After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 133385 - Performance is slow when using GOK with 'File types and Programs' application
Performance is slow when using GOK with 'File types and Programs' application
Status: RESOLVED FIXED
Product: gok
Classification: Deprecated
Component: performance
unspecified
Other Linux
: Normal normal
: ---
Assigned To: David Bolter
David Bolter
Depends on:
Blocks: 118134
 
 
Reported: 2004-02-04 11:51 UTC by Frances Keenan
Modified: 2004-12-22 21:47 UTC
See Also:
GNOME target: ---
GNOME version: ---



Description Frances Keenan 2004-02-04 11:51:24 UTC
Using GOK from CVS HEAD 2nd Feb 2004

-launch GOK
-launch 'File types and Programs'
 Applications->Desktop Preferences->Advanced->File types and Programs
-Notice GOK appears to freeze. After about 30-40 seconds, GOK comes 
 alive again. Everytime you activate or click any feature on the 
 application, GOK's performance is really slow.
-Movement should be quick and smooth.

running GOK on a machine :
Intel Pentium 4 CPU 2.00GHz
Comment 1 bill.haneman 2004-02-04 14:19:05 UTC
Fixing this may not be practical: File Types and Programs is a very
complex hierarchy.  Note that at-poke's 'Expand' button takes the same
amount of time to execute.

One thing I notice is that File Types and Programs has lots of
un-expanded items in the list; if GOK did not descend into lists that
weren't expanded, we'd avoid the time penalty but we might well
introduce new problems (since we'd then have to refresh our GUI
hierarchy on every 'expanded' state change!).


Comment 2 bill.haneman 2004-02-04 16:09:12 UTC
I put a workaround for this in cvs (which was fairly straightforward),
which eliminates the symptom in the case of this particular app. 
Padraig is looking at the gail implementation of that table to see if
behavior is correct.

We should probably file a separate bug for GOK's treatment of tables
in general, since the current behavior is not sufficient for the
general case.

Also, though it should be a separate bug, I note that we cannot at
present expand expandable table element via GOK.
Comment 3 korn 2004-02-04 19:38:37 UTC
Just out of curiosity, how does Gnopernicus flat review perform with 
this application?  This would be a good test...
Comment 4 bill.haneman 2004-02-05 12:44:25 UTC
Peter: don't know the answer, but bear in mind a significant
difference: gnopernicus flat review should use bounds checking of
tables to determine the visible elements, whereas GOK doesn't require
that actionable elements be visible in all cases.  Probably a
reasonable requirement in tables however, and the right long-term fix
for GOK IMO (i.e. check for MANAGES_DESCENDANTS, if it's present and
object is AccessibleTable, find out bounding row/col range for
visibility and restrict the search to that area).  However, that also
means that GOK would have to rebuild its component list whenever
visibility of a table changes (i.e. on scrolling), which could
introduce some even more serious performance issues.
Comment 5 korn 2004-02-05 23:56:27 UTC
Bill - yes, I am aware of the difference between the design of 
Gnopernicus with respect to flat review, clipping, and tables.  In 
fact, it is precisely because of this that I think this is an 
interesting test case, which is why I am suggesting that testing poke 
around in here with Gnopernicus.  It will be interesting to see how 
well those algorythms work in this case.