I need the token mappings in upgrade.c and wc_db.c in order to
populate the CONFLICT_VICTIM table.
There are some functions in upgrade.c to do the mapping, but the
values are slightly "off". I see no reason for differentiating the
values from what we store in the skels, so I want to use the existing
value mappings. Further, that skel storage/format will move to
old-and-busted.c. It is quite probable that tree_conflicts.[ch] will
go away entirely.
Cheers,
-g
On Mon, Sep 14, 2009 at 10:05, Julian Foad <julianfoad_at_btopenworld.com> wrote:
> On Sun, 2009-09-13, Greg Stein wrote:
>> Agreed.
>>
>> I had some reticence against this approach due to concerns around
>> cross-DLL linking of data variables, but Bert thinks we should be
>> okay. Potentially a few build tweaks to support the data bits, but
>> then we'll be good.
>>
>> I'm going to go ahead and commit the basic bits, and then we can grow
>> usage from there.
>
> Great. Thanks. That was r39273 and follow-ups. You rightly omitted the
> parts where in my patch I had used incorrect variants of the words,
> causing some test failures.
>
> Later, in r39282 you made the tree conflict description tokens
> semi-public:
>
> [[[
> A couple random things: constify some conflict stuff, and export
> conflict token mappings.
>
> * subversion/libsvn_wc/tree_conflicts.h:
> (svn_wc__operation_map, svn_wc__conflict_action_map,
> svn_wc__conflict_reason_map): exported for broader use
>
> [...]
> ]]]
>
> I think we should revert that part of that change, because these strings
> (which are part of the WC metadata format) are logically unrelated to
> any strings used for presenting the corresponding data to the user and
> should not need to be known outside that file.
>
> Would that be OK (patch attached) or did you have a different reason in
> mind?
>
> - Julian
>
>
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2394650
Received on 2009-09-14 16:52:01 CEST