Behavior in Chrome 19:
Renders correctly. Page contents IS NOT shifted to the left.
What steps will reproduce the problem?
1. Create EVE Online account or use an existing one.
2. Log in to EVE Gate (http://evegate.com).
3. Select a character.
4. Go to Profile > Character Sheet.
Attachments:
Screenshot-EVE Online | EVE Gate - Character Sheet - Google Chrome.png 213 KB
I don't want to give you password from my personal account(s) for obvious
reasons. But since you've created a trial already, give me login/pass and I
will create a char there. Just don't tell anyone, account sharing is
prohibited. ;-)
removing the release blocker until the user responds. will add it back
after confirming the bug. Feel free to add the blocker, if this is still
needed.
Does this happen in Safari? Has anyone tried bisecting nighties
(http://nightly.webkit.org/) or using Tools/Scripts/bisect-builds (or
git-bisect) to find a more specific revision? if this turns out to be a
WebKit trouble (which would very likely show up in Safari too), then we'll
need a bugs.webkit.org issue to fix.
WB eseidel! I've verified that http://trac.webkit.org/changeset/113885/ causes this regression, but I don't feel comfortable recommending we simply
roll that change out to fix this one. I'm trying to focus on remaining
sub-pixel layout issues if someone else wants to take a look at this. Eric?
Julien?
We need a reduction, to see what's actually changed. Have we verified that
the site looks correct in FireFox? Do we know if our new behavior is
actually incorrect? (i.e. is this actually a "regression" or just
a "behavior change"?)
OK. I can confirm that https://gate.eveonline.com/Profile/etouchqa looks
wrong in both Chrome dev channel, and WebKit nightly. (Reconfirming what
we already knew, that this is a WebKit issue.)
The page does not look "wrong" in Firefox. I have not tested IE or Opera.
Here is a first-pass reduction. This is still not small enough to really
understand what's going on, but the process of reducing is pretty
straightforward if someone would like to continue. I just saved the page
as "web page complete" from chrome, and started removing stuff until this
point.
Note: this is not a good test case yet, as it's not easy to understand what
it's supposed to do. I reduced it to this by comparing our rendering with
firefox's and kept removing stuff while they were still different.
I agree that this is a change in behavior, but it's not yet clear to me
that this is a regression, or that this should block any release.
I would not wait for a patch on this bug. A reduction blocks further
understanding of this issue. Without understanding, we can't decide if
this behavior change is intentional or not/good or not.
