There seems to be an issue with `darcs apply' when the long comment of a
`darcs send'^H^Ht' patch contains special characters.  

I am using 

        JustTesting chak 17 (.../src/c2hs): darcs --version
        1.0.4 (unknown)

When attempting to apply the attached patch, I get the appended error
message.  The patch itself must be ok, as `darcs pull'ing the same patch
works without problems.

Manuel

PS: I had a quick look at the bug tracker, but didn't see anything 
    similar yet.

-=- error message follows -=-

darcs: Error parsing patchinfo:
""
""
"[Only match in `checkForOneCUName' if there are no indirections"
"Jelmer Vernooij <jelmer@samba.org>**20051205172623"
""
" Without this patch, the following code:"
""
" #c"
" struct foostruct { char x; };"
" void bar2(struct foostruct **a);"
" #endc"
""
" {#pointer *foostruct as Foo newtype#}"
" test2 = {#call bar2#}"
""
" would give"
""
" foreign import ccall safe \"test-pp.h bar2\""
"   bar2 :: ((Foo) -> (IO ()))"
""
" with this patch, it generates:"
""
" foreign import ccall safe \"test-pp.h bar2\""
"   bar2 :: ((Pt"

Attachments

• unnamed

The patch seems to apply if you remove the part between the scissors

===============================8<===============================
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)

iD8DBQFDlHjjPa9Uoh7vUnYRAqWcAJ4yRohtKKIIS1hAK1iXyyjz6rIpcgCeNCmM
YL6J8iHouVKhdhcjgGhfFb0=
=rHPj
-----END PGP SIGNATURE-----

--=_--

.
===============================8<===============================

Is this related to issue26 (MIME stuff) at all?

I forgot to mention, that stuff immediately follows the patch bundle hash, so it
could just be our bundle-parsing code not expecting trailing characters.  No
comment on whether or not it is reasonable to do so.

I've (belatedly) found this bug.  We were just calling error instead of
returning Nothing in the PatchInfo parser.  Duh, we're foolish.  A simple grep
for the error message showed the problem!