GNOME Bugzilla – Bug 519849
Firefox regression test profile should define the font and size to be used(?)
Last modified: 2008-03-19 23:55:22 UTC
The window size for the Firefox regression tests is contained the localstore.rdf for the profile used by the harness. Will indicates that the current window size is too large for his laptop. We need a window size that is reasonable and also compatible with all testing environments. Once we establish the desired window size, a number of the regression tests (at least all line nav tests and the line-related flat review test) will need to be updated.
Will: Pick a size any size. ;-) Then I'll update the tests.
(In reply to comment #1) > Will: Pick a size any size. ;-) Then I'll update the tests. > I think the size that the tests currently use on your machine is fine, as long as it is the same as or smaller than 1024x768. :-)
According to the profile, that would be 994x687. I have no idea where those numbers came from. 994x687 is indeed smaller than 1024x768. However, a couple of times you have mentioned to me that you were having a problem with the FF tests on your laptop, and I *thought* that problem was the window size. If I misunderstood you, my apologies. If you are having other problems with the FF tests, please let me know what those problems are so that I can address them. Thanks!
(In reply to comment #3) > laptop, and I *thought* that problem was the window size. If I misunderstood > you, my apologies. If you are having other problems with the FF tests, please > let me know what those problems are so that I can address them. Thanks! Hmm...maybe it's not the window size. :-( Something is very strange about the flat_review_text_by_line.py test, and it looks to me as though it is a line wrapping difference between what is happening on your machine and what is happening on mine. My first guess was that it was a window sizing problem, and perhaps maybe my Minefield wasn't honoring the width/height things in the default profile. My next guess might be the default font. I'll try taking a screenshot of what the blockquotes.html file used by the test looks like on my machine when using the test profile.
Created attachment 106434 [details] Screenshot of blockquotes.html when using the test profile
Created attachment 106438 [details] my screenshot Looks like the font is indeed different. That will mess up all of the line related tests just as good as a window that is too big. :-( Thanks for the screenshot. Now to see if we can create a profile where the fonts are present on both platforms and the lines the same....
> Looks like the font is indeed different. Correction: Looks like the spacing (kerning perhaps??) is different. Will's also looks slightly bolder (smoothing differences??), but I don't believe that is significant. I just did an experiment, opening our screenshots in Gimp and placing one atop the other as separate layers. Then I made the top layer semi-transparent so that I could line up the letters at the beginning of various lines. The font appears to be the same and the size appears to be the same. But things get progressively more and more "off" as one moves to the right. If the problem was that the fonts were different, I would think it would be relatively easy to force things to match across platforms either through the FF profile or through CSS. In light of this discovery.... I'm not sure what the answer is. :-( I'll give it some more thought but, as usual, brilliant suggestions (or at the very least, entertainingly snarky comments) welcome. ;-) (I wonder if this problem is unique to Firefox.)
(In reply to comment #7) > > Looks like the font is indeed different. When I look at the bottom loop of the lower case e's and c's, it looks like a font difference to me. On Joanie's they curve up. On mine, they lay flat. The problem may be that Solaris systems alias the serif font to a Sun font. So, in order for us to be really sure, we probably should try to use more specific font references in the Firefox profile. I'll add these two lines to my test area's prefs.js file and then post a new screenshot: user_pref("font.name.sans-serif.x-western", "Helvetica"); user_pref("font.name.serif.x-western", "Times");
Created attachment 106484 [details] Screenshot of FF running with specific fonts in test profile
Thanks Will. I'm not sure that will work for Ubuntu because Ubuntu doesn't come with Helvetica.
Created attachment 106485 [details] Same thing, but using FreeSans and FreeSerif
Created attachment 106486 [details] used FreeSans and FreeSerif Will and I played "Go Fish" with fonts. Seems that Ubuntu, Solaris, and Indiana/Developer Preview/Whatever It's Called all share FreeSans and FreeSerif. Here's my screenshot.
Looks a bit different still (spacing between lines), BUT the lines seem to terminate with the same words. Yea!! Will, do you think it's safe for me to re-run the tests using this profile and update the full suite?
(In reply to comment #13) > Looks a bit different still (spacing between lines), BUT the lines seem to > terminate with the same words. Yea!! Will, do you think it's safe for me to > re-run the tests using this profile and update the full suite? > Probably. But...trying the flat review one(s) that use blockquotes.html first might give us some sanity checking before investing too much time.
First coarse pass at GNOME 2.24 planning.
Created attachment 107639 [details] [review] trial change This specifies the fonts and has a revised flat review test. Will, when you get a chance, mind running it on your Solaris box? I get no failures when I run it. Thanks!
(In reply to comment #16) > Created an attachment (id=107639) [edit] > trial change > > This specifies the fonts and has a revised flat review test. Will, when you > get a chance, mind running it on your Solaris box? I get no failures when I > run it. Thanks! > Running on SXDE 01/08 with atk/gail/at-spi/orca from trunk and Mozilla/5.0 (X11; U; SunOS i86pc; en-US; rv:1.9b5pre) Gecko/2008031001 Minefield/3.0b5pre. Yeeeeeeeee Haaaaaaaaaaaa! bash-3.2$ ./runone.sh ../keystrokes/firefox/flat_review_text_by_line.py firefox 0 starting test application firefox -profile /tmp/FirefoxProfile ... Test 1 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review current line Test 2 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 3 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 4 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 5 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 6 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 7 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 8 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 9 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 10 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 11 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review next line Test 12 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 13 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 14 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 15 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 16 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 17 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 18 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 19 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 20 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line Test 21 of 21 SUCCEEDED: ../keystrokes/firefox/flat_review_text_by_line.py:flat review previous line SUMMARY: 21 SUCCEEDED and 0 FAILED (0 UNEXPECTED) of 21 for ../keystrokes/firefox/flat_review_text_by_line.py
Woohoo! :-) Committed to trunk along with quite a few updated tests. There are still some tests that fail -- XUL tree tables are broken at the moment, some label guess bugs I need to work on, etc. But this should hopefully get us a lot closer to tests that work across platforms. Closing as FIXED.