Comment 1 Robert Longson 2011-10-03 20:17:07 PDT

I've just realised the patch should be fabs(patternMatrix->xx) and fabs(patternMatrix->yy) in case someone has written scale(-40) which should invert the image as well as scaling it. The patch as written would not render scale(-40).

Comment 2 Daniel Holbert [:dholbert] 2011-10-03 22:29:35 PDT

Comment on attachment 564442 [details] [diff] [review]
patch

Could you add a reftest for the situation described in comment 1?  (with inverting coming into play)  r=me with that.

Comment 3 Robert Longson 2011-10-04 00:08:43 PDT

Created attachment 564471 [details] [diff] [review]
address comments

Comment 4 Robert Longson 2011-10-04 01:40:55 PDT

pushed https://hg.mozilla.org/integration/mozilla-inbound/rev/9c7b5bdb4f04

Comment 5 Robert Longson 2011-10-04 03:01:32 PDT

And backed out as the reftest fails on Linux.

Comment 6 Robert Longson 2011-10-04 03:12:11 PDT

Created attachment 564502 [details] [diff] [review]
try to fix reftest failures

Comment 7 Mozilla RelEng Bot 2011-10-04 06:51:05 PDT

Try run for 696085573f75 is complete.
Detailed breakdown of the results available here:
    https://tbpl.mozilla.org/?tree=Try&rev=696085573f75
Results (out of 9 total builds):
    success: 4
    warnings: 5
Builds available at http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/longsonr@gmail.com-696085573f75

Comment 8 Mozilla RelEng Bot 2011-10-04 08:21:37 PDT

Try run for 6b114ae3b4c4 is complete.
Detailed breakdown of the results available here:
    https://tbpl.mozilla.org/?tree=Try&rev=6b114ae3b4c4
Results (out of 9 total builds):
    success: 9
Builds available at http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/longsonr@gmail.com-6b114ae3b4c4

Comment 9 Robert Longson 2011-10-04 08:24:07 PDT

fixed reftest and landed as https://hg.mozilla.org/integration/mozilla-inbound/rev/e05e1b35ebed

Comment 10 Robert Longson 2011-10-04 08:43:54 PDT

antialiasing sucks when you're writing reftests :-(

Comment 11 :Ehsan Akhgari 2011-10-04 13:47:41 PDT

Landing: https://hg.mozilla.org/mozilla-central/rev/9c7b5bdb4f04
Backout: https://hg.mozilla.org/mozilla-central/rev/8dac1be4c4c1

Comment 12 Marco Bonardo [::mak] 2011-10-05 05:03:53 PDT

https://hg.mozilla.org/mozilla-central/rev/e05e1b35ebed

Comment 13 Daniel Holbert [:dholbert] 2011-10-25 13:20:18 PDT

Looking back on this, I think we should have used patternMatrix->GetScaleFactors() to get the scale factors, instead of using patternMatrix->xx / patternMatrix->yy.  (assuming patternMatrix is non-singular)  Otherwise, I think rotations / skews could mess us up here.

Robert: am I missing something, or does that make sense?  I'm happy to file a followup on that, just wanted to sanity-check here first.

Comment 14 Robert Longson 2011-10-25 14:26:49 PDT

patternMatrix->ScaleFactors(true)

Comment 15 Daniel Holbert [:dholbert] 2011-10-25 14:49:28 PDT

Yup, sorry for getting the wrong name there.

Actually I'm getting less sure about ScaleFactors, though, from looking at the impl... I'm not sure we'd always be able to pass the right value of "xMajor". (the existing code also only has 2 calls to ScaleFactors, with 'true' in both places, but I don't think that's necessarily right in this spot)

So, I think I rescind comment 13 (at least partly)... sorry for my confusion on this.