vte for FC3/FC4 testing

Kjartan Maraas kmaraas at broadpark.no
Tue Mar 15 17:09:30 UTC 2005


man, 14,.03.2005 kl. 11.16 -1000, skrev Warren Togami:
> Warren Togami wrote:
> > Per Bjornsson wrote:
> > 
> >> On Tue, 2005-03-08 at 10:20 -1000, Warren Togami wrote:
> >>
> >>> I just pushed this package to FC4.  Not sure if it will make 
> >>> FC4test1, but we'll see.  I didn't get a chance to try your 0.11.13 
> >>> yet.  Maybe we can consider it for rawhide sometime after test1.
> >>
> >>
> >>
> >> Any chance to get an FC3 update as well? Not only is the vte in FC3
> >> slow, it also has really annoying rendering bugs (easily seen e.g. in
> >> nano, https://bugzilla.redhat.com/beta/show_bug.cgi?id=127972 ).
> >>
> >> Until today I was using a rawhide version I rebuilt some time ago (per
> >> comment in that bug report); I've given your new version (0.11.12-0.FC3)
> >> a spin and I haven't hit any problems yet. In fact, I think it fixes a
> >> couple of residual problems that I had (but I can't tell for sure yet
> >> since I couldn't figure out exactly how to reproduce them, it was mainly
> >> occasional screen corruption when scrolling a lot). It would be
> >> wonderful if you could shove this version into FC3 updates-testing ASAP
> >> and see if others have the same experience.
> >>
> > 
> > It seems that this 0.11.12 based package is much better than FC3, but I 
> > am more concerned about long-term runtime problems.  Keep testing it, 
> > and stress it hard, try to make it break.  If nobody complains for a 
> > while then we'll push this to FC3 updates.
> > 
> 
> Just as I had feared, long runtime with heavy activity shows signs of a 
> memory leak.  I don't have time to debug this anytime soon.
> 
Could you describe the activity pattern to make it easier to reproduce?
I've run gnome-terminal under valgrind with no signs of leaks, but not
for a long time.

> So it would be appropriate to push this to FC3 updates only if the 
> current FC3 vte also has memory leaks.  It probably does.  Can somebody 
> confirm?
> 
Also, I reverted the questionable fedora patch in CVS and added the
latest patch for the memory consumption problems. It now shares the
matching stuff and struct _vte_termcap between terminals and should use
a lot less memory with multiple tabs. I think one of the previous
versions of this patch had leaks that were fixed later, so that might be
worth checking out.

Cheers
Kjartan




More information about the fedora-devel-list mailing list