On Friday 19 March 2010 18:40:26 Martin Sucha wrote:
On Friday 19 March 2010 17:38:15 ~suv wrote:
On 19/3/10 17:06, Joshua A. Andler wrote:
On Fri, 2010-03-19 at 16:48 +0100, ~suv wrote:
Is it possible that this patch introduced a big performance loss when grouping a larger selection of objects? I just stumbled upon this when clipping ~1800 small objects (created with the spray tool) with a rectangle using the new feature to group the selection automatically before applying the clippath to the group: to my surprise this took very long (~45 seconds) whereas grouping the same selection happens almost instantaneously in Inkscape 0.47 (~2 seconds).
This appears to be accurate... would you mind filing a bug report and assigning to Martin Sucha?
Unfortunately, there is no way we can release with this big of a performance hit for such a common task. So if it doesn't get fixed before Feature Freeze I will have to revert it.
Bug #542004 https://bugs.launchpad.net/inkscape/+bug/542004 “performance regression when grouping a large selection set”
Interesting. I changed the code so I could reuse grouping from clipping function, but the logic behind grouping should be the same. Looking at the patch, I suspect the call to selection->clear() should be before the whole grouping logic (it was before) and I accidentally put that after. Will check if this causes the regression.
Yes, this causes the regression. I fixed this for grouping and now I'll look at the clipping code as it needs to be adjusted as well. Will send a patch when done.
Regards, Martin Sucha