Ben emailed me directly, so I'll attach the report for him - I'll try
and get someone to investigate if we've got some infrastructure issues
preventing his use of the tracker.

-- 

1. Summarise the issue (what were doing, what went wrong?)

> darcs diff test/validate/testSupport.h

darcs failed:  replace patch to ./src/snc/gen_ss_code.c couldn't apply.

Note the file it complains about is not teh one I gave on the command line.
(It's in the same repo, though).


2. What behaviour were you expecting instead?

This one:

> darcs-2.8.0 diff test/validate/testSupport.h
--- old-work/test/validate/testSupport.h        2012-10-04 14:36:35.0000000=
00
+0200
+++ new-work/test/validate/testSupport.h        2012-10-04 14:36:35.0000000=
00
+0200
@@ -9,12 +9,15 @@

 #include "epicsThread.h"
 #include "epicsMutex.h"
+#include "epicsEvent.h"
 #include "epicsUnitTest.h"

 #ifdef __cplusplus
 extern "C" {
 #endif

+extern epicsEventId this_test_done;
+
 void run_seq_test(seqProgram *seqProg, int adapt_priority);
 void seq_test_init(int num_tests);
 void seq_test_done(void);


Yes, this means it worked in darcs-2.8.0.


3. What darcs version are you using?

I am using the latest and greatest from the screned repo (2.9.5 + 33 or so
patches).

4. What operating system are you running?

Ubuntu 12.04 (64 bit)

Thanks, Owen, for helping out. I am using Firefox now (before used
Konqueror), maybe that works better.

I wanted to add that this happened after I unrecorded a large patch.
Also, to aid debugging, I can give you a tar ball of the repo where this
happened.

Hi Ben,

Yes, a tarball would be most useful, to allow us to distil a minimal
example of this behaviour.

Cheers,
Owen.

Here we go.

Attachments

• darcs-issue2251.tar.gz

This issue looks like it has as much to do with "replace" as "diff". 
Updating title accordingly.

BTW, this bug is still open. I just reproduced the problem with
darcs-2.10.3.