On 10 May 2010 16:52, Piotr Piastucki <[email protected]> wrote: > Hi, > > Thanks a lot for all the comments! I will try to improve the patch based on > your input. > However, I am going to keep merge_2_files() as it is because: > 1) I pass the original linediffer to Merger so the changes applied should be > exactly the same as the changes shown to the user > 2) it is about 10 times faster than calling replace_chunk in a loop on my > machine > I am attaching a patch that can be used to compare performance of both > approaches and two test files: > - run meld meldapp_1.py meldapp_1.py meldapp_2.py > - right click middle pane, select Pull all non-conflicting from right (see > console) > - right click middle pane again, select Pull all non-conflicting from left > (see console)
I tried this, but I don't get any console output, and the patch has no reference to replace_chunk. Since I couldn't check your approach, I'm just guessing here, but... I suspect that the only reason that there is any difference in speed is that you forgot to suspend the appropriate signal emission before starting the replace_chunk loop. If that's the case, then we're doing a whole update-diff, update-highlighting, etc. cycle for each chunk, and that *would* be slow. cheers, Kai _______________________________________________ meld-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/meld-list
