Re: 1.1rc1 performance regression in 'svn status' (and also 1.1rc2)

Peter N. Lundblad wrote:
> This was discussed earlier on this thread. If a pools parent is in a
> different thread, this will not work. You can't use a xlate handle in
> different threads at the same time, AFAICK.

maybe you can store store the threadID together with the xlate handle,
and only use the pool's xlate handle if the threadIDs match?
(assuming the threadid doesnt get re-used).

-

better solution would be:

create 2 new functions to initialize / exit
the subversion api. _if_ the init function is called,
the xlate handle can be stored in some thread local
storage. if it is _not_ called, everything remains like
it is: slow :)
afaik you are allowed to introduce _new_ functions
to the public api?

-

third option (better than to force everyone to a slow svn)

allow some config or environment variable to _disable_
the translations. svn would be plain english, but at least
it would be fast(er) again.
For everyone using an english windows, US locale, untranslated svn -
is there any reason why an xlate handle is created at all?
(currently I compile svn myself and withouth language support)

:-)
c.a.t.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Aug 13 11:42:26 2004