Bug#359909: subversion: FTBFS: Ruby test fails

[Peter Samuelson]

[Daniel Schepler]
> The patch didn't help.  After some more investigation, I found that
> installing official i386 ruby1.8 packages fixed the test failure, and
> in fact if I rebuild ruby1.8 locally adding "-fno-strict-aliasing" to
> the CFLAGS, the rebuilt packages also fix the test failure.

Thank you very much!  On the basis of those hints, I obtained access to
a Debian powerpc box this morning and did some tests.  My findings: at
least on powerpc, ruby1.8 compiled with gcc 4.0.3 works whereas ruby1.8
compiled with gcc 4.0.2 does not.  Of course, the version in sid was
compiled with gcc 4.0.2.

Interestingly, -fno-strict-aliasing didn't seem to have any effect,
only the compiler version.  (I'm talking about ruby1.8 itself - we've
always used -fno-strict-aliasing for the subversion ruby _bindings_.)

Steve Langasek has scheduled ruby1.8 to be rebuilt on powerpc, so we'll
soon see whether my observations are correct.

Thanks again,
Peter
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/pkg-subversion-maintainers/attachments/20060408/ca89f27c/attachment.pgp