Debuginfo packages for Java

[Andrew Haley]

Jason L Tibbitts III writes:
 > I'm having problems reviewing a package for some software written in
 > Java.  The problem is that the debuginfo package is generated without
 > any source.  find-debuginfo.sh prints out the following:
 > 
 > extracting debug info from /var/tmp/writer2latex-0.5-buildroot/usr/lib64/gcj/writer2latex/writer2latex-0.5.jar.so
 > cpio: writer2latex05/aot-compile-rpm/usr/lib64/gcj/writer2latex/writer2latex-0.5.jar.1.jar: Cannot stat: No such file or directory
 > cpio: writer2latex05/aot-compile-rpm/usr/lib64/gcj/writer2latex/writer2latex/Application.java: Cannot stat: No such file or directory
 > cpio: writer2latex05/aot-compile-rpm/usr/lib64/gcj/writer2latex/writer2latex/api/BatchConverter.java: Cannot stat: No such file or directory
 > 
 > Needless to say, the source isn't actually buried under
 > writer2latex05/aot-compile-rpm/usr/lib64/gcj/writer2latex.
 > 
 > Some time ago I reviewed some Java-using packages and the issue was
 > fixable with by making a symlink; see, for example, the ganymed-ssh2
 > package, which has:
 > 
 > # Link source files to fix -debuginfo generation.
 > rm -f ch
 > ln -s src/ch
 > 
 > but this method no longer works and in fact that ganymed-ssh2
 > debuginfo package currently includes no source.  Even if I cook up a
 > directory structure and symlinks so that
 > writer2latex05/aot-compile-rpm/usr/lib64/gcj/writer2latex exists and
 > holds all of the files cpio is complaining about, they still don't
 > make it into the debuginfo package.
 > 
 > So I'm at a loss.  We really need folks who understand java to come up
 > with guidelines and procedures that would answer the kinds of
 > questions which come up when reviewing Java-using packages.  There are
 > a number of them in the review queue but nobody understands how to
 > review them.  And it really wouldn't if someone documented just how
 > debuginfo generation works.
 > 
 > The review ticket in question is
 >   https://bugzilla.redhat.com/show_bug.cgi?id=386661
 > 
 > The ganymed-ssh2 ticket where I first tackled this:
 >   https://bugzilla.redhat.com/show_bug.cgi?id=191014
 > which includes some comments about debuginfo generation being busted
 > for Java.

I might be able to sort this out.  That ought only to be possible if
the package is built wrongly, perhaps with debug=off whan compiling.

I'll try to fix it sometime today.  Ping me if I don't reply.

Andrew.

-- 
Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, UK
Registered in England and Wales No. 3798903