Commit Graph

16 Commits

Author SHA1 Message Date
Aaron J. Seigo
c43ef1bcb8 make sure that we don't get hung up with old jobs that aren't enqueued anymore by checking for that case more often
svn path=/trunk/KDE/kdelibs/; revision=1053930
2009-11-25 01:54:44 +00:00
Aaron J. Seigo
5b61dd019e private headers
svn path=/trunk/KDE/kdelibs/; revision=1053919
2009-11-25 01:08:35 +00:00
Andreas Hartmetz
d0ecf4ae88 This hopefully fixes a double delete bug. I don't see what would prevent both
RunnerManager::jobDone() and DelayedJobCleaner::jobDone() from being invoked
if the JobDone() signal is emitted just at the right (wrong) time.
Both jobDone() slots delete (as in delete operator) the job.
CCMAIL:aseigo@kde.org

svn path=/trunk/KDE/kdelibs/; revision=1033035
2009-10-09 09:33:20 +00:00
Aaron J. Seigo
b32e268e15 meh; can't move these files into a subdir as-is; need to think of a more clever strategy
svn path=/trunk/KDE/kdelibs/; revision=1019657
2009-09-04 08:58:34 +00:00
Aaron J. Seigo
b711b96ee0 runner infra into a subdir
svn path=/trunk/KDE/kdelibs/; revision=1019561
2009-09-03 21:21:27 +00:00
Jacopo De Simoi
55169a2fda Style correction - thanks Dan
svn path=/trunk/KDE/kdelibs/; revision=957856
2009-04-23 02:13:54 +00:00
Jacopo De Simoi
745eeeb79a Optimization: don't bother running the performMatch if the context is already invalid.
svn path=/trunk/KDE/kdelibs/; revision=957814
2009-04-22 22:53:48 +00:00
Aaron J. Seigo
a96782cd4b if we have no more jobs, delete ourselves later
svn path=/trunk/KDE/kdelibs/; revision=938755
2009-03-12 22:02:31 +00:00
Aaron J. Seigo
8c30031635 * let the weaver own the jobs, so we can cean up successfully without crashing when the manager is deleted
* listen to jobDone's in the clean up class

svn path=/trunk/KDE/kdelibs/; revision=938752
2009-03-12 22:01:22 +00:00
Aaron J. Seigo
3ecfa9f6d5 if the RunnerManager is deleted but jobs are still processing, create a helper object to wait it out and clean them up when possible
svn path=/trunk/KDE/kdelibs/; revision=938728
2009-03-12 20:25:52 +00:00
Aaron J. Seigo
592ddc4e6e * emit a signal when our jobs are done but we have no matches
* use QSets instead of QLists where we are doing lots of random access and no sequential
* get rid of the stale marking as all items in the oldMatches list are, by definition, stale

svn path=/trunk/KDE/kdelibs/; revision=938708
2009-03-12 18:53:22 +00:00
Jacopo De Simoi
25a80b322c Create local copy of runnerContext in the runnerjob ctor
svn path=/trunk/KDE/kdelibs/; revision=938690
2009-03-12 17:25:13 +00:00
Jacopo De Simoi
0d12d6b3b8 Uff, I guess we will live with duplicates for some more time. My previous commit breaks krunner as it *only* runs stuff which is autocompleted. Will backrevert -
svn path=/trunk/KDE/kdelibs/; revision=935428
2009-03-05 06:02:11 +00:00
Jacopo De Simoi
8aa4045953 "...and they lived happily ever after with no more duplicate matches =)"
A local copy of the context is now created in the FindMatchesJob ctor, not anymore in AbstractRunner, this prevents old queries to be run with a valid dptr 

svn path=/trunk/KDE/kdelibs/; revision=935342
2009-03-04 21:03:33 +00:00
Jacopo De Simoi
235f81fa36 Added a stale flag to searchJobs; less pollution from old queries; fixed a misplaced brace in previous commit that made krunner unable to run any command :(
svn path=/trunk/KDE/kdelibs/; revision=934467
2009-03-03 08:02:13 +00:00
Ryan P. Bitanga
2da5f905d3 Change the queue policy for slow speed runners. They will now
wait 400ms before being executed by KRunner

svn path=/trunk/KDE/kdelibs/; revision=924670
2009-02-11 11:16:09 +00:00