[Webkit-unassigned] [Bug 25898] [Gtk] object:text-changed events should be emitted for entries and password text

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=25898





--- Comment #58 from Mario Sanchez Prada <msanchez at igalia.com>  2010-08-04 02:45:15 PST ---
(In reply to comment #55)
> Created an attachment (id=63314)
 --> (https://bugs.webkit.org/attachment.cgi?id=63314) [details]
> Single patch for fixing this bug
> 
> (In reply to comment #54)
> > Thanks for the feedback, I'll be back soon with the squashed patch
> 
> As promised, attached a single patch for this

I was doing some tests deleting huge amounts of text at once from input text fields (text areas), in order to see if my patch (specially because of the changes in WebCore/editing/DeleteSelectionCommand) degraded at some extent the performance, and the results are good, even quite surprising, I'd say:

 #chars       BEFORE      AFTER
================================
  10000        2 ms        2 ms
  20000        3 ms        1 ms
  30000        2 ms        1 ms
  40000        4 ms        1 ms
  50000        4 ms        2 ms
  60000        6 ms        1 ms
  70000        5 ms        2 ms
  80000        6 ms        2 ms
  90000        7 ms        2 ms
 100000        8 ms        2 ms
  ...          ...         ...
 250000       18 ms        5 ms
  ...          ...         ...
 500000       36 ms        9 ms
  ...          ...         ...
 750000       51 ms        14 ms
  ...          ...         ...
 900000       63 ms        17 ms
 910000       63 ms        17 ms
 920000       64 ms        16 ms
 930000       64 ms        16 ms
 940000       65 ms        17 ms
 950000       66 ms        16 ms
 960000       65 ms        17 ms
 970000       67 ms        17 ms
 980000       67 ms        16 ms
 990000       69 ms        17 ms
1000000       69 ms        17 ms


Explanation:
  - #chars: total number of characters in a text node which is going to be emptied
  - BEFORE: time involved in removing all the text in a row, WITHOUT the patch
  - AFTER: time involved in removing all the text in a row, WITH the patch

Notice that results are shown over 10000 characters being deleted because under that value the differences were practically inexistent.


As you can see there's no performance regression with the patch, but even an improvement, so I guess the patch is fine wrt performance. However, such an improvement is something I hardly can explain at this point, sorry... just let me say that I repeated this tests several times and the results were always similar for each situation. Perhaps, based on this information, someone else with more knowledge in how editing works can drop some light on this... or perhaps my tests are just crap, who knows :-)

Hope this information helps

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.