Hi Robert,

You did a lot of hard work again! I like the direction this is going.

1) I've made some progress regarding the slowdown problem with threads:

http://bugzilla.mkgnu.net/show_bug.cgi?id=264#c29

2) I can't figure out why we are getting a corrupted response from
$socket->peeraddr():

http://bugzilla.mkgnu.net/show_bug.cgi?id=264#c30
http://bugzilla.mkgnu.net/show_bug.cgi?id=264#c33

3) I did not move the $bugtracker variable so that it isn't global. 

First, invoking get_bugtracker() when a connection is processed is not
desirable, since this would let a user believe that the daemon started
correctly and be faced with a problem later on when committing (e.g. the
version of the bugtracker is not supported).

Second, I don't understand why this would give us "threading issues". It
seems that all thread data (even global data) would be locally copied:

http://perldoc.perl.org/perlthrtut.html#Shared-And-Unshared-Data

Does this mean that $request could have remained a global variable too ?
Just want to make sure I understand what's going on.

4) What's the harm if we do write a pid file for Windows ?

> Since we now have a threading solution I also had a look at changing the
> TestDirector interface to call the Win32::OLE interface directly (rather
> than exec another process).  Guess what, found out that Win32::OLE isn't
> thread safe either!

That's nasty.

> If you decide to merge this into the current version, is it possible for
> you to send me an updated release?

I'll be emailing you a private release for Windows. Let me know how it
works for you. Also, did threading really fix the crash problem you've
been experiencing ? Or do you need more time to test ? I'm curious -- we
haven't really figured out the reason we are crashing.

Thanks A LOT for all this work.

Kristis

_______________________________________________
scmbug-users mailing list
[email protected]
http://lists.mkgnu.net/cgi-bin/mailman/listinfo/scmbug-users

Reply via email to