Re: svn commit: r1498873 - in /subversion/trunk/subversion: libsvn_wc/entries.c libsvn_wc/tree_conflicts.c libsvn_wc/tree_conflicts.h libsvn_wc/upgrade.c tests/libsvn_wc/conflict-data-test.c

On Tue, Jul 2, 2013 at 2:23 PM, Stefan Sperling <stsp_at_elego.de> wrote:
> On Tue, Jul 02, 2013 at 02:09:32PM -0400, Greg Stein wrote:
>...
>> Maybe move v3 of this structure to private? Move to an API rather than
>> bare struct?
>
> I've been considering making it private. The reason I haven't yet done
> so is that I'm not sure where we'll go next with affected APIs.
>
> The most important public API affected by this is the conflict
> resolution callback. It needs this struct or needs to be changed when
> making the struct private, e.g. to expect some set of function parameters.

Well, I naturally just think of an opaque type with various accessors.
The accessors could have smaller structures that likely won't be
revised (such as svn_wc_conflict_version_t, or one returning
kind/action/reason). Or the accessors have multiple parameters (like
fetching all the associated paths).

The bare structures (entry_t, info_t, status_t, and
conflict_description_t) are the things that have historically given us
the most pain over the years.

Cheers,
-g
Received on 2013-07-02 22:24:10 CEST