Gump invokes Ant with -Dlib.dir=lib which then overrides the
> definition in properties.xml (that would point to ${axis.home}/lib
> using the absolute path).
>
> This in turn means that this part
>
>     <fileset dir="${lib.dir}">
>         <include name="**/*.jar"/>
>     </fileset>
>
> of path_refs.xml now is "lib" relative to the basedir of the build
> file - in build.xml it is axis/java/samples/transport/lib.  The
> problem we see.
>
> The "Caught exception" message comes from the <path> included, it
> doesn't hurt too much.
>
> The build failure happens, when the path gets used in the first
> taskdef from taskdefs_post_compile.
>
> After this analysis, -Dlib.dir=lib looks the problem and has been
> introduced a month ago [1].  I'd be content and happy, but then again
> - it gets built successfully in my Gump environment that uses the 1.5
> branch of Ant.  Honestly, I don't know why 8-)
>
> I'll investigate further, but thought I'd share what I've found so
> far.
>
> Stefan
>
> Footnotes:
> [1]
<http://cvs.apache.org/viewcvs.cgi/jakarta-gump/project/xml-axis.xml.diff?r1
=1.54&r2=1.55>
>
>
> --
> To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
>
>

Reply via email to