Greg Stein wrote: > On Wed, Mar 10, 2010 at 17:00, <ne...@apache.org> wrote: >> ... >> +++ subversion/trunk/subversion/libsvn_wc/adm_ops.c Wed Mar 10 22:00:38 2010 >> @@ -2208,17 +2208,39 @@ svn_wc__get_pristine_contents(svn_stream >> apr_pool_t *result_pool, >> apr_pool_t *scratch_pool) >> { >> + svn_wc__db_status_t status; >> const char *text_base; >> >> - SVN_ERR(svn_wc__text_base_path(&text_base, db, local_abspath, FALSE, >> - scratch_pool)); >> - >> - if (text_base == NULL) >> + SVN_ERR(svn_wc__db_read_info(&status, >> + NULL, NULL, NULL, NULL, NULL, NULL, NULL, >> NULL, >> + NULL, NULL, NULL, NULL, NULL, NULL, NULL, >> NULL, >> + NULL, NULL, NULL, NULL, NULL, NULL, NULL, >> + db, local_abspath, scratch_pool, >> scratch_pool)); > > You can read the KIND at the same time here, and early-out for non-files. > >> + if (status == svn_wc__db_status_added) > > You should also check for obstructed_add. This could be a local copy > of a file, and the user rm'd the file and put a directory there. You > still want access to the underlying pristine.
r921848, but since now bailing on non-files, obstructed_add should not happen -- thus just added an ASSERT for that. ~Neels
signature.asc
Description: OpenPGP digital signature