Merge tracking implementation issue - part 2

[Konstantin Kivi <kk...@yandex.ru>]

Some time ago I complained about mergeinfo property spreading
across files, spoiling log messages.  (http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=1594371&orderBy=createDate&orderType=desc)


It's pity that there is no traces
of this issue being addressed somehow.

Recently we hit another problem.
Our project has some 2000 files, 700 a  single directory and other split up
somehow.
Recently I created a branch of a trunk and made 2 commints on it,
touching not more than six files with total diffs amounting to couple of
screens.   After that I was not able to merge changes back to trunk,
even one by one. Svn quickly started to consume 100% CPU and it's
memory footprint grew up to 1.3G. I never managed to finish this operations
(tried both ubuntu 9.04 supplied client and 1.6.2 built from source).
My colleague  reported that he managed to complete this operation
 on slackware , but with his computer also starting to be very unresponsive.

Adding --ignore-ancestry switch fixed the problem.

I kindly ask developers of svn to rethink their design of svn mergeinfo
feature. As I wrote previously svnmerge utility approuch might be more
succesful.

Update:
The problem was caused by huge amount of file level mergeinfo. After
it had been cleared from them regular merge succeded.

Best regards, Konstantin Kivi

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2359598

To unsubscribe from this discussion, e-mail: [users-unsubscribe@subversion.tigris.org].