W3C home > Mailing lists > Public > www-html@w3.org > June 2003

OT: Re: Clarification needed on ALT/TITLE tooltips

From: Tantek Çelik <tantek@cs.stanford.edu>
Date: Thu, 26 Jun 2003 18:02:07 -0700
To: "L. David Baron" <dbaron@dbaron.org>
CC: Brian Bober <netdemonz@yahoo.com>, <www-html@w3.org>
Message-ID: <BB20E55B.29D22%tantek@cs.stanford.edu>
On 6/26/03 2:39 PM, "L. David Baron" <dbaron@dbaron.org> wrote:

> On Thursday 2003-06-26 11:52 -0700, Tantek Çelik wrote:
>> On 6/26/03 6:57 AM, "Brian Bober" <netdemonz@yahoo.com> wrote:
>>> I'm not trying to brag, but Microsoft could learn a bit from Mozilla having
> (Nor would you be in a position to brag.  Nor would I, since the
> cross-platform development model was well established long before I
> became involved in writing code for Mozilla.)
>>> the same codebase on every platform.
>> It's not as black and white as you put it.  Having a separate codebase
>> allowed Tasman to develop at much faster speed (than either IE/Windows or
>> Mozilla), and thus IE5/Mac shipped in March of 2000, with standards support
>> far ahead of any other browser (including Mozilla, whose NS6 release weeks
>> later merely proved my point, since reviews universally panned NS6 in
>> comparison to IE5/Mac), which was also confirmed in numerous reviews.
> Just because NS6 had serious problems doesn't mean that its
> cross-platform development model was bad.  (Cross-platform user
> interface and concurrent cross-platform development of core code are two
> totally different things.)

I've had experience with both cross-platform development (e.g. OpenDoc) and
single platform development (e.g. Tasman v0,v0.1), and my experience, and
the experience of everyone else I know that has done both, was that single
platform development can be done much much faster (not to mention, with
fewer people, which also often means much faster).

How much effort was spent on NS6's cross-platform development, when that
energy could have been put into better standards support instead?

And speed isn't the only issue.  Mac users and developers *know* that Mac
first, Mac only, or Mac specific apps are almost always of higher quality,
and provide a superior user experience than "ported" apps.

> Furthermore, claiming that IE5/Mac's standards support was ahead of
> Mozilla's at the time is a difficult claim to defend,

Not if you check the reviews of both from the time - that's what I meant by
"was also confirmed in numerous reviews". E.g.


Including, for example, Eric Meyer's CSS1 Leaderboard.
Warning, old hyperlink: http://www.webreview2.com/style/leaderboard.shtml

It is unfortunately no longer accessible since Webreview.com morphed.
Enclosed is a snapshot of the .shtml file from 29 November 2000, and an
image grab (.gif) from 8 August 2001 (which also included Opera5).  I'm sure
Eric Meyer has the latest Leaderboard that was published which IE5/Mac
*still* led, even in comparison to v6 browsers.

I grabbed URLs and quotes at the time from other pubs also (e.g. LA Times,
"Create" magazine in France).


> and is also
> unrelated to NS6 having problems in other areas.  (I'm certainly not
> convinced.)  IE5/Mac may have done better on the CSS1 test suite, but
> Mozilla did better on many other tests.  See, for example, my archived
> CSS test results from May of 2000 [1].
>  Which is better depends on how
> support of different things is weighed.

Unfortunately, in 2000, CSS2 was *not* a reliable standard, by any
definition, even in terms of its own contradictions and ambiguities - your
test results page showed test results of a specification (CSS2) which the
CSS working group is *still* in the process of fixing (as we both know all
too well) - and the group is *close*, now, more than three years later.

And your tests, though certainly useful, focused heavily on obscure
unrealistic edge cases (as I've pointed out in the past), rather than trying
for more broad coverage (like the CSS1 test suite), or "real world" cases.

Don't get me wrong, I used your tests to fix numerous bugs, fixes that made
it into IE5.1/Mac (Tasman v0.1), I just don't think any yardstick of CSS2
was really meaningful for measuring "standards support" (in contrast to

>> Also, IE6/Windows benefited from many of the innovations that Tasman was
>> able to make since it was a separate code base that could attempt solving
>> problems that were deemed risky or even impossible (e.g. fixing the box
>> model, DOCTYPE switching etc.), solutions that were also copied by Mozilla
>> (and other browsers) for that matter.
> Claiming that they were copied by Mozilla is a little difficult since
> * Mozilla was planning DOCTYPE switching from at least November 1998 [2]
>  and it was shipped in Mozilla M11 in November 1999.

There is a world of difference between an idea (i.e. planning), and an
actual working implementation shipped to the market.  Although, frankly, I'm
not surprised to see this kind of confusion or implication among the open
source crowd - it seems to crop up all the time.

I'm certainly not taking credit for the "idea" of DOCTYPE switching (which I
don't think came from Mozilla either, but rather Todd Fahrner), just that
IE5/Mac shipped the first DOCTYPE switch (with particular switches between
HTML4, Transitional, XHTML, XML, and effects on the box model, font sizing
etc.), and we certainly had beta versions shipped to beta testers in early
to mid 1999 with it, if you want to compare preview/milestone releases.

I'm claiming that Mozilla copied the DOCTYPE switch in IE5/Mac because
that's pretty much what I saw happen after IE5/Mac shipped.  The particular
choices that IE5/Mac had made for the most part (which DOCTYPES meant what,
and what effects happened) seemed to show up in NS6+ after IE5/Mac shipped
and demonstrated in the market that they worked.

It didn't surprise me that such reverse-engineering took place.  I'm
surprised when I see subtle differences in fact, like treating HTML4
Transitional with URL DOCTYPE as quirky rather than standards (especially
since the CSS1 Test Suite has used that particular DOCTYPE since its

> * Mozilla supported the correct box model since before Mozilla M3 in
>  March 1999 (the oldest release still available).  My memory (which
>  may be incorrect) was that it supported the box model correctly when
>  I first tested in it August/September 1998.

As actually did the *shipping* IE4.01/Mac back in May of 1998 (with
limitations on elements in general beyond DIVs).

But I was specifically referring to the box model _switching_ between CSS1
model and "IE4/Windows" model depending on the DOCTYPE.  AFAIK, IE5/Mac was
the first to implement and ship that.

> (This is off-topic, but I have trouble leaving these statements
> unanswered.

I understand and appreciate the need to clarify history.

> However, I think Tantek was right to respond since I think
> the criticism of Microsoft in the previous messages was inappropriate
> for this list.)


If you look at traffic on the subject, especially among web standards lists
that seem to draw the type of personality that likes to bash Microsoft for
sport (e.g. /. types), there is a very strong weighting of uninformed
(underinformed or misinformed) folks claiming that Mozilla invented
everything, did everything first, and never did any wrong, and the opposite
for Microsoft, which is clearly false.

Most of the time these folks ignore IE5/Mac (and in fact, the Mac in
general), as being irrelevant, or treating it as if it never existed,
because it blows a hole in their black and white vision of the world.  Read
the beginnings of this thread for example, where folks referred to "IE" as
there was only one and without any knowledge of IE5/Mac.

> -David
> [1] http://dbaron.org/css/test/results_20000505
> [2] http://bugzilla.mozilla.org/show_bug.cgi?id=1312


(image/gif attachment: leaderboard.gif)

Received on Thursday, 26 June 2003 21:02:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:15:55 GMT