[packagekit] Segfault

Richard Hughes hughsient at gmail.com
Mon Nov 5 23:34:32 PST 2007

On Tue, 2007-11-06 at 07:15 +0000, Grzegorz Dąbrowski wrote:
> On Mon, 2007-11-05 at 22:35 +0000, Richard Hughes wrote:
> > Ahh, there it is. The problem was that the cancel not being implemented
> > was causing the first transaction to error out and be marked as
> > finished, when it was still running it's thread. Therefor you got two
> > finished signals. As cancel is not something that uses a fresh tid, we
> > shouldn't try to finish it on error. I've hopefully fixed this up in
> > commit a38c6745e0d4c1348b68ad6b4c2d04e757823e43. Can you confirm the
> > warning is now gone?
> Yes, it's gone. Thanks!

Super cool.

> > Also, how did you cancel the transaction when cancel wasn't available?
> > Did you click something in the GUI or do it in the code?
> In the GUI I was moving up/down (fast) on packages list (search result).

Ahh gotcha. Is there a reason you can't use Cancel? Can box cope with
command-C when searching? If it can then we should probably hook up
Cancel to a thread cancel operation.

It's probably also a good idea to not call Cancel in the GUI tools if we
know we can't support it. I'll do this now.



More information about the PackageKit mailing list