Mailinglist Archive: opensuse-factory (340 mails)

< Previous Next >
Re: [opensuse-factory] Certain patch files fail due to new version of Patch in upcoming Factory
  • From: Andreas Gruenbacher <agruen@xxxxxxx>
  • Date: Fri, 3 Apr 2009 20:46:04 +0200
  • Message-id: <200904032046.04730.agruen@xxxxxxx>
On Friday, 3 April 2009 20:38:05 Raymond Wooninck wrote:
On 03/04/09 20:05, Andreas Gruenbacher wrote:
On Friday, 3 April 2009 19:54:42 Raymond Wooninck wrote:

Send along a few examples ...

Ok, here is one. I have attached two files.

Here is another oe that fails:

--- trunk.orig/src/ausearch-options.c
+++ trunk/src/ausearch-options.c
@@ -51,5 +51,6 @@ int event_ua = 0, event_ga = 0, event_se
int just_one = 0;
int event_session_id = -1;
+int line_buffered = 0;
const char *event_key = NULL;
const char *event_node = NULL;
const char *event_filename = NULL;

This hunk has fewer prefix than suffix lines, and patch assumes that this
means that the hunk must apply at the beginning of the file. The latest
upstream version of patch (2.5.9) would apply such a hunk with fuzz 1, at
which point the prefix and suffix context would have equal length again.
Previous SUSE versions of patch would try to apply this hunk anywhere, but
this can lead to weird misapplied hunks, so this was fixed.

The beginning-of-file case should be easy to detect from the line numbers. The
same problem exists at the end of the file though (as in your example), and
there, line numbers won't help.

I'm not sure how to best handle this case. Maybe go with the upstream behavior
and document the asymmetric case / add a message telling what's going wrong?

Thanks,
Andreas
--
To unsubscribe, e-mail: opensuse-factory+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse-factory+help@xxxxxxxxxxxx

< Previous Next >
References