Editor performance issue since revision d98df4092e16 #1064
Comments
|
From pierre.raybaut on 2012-05-28T06:03:23Z For example, if I have a dozen files opened and I try to split the editor window or to create a new window, it takes more than 8 seconds to process my request. |
|
From pierre.raybaut on 2012-05-28T06:13:58Z Apparently, this is not the "load" feature that takes time because I ran the same benchmark in v2.2 and v2.1 and obtained the same results (see revision 847f8f3312da and https://code.google.com/p/spyderlib/source/detail?r=267c2476df62a8df4b4906fea6887651e7a5b103&repo=v21 ). |
|
From techtonik@gmail.com on 2012-05-28T06:18:32Z It doesn't happen on my machine with 12+ tabs, although splitting does take about 3 seconds. It may happen that editor.rehighlight() is being called twice at some point - it is rather hard to backtrace all execution paths for it. I guess the fastest way is to update to r5e0657290a9bc02b4b48e41044623896ac9d0b6a to see if the loading time differs? |
|
From pierre.raybaut on 2012-05-28T06:20:01Z I just ran the following test:
spyderlib/widgets/sourcecode/base.py
v2.1.10: it takes 6.6 seconds on my [slow] machine here -- it's very long but that's what it is |
|
From pierre.raybaut on 2012-05-28T06:21:36Z (I'm developping with a Pentium M 1.7Ghz... it's slow but I never found a more effective way to optimize code!) |
|
From pierre.raybaut on 2012-05-28T06:48:27Z As I mentionned in comment Curiously I can't reproduce the bad results of comment v2.1.10: 6.7s Revision 5e0657290a9b : 6.3s (i.e. a little bit faster) Revision 18abcf679c95 : 6.5s (i.e. no difference) Revision 28923a99b1d5 : 11.3s (i.e. almost two times slower) Revision 847f8f3312da : 11.4s (i.e. no difference) So, apparently between v2.1.10 and revision 18abcf679c95 , there is no significative change. But after revision 28923a99b1d5 , it is two times slower than before. Before trying to fix this bug, please try to do a benchmark to show this performance difference between revision 0240fa73feed and revision d98df4092e16 . Thanks! Summary: Editor performance issue since revision d98df4092e16 |
|
From pierre.raybaut on 2012-09-23T05:35:49Z Labels: -Priority-High Priority-Critical |
|
From pierre.raybaut on 2012-09-23T06:57:12Z Fixed with the following changesets: Revision 3b1ca9223a3e Revision 7676ebbd8446 Revision 856622499480 Revision c53255bbefc4 I have simply backed out changesets because this performance issue was introduced with revision d98df4092e16 , a revision which should never have been commited to the main branch, at least not without some debugging afterwards (nothing has been done for 4 months now). Something needed to be done: I'm tired of waiting a few seconds each time I'm splitting editor widget in two. Status: Fixed |
|
From techtonik@gmail.com on 2012-09-29T05:23:10Z Assigning this to myself to investigate why refactoring slowed down the performance. Status: New |
|
From ccordoba12 on 2013-02-11T08:50:19Z Anatoly, do you plan to check this for 2.2? If not, please move it to 2.3 |
|
From techtonik@gmail.com on 2013-02-11T09:56:09Z I am not using Spyder actively anymore since PySide development stalled. Reverted changesets will still be useful for refactoring highlighting code, but for now I don't have enough energy to proceed. Closing. Thanks for the care. Status: Fixed |
|
From ccordoba12 on 2013-02-11T10:13:48Z I'm very sorry to hear that. Hope to see you soon, when PySide continues development. |
|
From techtonik@gmail.com on 2013-02-11T10:26:25Z I might pop up earlier that PySide restores. =) |
|
From ccordoba12 on 2013-02-11T10:40:15Z How's so? I thought PySide had a great group of Brazilian devs behind it. |
|
From techtonik@gmail.com on 2013-02-11T10:59:53Z Nokia seized project financing after the deal with MS. |
From pierre.raybaut on 2012-05-28T08:01:47Z
Anatoly, I think you have done some things lately regarding the code editor syntax highlighting features. I don't know if it's related, but I'm experiencing a significative difference of loading time between v2.1.10 and v2.2.0dev editors.
It may be worth doing some benchmark test. It's becoming annoying: I'm currently developping on a slow machine which had no problem running Spyder v2.1 but it is struggling to run v2.2...
Any idea?
Original issue: http://code.google.com/p/spyderlib/issues/detail?id=1064
The text was updated successfully, but these errors were encountered: