[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
[nebula-dev] [CompositeTable] very bad scalability
|
Hi Dave
I just implemented a first version of a editpart that uses the
compositeTable as table widget and my first 'real-life' results are
unfortunately very bad. My table has to show up to 30 columns and up to 30
rows (900 widgets!! autsch! I didn't think of that, unfortunately...).
There are 2 issues that have the potential to stop my efforts:
1.) the whole rcp app dies with SWTError: no more handles
I'll analyze this by using sleak but I suppose that this will not help a
lot as i'll reach the limit very soon by opening a second editor-tab.
This lets me think that it's not possible to display cells by individual
controls. On the other hand traversal relies on these controls. Any ideas
here?
2.) layout (I use native header) is terribly slow. The table takes up to 2
minutes to display and resize.
I probably could solve that performance issue by improving the resizing or
using another header. any ideas?
I'd very very appreciate any hints that could help me to overcome this issue!
Thanks in advance & Regards
André
> Hi Tom
>
> thanks a lot for the precise feedback! Good to hear the bug is gone on
> OSX, too :-)
>
> Regards
> André
>
>> Hi,
>>
>> Ok. I see you created a patch only for the single file and so the
>> module path was not included in your patch and I had to selected the
>> file itself to apply the patch.
>>
>> Everything working smoothly on MacOSX, the only very small issue I see
>> is a white small white line in the Table row directly above the text
>> which is coming from the text-elements border I guess and you can't do
>> anything against it.
>>
>> Another small problem is that the row is only highlighted when you
>> click
>> in an area with a control e.g. the text control doesn't fill the
>> complete hight on MacOSX and when clicking in the area above the text
>> widget the row is not colored. Maybe the mouse listener should also
>> listen to clicks outside the control area.
>>
>> Nevertheless nice work.
>>
>> Tom
>>
>> André Dietisheim schrieb:
>>> Hi Tom,
>>>
>>> yes, that's what I did (checked out the latest cvs, inserted my
>>> changes, created a patch..
>>> I just checked out the latest cvs, applied my patch and did not have
>>> any troubles. Did you apply the 2nd patch
>>> (not the first one in the list. I included this to show only what
>>> changes I inserted. The current changes rely another patch I did in
>>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=173558 and I therefore
>>> put a second patch, that includes both changes (patch to bug 173558 &
>>> the current bug)
>>>
>>> Greetings
>>> André
>>>
>>>> Hi Andre,
>>>>
>>>> I'm having problems applying your patch against the latest checkout
>>>> from nebula. I get merging conflicts all over the
>>>> InternalCompositeTable. Does the patch apply for you cleanly on a
>>>> checkout from cvs?
>>>>
>>>> Tom
>>>>
>>>> André Dietisheim schrieb:
>>>>> Hi tom
>>>>>
>>>>> I just checked my patch and a demo-snippet into bugzilla:
>>>>>
>>>>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=176466
>>>>>
>>>>> I couldn't add you to cc's as your list-email was not accepted.
>>>>> Thanks for the testing!!!
>>>>>
>>>>> Regards
>>>>> André
>>>>>
>>>>>
>>>>>> yeah. since yesterday ;-) Should I check out from CVS and do what
>>>>>> afterwards?
>>>>>>
>>>>>> Tom
>>>>>>
>>>>>> David J. Orme schrieb:
>>>>>>> This sounds correct. Wow, you really have deeply understood how
>>>>>>> CompositeTable works. Please file a bug and I'll test on Win32.
>>>>>>> Is there anyone out there with a Mac?
>>>>>>>
>>>>>>>
>>>>>>> Regards,
>>>>>>>
>>>>>>> Dave Orme
>>>>>>>
>>>>>>> -------
>>>>>>> thanks a lot for the positive feedback! Always nice to hear :-)
>>>>>>>
>>>>>>> I have a more pickier problem I just managed to solve for GTK+
>>>>>>> (I'll test things on Win32 asap) and I would appreciate a lot
>>>>>>> your ideas on this issue & my possible fix as this one gets
>>>>>>> deeper into compositeTable and SWT. But take your time! I really
>>>>>>> do not want to put more stress on you than you certainly already
>>>>>>> have :-(
>>>>>>>
>>>>>>> The issue can be experienced when you scroll the table by
>>>>>>> mouse-wheel very fast and you scroll the current row outside the
>>>>>>> viewport. It gets more frequent the more widgets are visible (the
>>>>>>> more complex a row is). The issue is that an arbitrary
>>>>>>> scrolled-in row (after the current row was scrolled out) gets
>>>>>>> focused and gets 'current row' - I clearly tracked it down to
>>>>>>> focusGained() in InternalComposite. I believe that the newly
>>>>>>> focused row is
>>>>>>> actually the one that got scrolled out and was repositioned to
>>>>>>> the top (and refreshed with new model-content). I believe that
>>>>>>> this is due to deferredSetFocus() that gets delayed a lot and
>>>>>>> gets
>>>>>>> executed only when the row was repositioned and scrolled in (on
>>>>>>> the top) again. My possible fix (tested for GTK) is to keep track
>>>>>>> of the control to focus delayedly (asyncExec) by setting the
>>>>>>> instance variable this.toFocus. Furthermore to clear that
>>>>>>> instance variable on row departure & fire a focusOut-event on the
>>>>>>> other hand. I try to prevent any deferredSetFocus() that would
>>>>>>> focus any control that was departed and is visible again, when
>>>>>>> asyncExec finally gets executed.
>>>>>>>
>>>>>>>
>>>>>>> ------------------------------------------------------------------------
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> nebula-dev mailing list
>>>>>>> nebula-dev@xxxxxxxxxxx
>>>>>>> https://dev.eclipse.org/mailman/listinfo/nebula-dev
>>>>>> _______________________________________________
>>>>>> nebula-dev mailing list
>>>>>> nebula-dev@xxxxxxxxxxx
>>>>>> https://dev.eclipse.org/mailman/listinfo/nebula-dev
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>>
>>
>
>
> --
> André Dietisheim
> Stv-Bereichsleiter Products
>
> Puzzle ITC GmbH
> Eigerplatz 4
> CH-3007 Bern
> Telefon +41 31 370 22 00
> Mobile +41 76 423 03 02
> Fax +41 31 370 22 01
>
> Puzzle ist Mitglied der ODF Alliance:
> <http://www.puzzle.ch/odfalliance/>
>
>
> _______________________________________________
> nebula-dev mailing list
> nebula-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/nebula-dev
>
>
> !DSPAM:45ed895d210036786915523!
--
André Dietisheim
Stv-Bereichsleiter Products
Puzzle ITC GmbH
Eigerplatz 4
CH-3007 Bern
Telefon +41 31 370 22 00
Mobile +41 76 423 03 02
Fax +41 31 370 22 01
Puzzle ist Mitglied der ODF Alliance:
<http://www.puzzle.ch/odfalliance/>
