Julian Foad wrote: > On trunk, it looks like the following doc fixes are required to match the > code > changes made in r1570642. [...]
And a bit more: [[[ Index: subversion/libsvn_fs_fs/fs.h =================================================================== --- subversion/libsvn_fs_fs/fs.h (revision 1586278) +++ subversion/libsvn_fs_fs/fs.h (working copy) @@ -339,7 +339,7 @@ typedef struct fs_fs_data_t svn_cache__t *dir_cache; /* Fulltext cache; currently only used with memcached. Maps from - rep key (revision/offset) to svn_string_t. */ + rep key (revision/offset) to svn_stringbuf_t. */ svn_cache__t *fulltext_cache; /* Access object to the atomics namespace used by revprop caching. Index: subversion/libsvn_fs_x/fs.h =================================================================== --- subversion/libsvn_fs_x/fs.h (revision 1586278) +++ subversion/libsvn_fs_x/fs.h (working copy) @@ -276,7 +276,7 @@ typedef struct fs_x_data_t svn_cache__t *dir_cache; /* Fulltext cache; currently only used with memcached. Maps from - rep key (revision/offset) to svn_string_t. */ + rep key (revision/offset) to svn_stringbuf_t. */ svn_cache__t *fulltext_cache; /* Access object to the atomics namespace used by revprop caching. ]]] I should have said, I reviewed this nomination by static code examination, searching for all uses of caches where no serializer and deserialize are provided, and checking that they correctly use svn_stringbuf_t values. It all looks good, apart from the documentation updates. I can vote +1 for this once I work out the magic incantation to test it. I'd also like to see these doc fixes in trunk and preferably included in the backport too. I can commit them if one of you can give the nod for reassurance, although I'm pretty sure they're right. - Julian
