134942 - [meta] single pixel rounding error tracking bug

Status: REOPENED

(Core :: Layout, defect)

Description

[Andrew Hagen]

This will be a tracking bug for the many bugs that are _potentially_ duplicates
of one another. In any case, these bugs are definitely related to each other, at
least in their symptoms.

Comment 1

[Andrew Hagen]

Please feel free to remove and add bugs to the dependency list.

It's likely that all these reported problems are caused by one or two bugs that
are rounding errors. Adding keyword qawanted to try to narrow the issues.

Comment 2

[Andrew Hagen]

A recently fixed bug in this area is bug 83289. 

Comment 3

[Michiel van Leeuwen (email: mvl+moz@)]

Other candidates: bug 131382 and bug 35735

Comment 4

[Andrew Hagen]

Thanks. Adding.

Comment 5

[Warner Young]

I can't find the bug number now, but there used to be one for the scrollbar
"thumb" being drawn wrong.  That was sort of fixed, but once in a while, the
vertical scrollbar thumb leaves a 1 pixel gap at the bottom.

Comment 6

[Andrew Hagen]

Maybe that is bug 80292. 

Comment 7

[Andrew Hagen]

taking

Comment 8

[Markus Hübner]

Adding bug 147547.

Comment 9

[Zac Jacobson]

*** Bug 147453 has been marked as a duplicate of this bug. ***

Comment 10

[Sean Proctor]

bug 130767 should be added to this list

Comment 11

[Andrew Hagen]

I don't have time to keep this up to date anymore. If someone wants to maintain
it, please take it over. Thanks.

Comment 12

[Samir Gehani]

adt: removing nsbeta1 nomination; nominate bugs indiviually as appropriate.

Comment 13

[nemo]

This seems most closely related to bug 147453, although I'm not certain which
one it falls under, so putting this in the tracker.
Seeing this, no matter what DPI I seem to use, in varying levels of severity,
irregardless of font.
However, smaller fonts, unsurprisingly, hide the rounding error.

http://www.meyerweb.com/eric/css/edge/slantastic/demo.html
for example, I see things like this:
http://m8y.org/tmp/screengrab.png  (that was a size 13 Tahoma)

Tried a range of DPIs, from 85 to 120, a number of fonts in various sizes, and
only tended to vary the amount of white showing until I drop below a critical
font size.
For above page, fonts ~<11 display the page fine on my machine, those above,
mess it up.

Windows 2000 SP3 running a nightly from 2003-04-18

Comment 14

[Erich 'Ricky' Iseli]

Adding some bugs which I found to be depending from this meta-bug.

Comment 15

[Christian Hamacher]

I added bugs 135012, 136012, 172162, 171282, 217313, 217825. They are all
related to font corruption by one pixel shifts.

Like many here I believe that all or most of these bugs are either caused by the
same underlying problem, or are extremely closely related. However, the
descriptions are sufficiently different (some talk about only scrolling causing
the problem, some seem to be fixed by special display resolutions, etc.) that
I'm hesitant to close them as duplicates, in order to not 'lose' important
possibly useful debug information that has accumulated in all of the bugs.

Comment 16

[Christian Hamacher]

CC'ing [e-mail] and [e-mail], since they are QA contact and
assignee for many of the individual bugs in the dependency list. Perhaps
knowledge of other, related bugs is in any way useful for them.

Comment 17

[nemo]

http://www.infimum.dk/HTML/slantinfo.html
Just to add to the slantastic problem in an interesting way.
This demo seems to render fine in Mozilla under windows.
Under linux, regardless of what font size I choose, I get white lines appearing
randomly throughout the demo.

Comment 18

[Prognathous]

Bug 98564 ("caret overlaps the last character in textfield - if positioned after
the last char") may also be related.

Prog.

Comment 19

[Jaime Mitchell (use [e-mail] for email)]

*** Bug 295823 has been marked as a duplicate of this bug. ***

Comment 20

[Thomas Goldstein]

Are these dupes of this bug?

https://bugzilla.mozilla.org/show_bug.cgi?id=224753
https://bugzilla.mozilla.org/show_bug.cgi?id=250051
https://bugzilla.mozilla.org/show_bug.cgi?id=254214
https://bugzilla.mozilla.org/show_bug.cgi?id=303558

Also, I made this little test case not long ago: http://stifu.free.fr/dl/temp/1px-bug.html
I assume it is related to this bug?

Comment 21

[Arthur]

Hi Thomas. The bugs you mentioned are all blocking this tracking (or meta) bug. You can see this at the top of the bug entry under "depends on". As the name indicates, its meant to ease keeping track of pixel rounding bugs. Probably one of those dependant bugs covers your testcase. And you can always check whether the latest nightly builds ( http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk/, back up your profile!) still expose the issue at hand.

Comment 22

[Thomas Goldstein]

I totally overlooked the "Depends on" section, sorry about that. I need sleep. I just assumed the bug basically only needed to be fixed in one location, not separately for so many different cases. Thanks for clearing that up.

I'm already running Minefield, and actually suffer from the rounding bug on my personal site, which is why I'm interested. So yes, my test case still shows the bug in the latest nightly.

Comment 23

[Thomas Goldstein]

I just thought I'd point out I just noticed the rounding errors I experienced are now gone in the latest trunk builds. The testcase I posted above (http://stifu.free.fr/dl/temp/1px-bug.html) also works fine now. I thought this was surprising, as none of the rounding bugs got marked as fixed recently.

Comment 24

[Cees T.]

Bug 375240 seems related.

Comment 25

[Ioana (away)]

Went through the dependency list and updated the bugs where possible.

Comment 26

[BMO Automation]

Marking all tracking bugs which haven't been updated since 2014 as INCOMPLETE.
If this bug is still relevant, please reopen it and move it into a bugzilla component related to the work
being tracked. The Core: Tracking component will no longer be used.

Comment 27

[hrustlovro]

I believe bug I noticed may be classified to belong here.
I am experiencing a bug when triangular svg:

<svg width="684.6" height="17.6" class="inclined right" version="1.1" viewBox="0 0 181.1 4.657" xmlns="http://www.w3.org/2000/svg"><path d="m181.1 0v4.657h-181.1z" strokewidth="1.604"></path></svg>

is placed above square div content, one pixel line appears between the two.

Bug is present on desktop browser Firefox for developers, 90.0b3 (and earlier versions)
In other desktop browsers, i.e., Chrome, Edge, bug does not appear, i.e. there is no 1px line between svg and div box.

When I tweak svg (i.e. set d attribute to d="m181.1 0v4.7h-181.1z"), 1px line disappears. It is clearly some sort of a rounding problem.