Gérard Talbot

Hello Mark Wilton-Jones,

I've written an article on IE7, CSS 2.1 and I am quoting you in this webpage:

[URL]

I will eventually re-structure my own
[URL]
section and add many other bugs I found.

I will (starting this evening) also list your numerous webpage on IE7 bugs
(idem est: http://www.howtocreate.co.uk/wrongWithIE/)
at the bottom of
[URL]
and
[URL]
webpages.

We, web developers, need to stand connected+together and strong on this
issue. It's not good enough to want users to use Firefox 1.5+ or Safari
2.x or Opera 9+; we would need lots of amateur web authors to use better
editing tools, validators, etc..

Take care of yourself and please receive my warmest thanks for the
wonderful, useful, relevant website you maintain :)

Gérard Talbot
--
I truly agree with you when you say: "the latest IE release concentrated
on user interface, and security fixes and ignored the lack of standards
support."

On the vertical-align: middle bug, you say
{
The same applies to align="middle".
}

but you should instead say
{
The same applies to valign="middle".
}

Also, you say
{
Block elements no longer detect hover or click unless the mouse is over
some text within them (this is an old bug that got fixed in IE 7 and
broken again in IE 8).
}

but, in some particular cases, like links with style display: block in
some precise contexts, there were bugs in IE 7:
[URLs]


I have linked your IE 8 browser bug collection in my IE 8 bug webpage:

[URL]

Best regards, Gérard Talbot

Hello Tarquin,

Ok. You're right! :)
But CSS vertical-align: middle for table cells and valign="middle" for
table cells do not render as expected in IE 8 (I know I have tested for
those explicitly):

[URL]

At least, valign="middle" for table header cells is no longer the default
in IE 8.

For images, I have:
[URL]

> In the case of the bugs listed on my IE 8 page, they are only the
> regressions

Ok. Noted. This is also what I thought.

> failing in IE 8. Preparing a list of bugs that also affect earlier
> versions (such as your own list) is out of scope

I understand.
It took me basically 18-24 months to build my IE 7 bug collection ...
while trying to avoid repeating already reported elsewhere IE bugs in that
page.

Best regards,

Gérard Talbot

Hello Tarquin,

When I load
http://www.howtocreate.co.uk/ieBugs/doublebullet.html
in IE 8 beta 1 (XP Pro SP2), then there is no bullet point to the left of
the second sentence, and reloading the page does not change that. So, it
appears, at first, that the testcase is not demonstrating your issue...
But when I select/highlight (with the mouse, by mouse-dragging) the text
of the second sentence, then the list-marker bullet appears. The
list-marker bullet of the 2nd sentence also appears after double-clicking
a word in that sentence.

Furthermore, if I select/highlight (with the mouse, by mouse-dragging) the
list-marker bullet itself of that 2nd sentence, then the list-marker
bullet disappears.

I've tested this very carefully and I can reproduce these "phantomatic"
behaviors consistently: these behaviors are reproducible 100% of the time.

Speaking of list-markers appearing and disappearing, I have a testcase
that demonstrate such behavior:

If you hover the mouse cursor pointer over the text of links in these 2
testcases:

[URL]
[URL]

then you will see an extra, additional list-marker appearing in the
previous list-item. E.g. Hover the mouse over a 2nd link text and 2
list-markers will appear for the first list-item.

Regards, Gérard

Hello Tarquin,

Well, I use XP Pro SP2 and reloading does not make it appear. Just
informing you about this.

>> but it is a redraw issue.

IE 7 had some stacking bugs but now, it seems there are more layer
stacking bugs in IE8b1. I have several examples of this.

[URL]

[URL]
being one of the worse

>> There are multiple ways to produce the same bug

Just curious, Tarquin: do you see an extra, phantom list-marker when you
hover the mouse cursor over the 2nd link in this testcase:

[URL]

...and do you see extra, phantom list-markers appearing (at previous
list-item) when you hover the mouse cursors over the links in this
testcase:

[URL]

I'm wondering if you would be interested in linking your IE8b1 collection
of bugs webpage to mine.

One last question... but an important one. I'm more and more wondering if
creating a web ring of IE8 collection of bugs is an idea that people (like
yourself) would support and join. All that would be needed for
participants would be
- add a particular image (gif, png) indicating its participation into such
web ring of IE8 collection of bugs
- a participant would have to provide at least 2 bugs with respective
testcase
- a participant would have to list links to other web-ring-participants'
IE8 bug collections.

I'm pretty sure several others would join. It certainly would compensate
for absence of IE database bug reporting system (when such semi-public
online database is off) or its current miserability (poor usability, many
annoyances, many bugs being closed instead of being futured, etc.).

Bugs would have to be related to W3C web standards (or official standards
bodies), preferably stable, final and official releases (not working
drafts).

I dearly value W3C web standards. I personally use and prefer Firefox but
I certainly do not object or mind against other browsers, including IE,
achieving compliance, conformance of W3C web standards and more
correctness in its implementation of such web standards. I'm happy when I
read that Safari nightly build passes acid3 test or that an internal
nightly build of Opera passes acid3. I think all people who value web
standards should.

Best regards, Gérard

Hello Mark "Tarquin" Wilton-Jones,

I am in charge of creating CSS tests for the upcoming CSS 2.1 test suite
http://www.w3.org/Style/CSS/Test/CSS2.1/current/
which may originally be coming from outside the W3C community and from
outside of browser manufacturers in the form of bug reports or webpages
showing problems in a particular browser. My work involves turning,
transforming these into usable CSS2.1 test cases which comply with the
required format.

CSS 2.1 test suite, Test Format
[URL]

Once done according to all test format requirements, I try to contact
the people originally involved with the discovery, the investigation of
a test. The test case I do may not be like the original webpage: often,
the test case is small, terse.

Now, on your webpage
http://www.howtocreate.co.uk/ie8.html
I see your min-width test case and I think it would be a great
contribution to the CSS 2.1 test suite. Your min-width test
http://www.howtocreate.co.uk/ieBugs/floatminwidth.html
is already excellent, well built.

I already submitted 14 tests for [name], about 20 tests for
[name] and several others. They are all identifying and crediting
accordingly the people involved and the people who deserve to be
credited.

Here's my "home" basis:
Contributions to CSS 2.1 test suite
[URL]

and the submitted tests so far

[URL]

There is also an incoming folder
[URL]
which just is sort of a "standby" folder for some tests or because they
are not ready to be submitted yet.

The submitted folder is
[URL]
and have all tests which are ready for final reviewing.

I am willing and, in fact, I would be happy to add your name under a
<link> for credit into the test. This is possible under the current CSS
2.1 test suite, file format:

 "
 A test can have multiple author credits if necessary.
 "

 CSS 2.1 Test Suite, Test Format: credits
 [URL]

Once the test(s) have been submitted, then you (and any other people
involved into the discovery of the bug or the making of the tests) have
to sign the W3C form about licensing:
[URL]

... otherwise I can not add your name to credit your involvement.

So far, all of the people ([names]) who accepted to cooperate
with me have signed the W3C form and their form is available via the W3C
archived mailing list. You can view how I filled up this form myself:

[URL]

and how anyone/everyone else have filled up the form by going through the

[URL]

mailing list and by looking for entries which start with

"
[wbs] response to 'Grant II: Grant of License for Contributed Test Cases
Published Outside a W3C Recommendation' WBS Mailer on behalf of
"

Another thing about the filling of the form is that you can update the
data of the form later (until october 2010) or submit an additional form
to complete, update the previous one.

If you have any question or if you would like to submit more/other tests
to be included in the test suite, please let me know about it.

best regards and season's greetings, Gérard Talbot

Tarquin,

One problem that is now occuring is that [company] submitted 7201 test
cases to the CSS 2.1 test suite. So, before even thinking of submitting
a test case, you have to search and see if the test(s) you would want to
submit has not been already submitted to the CSS 2.1 test suite by
Microsoft. This step is now becoming more and more time consuming.

> > In most cases, however, I dislike submitting
> > my work to public archives - I write it either for my employer or for my
> > site.

The CSS 2.1 test suite is not just any ordinary public archive. When
submitting a testcase (or several), you have to fill up a form stating
that you are the author or a contributing author and then give details,
specifics about this. I think you can trust me. You can see how I fill
up that W3C form and see how the others ([names]) have signed that form,
have filled up that form.

That W3C form is mostly to credit people who deserve credit but also to
avoid any battle or fighting over who created the test or discovered +
reported a browser bug one day.

> > In this particular case though, so many browsers make such a mess with
> > this particular piece of CSS that I really would like to see them all
> > fix it.

When I started creating test cases for the CSS 2.1 test suite, it was
because I wanted browsers, in particular upcoming IE9, to fix many CSS
bugs we found in IE8. And one of the only assured, guaranteed way to
have such bugs fixed was to submit relevant test cases to the CSS 2.1
test suite.... as Microsoft would presumably not tolerate, not accept to
be told/replied later something like "Well, you still fail to pass the
CSS 2.1 test suite, ... IE9 fails a bunch of CSS 2.1 tests... Firefox 4
achieves a better score than IE9 in the CSS 2.1 test suite.. etc."

> > Please do not force me to sign up

If you sign (now or later) the W3C form, then I can (and will) add a
<link> crediting you/your involvement in that test.

If you do NOT sign the W3C form, then I can NOT add a <link> crediting
you/your involvement in that test.

> > Use my testcase as a basis to make your own,

This is what I usually do. Then I put such test case in the incoming
folder so that the contributing author (you!) and I can make adjustments
or anything. Last night, I just did that for [name]:

[URL]
[URL]

[name] original test:
[URL]

and today I got an email from [name] saying that he's happy with
those 6 (border-style: hidden on sub-table elements) tests.

> > but leave me out of it. I am signed up to too many systems
> > already.

Fine. No problem. If one day you ever want to change your mind about
this, there won't be a problem either.

My copy of the test case made from your min-width test case will have a
<link> reference crediting you... but the submitted test case to the CSS
2.1 test suite will not have a <link> reference crediting you. This is
what you request.

>> >> Your min-width test is already excellent, well built.
> >
> > It has several problems

Compared to what I've seen before, it is well built :)

> > that will need to be sorted out before it can be
> > used in such a generic suite as the one you are maintaining:

[name] is currently the maintainor of the CSS 2.1 test suite.

> > 2. The test cannot ever pass if the window is narrow

2. browser window viewport width
--------------------------------
 it's possible to edit a PREREQUISITE note in the test regarding medium
(excluding mobile or small screen rendering devices) or imposing some
kind of constraints such as minimum viewport width; right now, it is a
bit early to do that. Also, the uncommon assumptions text of CSS 2.1
test suite states:

"
Uncommon Assumptions

In addition, some of the tests make one or more of the following
assumptions:
(...)
    * The device has a viewport width of at least 640px (approx).
"

> > 3. It will always pass on an extremely wide screen

3. very large browser window viewport width
-------------------------------------------
Believe it or not, we had a discussion about this (what would be a
minimum for a maximum window viewport width) as people are using more
and more larger monitor screen and we want the CSS 2.1 test suite to be
useful, helpful, still reliable and meaningful for at least 6 years and
for various media.

The last email on this particular point was:

[URL]

In some test cases where content must overflow and involving overflow:
auto, then an assumption must be made about the maximum window width of
the UA. We just can not create a single 10,000 character long word
without a white-space and think that this is a good test :)

Right now, I think the assumed maximum window width will be 2048px but
this is not my decision.

> > but this depends on the browser's default font.

Correct. Font-size has an impact in such cases. In all tests from
[name] (see all of the max-*, min-* tests submitted), I used a
(color green and red) font: 200px/1 Ahem;

> > However, I can't force my visitors to install that font,

Ahem is a requirement (to install) for the CSS 2.1 test suite; there is
just too many tests using it.

> > Perhaps you can use Ahem in the W3C suite.

We do use/rely on/ require the Ahem font in the CSS 2.1 test suite.
"many of the tests require the Ahem font to be installed."
http://www.w3.org/Style/CSS/Test/CSS2.1/current/
Maybe 1 test out of 3 I submitted requires the Ahem font to be installed.

Tarquin,
I'll start doing min-width tests based on your test and will let you
know about my progress in the next few days (maybe even today!). The
min-width tests will first appear in
[URL]
and then in
[URL]
and then finally in
[URL]

best regards and season's greetings, Gérard

Tarquin,

I have started to work on the CSS 2.1 min-width tests.

I believe your current

http://www.howtocreate.co.uk/ieBugs/floatminwidth.html

test should explicitly indicate (as an expected result) that the tested
sentence should appear on the left side of the page and not at the right
side of the page.

Also, the ideal font-size unit for this kind of test could be the CSS 3
vw unit defined as:

"The 'vw' unit is relative to the viewport's width. The viewport's width
is equal to 100 'vw' units."
CSS3 Values and Units
http://www.w3.org/TR/css3-values/#relative0

I have started doing tests, experimenting:

[URLs]

I'd say so far the worse browser is Safari 4.0.4; also Konqueror 4.3.4
is not doing so great in those tests.

This may take me more time than expected.

I have been called to work tomorrow (I work after snow storm in [location]
during several days and for approx 14 hours a day) so working on the
min-width tests will take me more than a week. I'm merely notifying you
here.

best regards and all the best for you in 2010,

Gérard Talbot