Re: CHANGEs File for Subversion 1.6 - Questions - #3349

Nope, no API change. Just a bugfix to a codepath folks aren't supposed to
be executing anyway.

Karl Heinz Marbaise wrote:
> Hi,
>
> I found the following:
>
> In theory, BDB backend could allow `changes' table out-of-sync with
> actual changes (#3349)
>
> A comment in the #3349 issue:
> -------------
> NOTE: The API usage demonstrated in the regression test is not valid
> per the instructions on the APIs given in svn_fs.h, which say that you
> must send windows including the NULL terminating window through the
> returned handlers before making other changes to the transaction. FSFS
> seems to enforce this, but BDB does not. In BDB, this code path will
> result in node revisions that still have dangling
> representations-in-progress, but that was true before my changes, too.
> -------------
>
> Based on the diff
> svn diff -r 34896:34897 http://svn.collab.net/repos/svn
> i don't see a real change in an API ?
>
> Or did i misunderstand a thing ?
>
> Kind regards
> Karl Heinz Marbaise

-- 
C. Michael Pilato <cmpilato_at_collab.net>
CollabNet   <>   www.collab.net   <>   Distributed Development On Demand
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=1334503