Not really related to this bug, but preparatory cleanup.

I missed this during review: nsPresContext::AppUnitsPerCSSPixel() is a static method and always returns 60, we don't need to store the result.

Comment on attachment 8487132 [details] [diff] [review] part 1: mUserSpaceBounds shouldn't change its meaning temporarily Review of attachment 8487132 [details] [diff] [review]: ----------------------------------------------------------------- Looks better! ::: layout/svg/nsSVGFilterInstance.cpp @@ +104,5 @@ > // The filter region is way too big if there is float -> int overflow. > return NS_ERROR_FAILURE; > } > > // Undo the temporary transformation of the user space bounds. This comment mentions the "temporary transformation" we're not doing anymore. I would either remove it, change it to "Set the user space bounds based on the rounded-out filter space bounds.", or something else.

Comment on attachment 8487134 [details] [diff] [review] part 2: Only construct FilterDescription once Review of attachment 8487134 [details] [diff] [review]: ----------------------------------------------------------------- Nice!

Actually, let's cut this bug here, I'll file a new bug for the rest. Thanks for the quick reviews!

https://hg.mozilla.org/integration/mozilla-inbound/rev/7de55e234f2d https://hg.mozilla.org/integration/mozilla-inbound/rev/84e7c9da710e https://hg.mozilla.org/integration/mozilla-inbound/rev/0703f395ca30

https://hg.mozilla.org/mozilla-central/rev/7de55e234f2d https://hg.mozilla.org/mozilla-central/rev/84e7c9da710e https://hg.mozilla.org/mozilla-central/rev/0703f395ca30