Re: svn commit: rev 2428 - trunk/subversion/include trunk/subversion/libsvn_fs trunk/subversion/libsvn_wc trunk/subversion/libsvn_subr trunk/subversion/libsvn_client

Karl Fogel <kfogel@newton.ch.collab.net> writes:

> Philip Martin <philip@codematters.co.uk> writes:
> > I'm seeing lots of failures in the regression tests when configured
> > without --enable-utf8. Recompiling only utf.c with -DSVN_UTF8 (as
> > well as relinking) appears to solve the problem. So, although this
> > patch introduced the failures, I suspect utf.c rather than the files
> > changed here.
> >
> > The first test that fails is basic_tests.py number 4.
>
> Yes, Mike Pilato just reported similar lossage to me when I got in
> this morning.
>
> Sorry about that, folks.
>
> I'm going to revert right away, then work on the problem.

I suspect the problem is that the non-utf8 functions in utf.c
don't always allocate memory:

Index: ../svn/subversion/libsvn_subr/utf.c
===================================================================
--- ../svn/subversion/libsvn_subr/utf.c
+++ ../svn/subversion/libsvn_subr/utf.c Mon Jul 8 18:08:59 2002
@@ -365,8 +365,9 @@
                          const char **dest,
                          apr_pool_t *pool)
 {
- SVN_ERR (check_non_ascii (src, strlen (src), pool));
- *dest = src;
+ apr_size_t len = strlen (src);
+ SVN_ERR (check_non_ascii (src, len, pool));
+ *dest = apr_pmemdup (pool, src, len + 1);
   return SVN_NO_ERROR;
 }

-- 
Philip
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org