browser detection and redirection

[Spartanicus]

You are over-exaggerating here. Internet Explorer 5.x and IE 6 are prone
to spywares, have unpatched security weaknesses and do make computers
unsafe. Even US-CERT has said so in broad daylight. An objective,
neutral examination of secunia.com website and a wide majority of
security columnists will side with me on this.

I suspect that you are offended when you encounter a message that asks
you to use IE instead of the browser you normally use on the web. You'd
feel that they are trying to push their preference onto you, yet you are
doing the exact same thing to others. You'd splutter something like "but
*I* have my visitor's best interests at heart, the others are plain
wrong". Here's a flash: the others would say the same thing about you.

What is the "best" browser to use is highly debatable depending on your
point of view, it is only a matter of opinion.

Read me again. I do say "please consider" in the above text.

Including "please" when asking someone who walks into a store to
consider changing the colour of their clothes doesn't make the message
any less offensive.

Can *YOU* show me where I actually reject IE on my website??

The issue is your choice of browser, how that is your choice to make,
and how you should not question the choice made by others, or target a
message specifically to people who have made a different choice than you
informing them that you don't agree with their choice.

I don't understand your agressive response to my post.

I'm offended by anyone trying to push their opinion on me, or onto
others.

 

[dorayme]

What is the "best" browser to use is highly debatable depending on your
point of view, it is only a matter of opinion.

The bit after the comma is plain bullshit.

 

[=?ISO-8859-1?Q?G=E9rard_Talbot?=]

Spartanicus wrote :

I suspect that you are offended when you encounter a message that asks
you to use IE instead of the browser you normally use on the web.

No I'm not. I just think these people don't know much about browsers.
And frankly, I have not seen a lot of those websites since 2003 or so.

You'd

feel that they are trying to push their preference onto you, yet you are
doing the exact same thing to others.

No I don't. Go and visit my own website.

http://www.gtalbot.org/

You'll even find at least 20 links to download IE 6. It's been like that
for years too.

You'd splutter something like "but
*I* have my visitor's best interests at heart, the others are plain
wrong". Here's a flash: the others would say the same thing about you.

You totally misunderstand what I do on my website. Or you never visited it.

What is the "best" browser to use is highly debatable depending on your
point of view, it is only a matter of opinion.

I did not create the US-CERT advisory in June 2004. I have not created
the secunia.com website.
CSS 1 and CSS 2.1 support and compliance can be tested, measured and
quantified: E. Meyer and Ian Hickson have done so and others have too.
There is a wide consensus among experts that IE 6 and IE 7 have poor web
standards support, compliance: no one seriously claims otherwise, even
Chris Wilson and David Massy.
HTML 4.01 support and compliance can be tested, measured and quantified:
R. Lionheart has done that.
Number of flaws, vulnerabilities (with demo, proof-of-concept code),
objective severity, gravity, time to patch, etc.. all of this can be
measured, assessed, quantified.

Publicly disclosed security vulnerability without a patch by browsers
http://www.webdevout.net/security_summary.php#public

http://www.webdevout.net/browser_support.php

http://www.webdevout.net/browser_support_summary.php

http://en.wikipedia.org/wiki/Comparison_of_web_browsers

http://en.wikipedia.org/wiki/Comparison_of_layout_engines

Techies Vote: Anything But IE
http://www.emailbattles.com/2006/01/04/browsers_aacfccahdi_hi/

Microsoft Employees Rage As Internet Explorer Ship Sinks
http://www.emailbattles.com/2006/01/03/browsers_aacehieihi_gd/

Sick of IE 6
http://www.designdetector.com/archives/04/06/SickOfIE6.php

Why You Should Dump Internet Explorer
written by a MCSE
http://www.lockergnome.com/nexus/news/2004/06/15/why-you-should-dump-internet-explorer/

It's not just a mere opinion. It's been now 2 days that IE 7 final
release is out and already, 2 security flaws have been confirmed at
secunia.com. I am a bit surprised... but then not *_that much_* surprised.

Including "please" when asking someone who walks into a store to
consider changing the colour of their clothes doesn't make the message
any less offensive.

If I was to promote Opera 9.1 or Firefox 2, I wouldn't be selling
anything. I don't work for them; I am not a share holder of Opera
either. And a simple message promoting either or both product at the
bottom of a webpage wouldn't be invasive or anything like "agressively
pushing". You over-exaggerate here again: what's so offensive with a
browsehappy button or a "download Opera 9" image??

The issue is your choice of browser, how that is your choice to make,
and how you should not question the choice made by others,

I don't question the choice of Safari users. I don't question the choice
of Icab users. I don't question the choice of Dillo users. etc. etc. I
don't question the choice of any users using any browser actually.
Again, you can check my very own website.

or target a

message specifically to people who have made a different choice than you
informing them that you don't agree with their choice.

Basically I would like to say to IE 6 users this: there are better
browsers than IE 6 available out there and there are better browsers
than IE 7.

I'm offended by anyone trying to push their opinion on me, or onto
others.

Why you would feel offended by someone promoting another browser than IE
6 is beyond my comprehension.

All browser manufacturers have promotion banners and promotion buttons.
Even non-browser softwares. Are they all nuts, way out of line???

Gérard

 

[Spartanicus]

No I'm not. I just think these people don't know much about browsers.

Again: the people who advocate any other browser than you are using
think the same about you. You think that they are wrong, they think you
are wrong, actually you are both equally wrong.

It is rude and offensive to ask users who don't use IE to change their
browser to IE, it is no less rude and offensive to ask IE users to
change to something else. There are arguments supporting both requests,
neither set of arguments can claim to be more true than the other. Which
set of arguments matters to a user is 100% dependent on personal
circumstances.

You cannot claim the moral high ground by saying that everyone needs a
secure browser, no browser is secure. The vast majority of IE users have
never suffered any security problems that can be attributed to their
choice of browser. Your claims about IE's security problems amount to
scaremongering, you make this claim under the false pretence that you
have the user's best interest at heart, in reality you are pushing a
private agenda. You are afraid that if you were to reveal the real
reason why you are pestering IE users they'd not take any notice, so
you've resorted to trying to scare them.

You'd

No I don't. Go and visit my own website.

If the code and content you posted is somewhere on your website then
that's all I need to know, the context in which it is displayed is of no
relevance.

You'll even find at least 20 links to download IE 6. It's been like that
for years too.

I ask visitors to my library to change their clothes if I don't like the
colour, but I also point them to places where they can get clothes in
that colour, so that demonstrates that I am not pushing my preference
onto them.

I did not create the US-CERT advisory in June 2004. I have not created
the secunia.com website.
CSS 1 and CSS 2.1 support and compliance can be tested, measured and
quantified: E. Meyer and Ian Hickson have done so and others have too.
There is a wide consensus among experts that IE 6 and IE 7 have poor web
standards support, compliance: no one seriously claims otherwise, even
Chris Wilson and David Massy.
HTML 4.01 support and compliance can be tested, measured and quantified:
R. Lionheart has done that.
Number of flaws, vulnerabilities (with demo, proof-of-concept code),
objective severity, gravity, time to patch, etc.. all of this can be
measured, assessed, quantified.

More personal opinions about what users should care about, the fact that
some others have a similar view doesn't change the fact that even if all
these claims are true, they could well be totally irrelevant to someone
else's choice of browser. You fail to grasp the fundamental point: it's
none of your bleeping business telling others what they should care for
in a browser.

If I was to promote Opera 9.1 or Firefox 2, I wouldn't be selling
anything. I don't work for them; I am not a share holder of Opera
either.

See the previous library example.

And a simple message promoting either or both product at the
bottom of a webpage wouldn't be invasive or anything like "agressively
pushing". You over-exaggerate here again: what's so offensive with a
browsehappy button or a "download Opera 9" image??

The fact that you think that it is your right to pester others about
having made a choice you don't like. You do this using deplorable
tactics and delivery methods.

I don't question the choice of Safari users. I don't question the choice
of Icab users. I don't question the choice of Dillo users. etc. etc. I
don't question the choice of any users using any browser actually.

I don't dislike brown, purple, yellow, I only ask you to change your
clothes when they are red.

Basically I would like to say to IE 6 users this: there are better
browsers than IE 6 available out there and there are better browsers
than IE 7.

What you want is to dictate to others what they should care for in a
browser. You "better" qualification is a personal one. To do so is
offensive and unacceptable in a free society.

Why you would feel offended by someone promoting another browser than IE
6 is beyond my comprehension.

Your comprehension needs enlarging.

All browser manufacturers have promotion banners and promotion buttons.
Even non-browser softwares. Are they all nuts, way out of line???

All clothes manufacturers promote their chosen colour of clothes, that's
fine. This doesn't give the librarian the right to bother people coming
into his library who wear a colour he doesn't like.

 

[dorayme]

It is rude and offensive to ask users who don't use IE to change their
browser to IE, it is no less rude and offensive to ask IE users to

This is a hysterical point of view. It is not rude or offensive
by mere fact of a recommendation that a site is best viewed in
such and such and that this is recommended. It might not be very
practical or effective. But it is not rude. This is really quite
beyond the pale. I am rude. GT is not.

I ask visitors to my library to change their clothes if I don't like the
colour, but I also point them to places where they can get clothes in
that colour, so that demonstrates that I am not pushing my preference
onto them.

This is ridiculous. The colour is totally without connection to
the library. If the library were somehow special and needed some
item that helped in its navigation or if the visitors were
required to remove their shoes because of the silence thereby
promoted, we would be at least getting to the ball park in the
analogies...

You fail to grasp the fundamental point: it's
none of your bleeping business telling others what they should care for
in a browser.

It is not necessarily a point so etched in stone that needs
grasping... It is not obvious at all that it cannot be made, in a
good natured friendly way, the website maker's business. I am not
recommending it, I just don't think your pronouncements are any
kind of argument.

 

[Andy Dingley]

How about telling visitors using buggy, old, non-web-standards-compliant
browsers

How do we identify them? This isn't easy if it's to be robust.

It's practical to identify NS4, but is it really necessary to? NS4
users will have noticed by now that things don't always work tooo well,
and if they're still using it, then presumably they have some other
good reason for sticking with it. Why should one more random website be
a trigger to cause them to upgrade ?

My real objection is to sniffing for IE6, or IE7 build n-1. This sort
of fine-grained sniffing that claims last week's build is now terrible
is the real problem. This sort of test is _never_ maintained and up to
date. A bad or outdated implementation of it is worse than no
implementation.

they may consider switching if they want his webpage code
to render as expected (layout, formating, functionality)?

Advertise the best browsers by all means. Now's a particularly good
time to plug Firefox 2.0
But do this objectively -- "Use Firefox, it's good" rather than "Your
browser sucks, go away"

What's so wrong with such invitation?

The OP is presumably trustworthy -- but should we encourage an attitude
that random websites recommend you to install random browsers that may
very well be dubious?

 

[John Dunlop]

Gérard Talbot:

How about telling visitors using buggy, old, non-web-standards-compliant
browsers

You wouldn't even need browser sniffing to do that.

that they may consider switching if they want his webpage code
to render as expected (layout, formating, functionality)?

I'm not sure what distinction you're drawing between layout and
formatting, but functionality isn't rendered. The HTML *document* is
rendered, rendered according to the *user-agent*.

Who is to say what counts as 'rendered as expected'? Expected by who?
If someone takes it upon themselves to prescribe and proscribe
different renderings, who granted them the authority to do so?

The interworking specifications do not restrict the rendering of HTML
documents, but actually allow for different renderings.

'We do not recommend that authors limit their creativity, only that
they consider alternate renderings in their design.' (HTML4.01: 2.4.2)
http://www.w3.org/TR/html401/intro/intro.html#h-2.4.2

'Providing access to content ... includes enabling users to configure
and control its rendering' (User Agent Accessibility Guidelines 1.0)
http://www.w3.org/TR/WAI-USERAGENT/guidelines.html#gl-user-control-styles

What's so wrong with such invitation?

Nothing is wrong, nothing is right; "wrong" is reductionistic.

Telling users that 'they may consider switching' browsers because of X,
Y, or Z would distract them from the real content - the reason they're
there - in much the same way as any mention of the mechanics would do.

Trouble making your pages backwards-compatible?

 

[Ben C]

Gérard Talbot:


You wouldn't even need browser sniffing to do that.


I'm not sure what distinction you're drawing between layout and
formatting, but functionality isn't rendered. The HTML *document* is
rendered, rendered according to the *user-agent*.

Who is to say what counts as 'rendered as expected'? Expected by who?
W3C.

If someone takes it upon themselves to prescribe and proscribe
different renderings, who granted them the authority to do so?

They do that for HTML, although I don't know if they were actually ever
"granted the authority".

I think they did just take it upon themselves.

The interworking specifications do not restrict the rendering of HTML
documents, but actually allow for different renderings.

Yes, but there are W3C standards for rendering as well.

See http://www.w3.org/TR/CSS21/ which describes the box model in detail.

CSS 2.1 does explictly leave certain things "up to the UA" (exactly
where you put the list item marker bullets for example), and some other
things it just doesn't mention.

But it covers most things, is generally not ambiguous, and Opera and
Firefox conform very well to it. Others have said that even IE is not
that bad in strict mode.

You're right though that if you publish HTML with no styles, you should
have few or no expectations about rendering.

 

[John Dunlop]

Ben C:

[John Dunlop:]

Who is to say what counts as 'rendered as expected'? Expected by who?

W3C.

The W3C neither holds expectations about nor dictates how an HTML
document, even with an associated author stylesheet, should be
rendered.

They do that for HTML,

The W3C do not prescribe and proscribe different renderings for HTML.

although I don't know if they were actually ever "granted the authority".

Bearing in mind that the W3C's founder, Tim Berners-Lee, invented the
World Wide Web, you could argue that they, the consortium, have as much
"right" to make such decrees as anyone.

Yes, but there are W3C standards for rendering as well.

Not so much standards, in so far as the W3C isn't a standards body, as
specifications; but yes, I am aware of the CSS recommendations.

However, stylesheets can be stripped out, turned off, unsupported in
whole or in part, overriden, or even have no bearing on a particular
user-agent (or medium). CSS2.1 defines CSS2.1; it doesn't define how a
document will be rendered. No "standard", W3C-endorsed or otherwise,
defines how documents *will* be rendered.

You're right though that if you publish HTML with no styles, you should
have few or no expectations about rendering.

Well, CSS2.1 offers a default stylesheet for HTML4.01 (Appendix D), so
you could make reasonable guesses about how a document would be
rendered if the only stylesheet applied was a user-agent one *and* you
were familiar with the user-agent and medium in question. But even
with an author stylesheet covering the gamut of HTML elements, I would
hold no expectations about rendering.

 

[Ben C]

Ben C:

[John Dunlop:]

Who is to say what counts as 'rendered as expected'? Expected by who?

W3C.

The W3C neither holds expectations about nor dictates how an HTML
document, even with an associated author stylesheet, should be
rendered.

They do that for HTML,

The W3C do not prescribe and proscribe different renderings for HTML.

I was a bit unclear there. I just meant they take it upon themselves to
make standards.

And inasmuch as they have "authority" to do that for HTML, they also
have it for CSS.

[snip]

Not so much standards, in so far as the W3C isn't a standards body, as
specifications; but yes, I am aware of the CSS recommendations.

However, stylesheets can be stripped out, turned off, unsupported in
whole or in part, overriden, or even have no bearing on a particular
user-agent (or medium). CSS2.1 defines CSS2.1; it doesn't define how a
document will be rendered.

But it does define things about how it will be rendered (or laid out, or
formatted, or whatever you want to call it).

You can safely say, for example, that a conforming UA will not place
right floats to the left of left floats (in the same block formatting
context). If that isn't an expectation about how a document will be
rendered, then I don't know what is.

No "standard", W3C-endorsed or otherwise,
defines how documents *will* be rendered.


Well, CSS2.1 offers a default stylesheet for HTML4.01 (Appendix D), so
you could make reasonable guesses about how a document would be
rendered if the only stylesheet applied was a user-agent one *and* you
were familiar with the user-agent and medium in question.

Good point.

But even with an author stylesheet covering the gamut of HTML
elements, I would hold no expectations about rendering.

You would be justified in holding some, and I think most people probably
do.

 

[dorayme]

Nothing is wrong, nothing is right; "wrong" is reductionistic.

What nonsense Jock!

Telling users that 'they may consider switching' browsers because of X,
Y, or Z would distract them from the real content - the reason they're
there - in much the same way as any mention of the mechanics would do.

You can take too hyterical a view of these things. Humans are
distractable but they are not all ADD sufferers requiring Ritalin.

 

[=?ISO-8859-1?Q?G=E9rard_Talbot?=]

Andy Dingley wrote :

How do we identify them? This isn't easy if it's to be robust.

It's practical to identify NS4, but is it really necessary to?

No, it's not necessary.

NS4
users will have noticed by now that things don't always work tooo well,

Exactly. They are used to/they recognize the quirks of NS 4.

and if they're still using it, then presumably they have some other
good reason for sticking with it.

My guess is it's often lack of knowledge, fear (of losing personal data,
emails, profile info, bookmarks, etc), hardware/memory limitations, cpu
limitations, or visiting only a short list of sites which work well with
their NS 4 (so then, there is no need to upgrade), who knows..
World stats regarding NS 4 put it at under 0.2% and declining.

Why should one more random website be

a trigger to cause them to upgrade ?

I agree. They must have already notice that the layout is often
misaligned, misplaced, jagged, etc... during the last 3-5 years for
sure. And they must have heard of IE 6 or Firefox or Mozilla before. So,
they must have a reason to still use NS 4. I have brought up the same
argument regarding detection of IE5.x: it's rather useless to detect
them and then invite them to upgrade when pretty much everything around
them for the last 5 years must have invited them to upgrade.

My real objection is to sniffing for IE6, or IE7 build n-1. This sort
of fine-grained sniffing that claims last week's build is now terrible
is the real problem. This sort of test is _never_ maintained and up to
date.

Well, this one should be easy to do, easy to maintain (and it does not
rely on js):

<!--[if gte IE 6]>
<p><a href="[link to disclaimer page]">Recommended browsers [or some
other text]</a><p>
<![endif]-->

And disclaimer page says something like
"The webpages of this website use valid markup code, semantic markup
code and valid CSS code. Despite our best coding, testing and technical
efforts, the layout/formating of webpages may not render as intended in
browsers which have an incomplete support of W3C web standards or an
incorrect implementation of W3C web standards. For best viewing
conditions and for best security, we recommend Firefox 2.0 or Opera 9.1
or ..."

A bad or outdated implementation of it is worse than no
implementation.

Of course, very recently at IE blog (can't find where exactly anymore),
it was said that about 33% of all webpages fail to correctly identify
MSIE7; so IE7 users visit webpages which in turn tell them to upgrade to
MSIE6!

Advertise the best browsers by all means.

I agree. In advertising, it's called positive advertisement.

Now's a particularly good
time to plug Firefox 2.0
But do this objectively -- "Use Firefox, it's good" rather than "Your
browser sucks, go away"

Agreed.

Gérard

 

[=?ISO-8859-1?Q?G=E9rard_Talbot?=]

John Dunlop wrote :

Gérard Talbot:


You wouldn't even need browser sniffing to do that.


I'm not sure what distinction you're drawing between layout and
formatting,

As far as my english understanding goes, I make no distinction between
layout and formatting.

but functionality isn't rendered. The HTML *document* is
rendered, rendered according to the *user-agent*.
Correct.


Who is to say what counts as 'rendered as expected'? Expected by who?

By the web author.

If someone takes it upon themselves to prescribe and proscribe
different renderings, who granted them the authority to do so?

W3C.

Let me give you an example, a real one. Just today, I dealt with someone
who had this in his webpage:

<DIV style="position: absolute; top: 220 px; left: 20px; ">... some
image ...</DIV>

When the browser is triggered into standards compliant rendering mode,
the rendering is [roughly] the same in 94% of all CSS-capable web
browsers in use out there. But when the browser is triggering backward
compatible rendering mode, then MSIE 6 and MSIE 7 will *_honor_* the
top: 220 px declaration when it should NOT according to CSS1 and CSS 2.1
parsing error conditions and recovery. This is clearly a case covered by
specs. The
top: 220 px;
declaration must be ignored according to CSS specifications.... but it's
not ignored in MSIE 6 and MSIE 7 in bugward compatible more.

The interworking specifications do not restrict the rendering of HTML
documents, but actually allow for different renderings.

Yes. There is a margin of flexibility, especially in different media.
Visual browsers can have different default values for many CSS
properties. E.g.: on MSIE 6, the default margins on the body were 15px
10px and in MSIE 7, they are 8px (for all 4 sides), just like Netscape
6+, Mozilla Suite, Seamonkey, Safari 2.x, Firefox, etc..

'We do not recommend that authors limit their creativity, only that
they consider alternate renderings in their design.' (HTML4.01: 2.4.2)
http://www.w3.org/TR/html401/intro/intro.html#h-2.4.2

'Providing access to content ... includes enabling users to configure
and control its rendering' (User Agent Accessibility Guidelines 1.0)
http://www.w3.org/TR/WAI-USERAGENT/guidelines.html#gl-user-control-styles


Nothing is wrong, nothing is right; "wrong" is reductionistic.

Telling users that 'they may consider switching' browsers because of X,
Y, or Z would distract them from the real content - the reason they're
there - in much the same way as any mention of the mechanics would do.

Trouble making your pages backwards-compatible?

No trouble. Content and navigation must always be ensured when
designing. Presentation and formatting as intended by the author can not
be ensured though.

Gérard

 

[John Dunlop]

Ben C:

[re CSS2.1 spec]

But it does define things about how it will be rendered

Ok. Please quote which part of the CSS2.1 specification you think
defines how a user-agent *will* or even *should* render an HTML
document. The CSS2.1 specification defines the syntax and semantics of
CSS2.1, nothing more, nothing less.

 

[John Dunlop]

Gérard Talbot:

Correct.

Fine, but...

By the web author.

Well, that's their prerogative, but why would they hold expectations
about how their document will be rendered if they realise that
documents are rendered according to the user-agent? It is impossible,
for example, to know how a document will be rendered by a user-agent on
a platform or medium that is beyond your ken.

W3C.

See my responses to Ben C.

Let me give you an example, a real one. Just today, I dealt with someone
who had this in his webpage:

<DIV style="position: absolute; top: 220 px; left: 20px; ">... some
image ...</DIV>

Imagine a user-agent where (spatial) position makes no sense; imagine a
user-agent where stylesheets are off; imagine a user-agent that doesn't
support absolute positioning; imagine a user-agent who only reads the
HTML document because all author and user stylesheets have been
removed; imagine a user-agent with images off; imagine a user-agent
that doesn't support images. I am sure there are countless scenarios
beyond my imagination.

Content and navigation must always be ensured when designing.

Sound advice.

Presentation and formatting as intended by the author can not be
ensured though.

Agreed.

 

[Ben C]

Ben C:

[re CSS2.1 spec]

But it does define things about how it will be rendered

Ok. Please quote which part of the CSS2.1 specification you think
defines how a user-agent *will* or even *should* render an HTML
document.

I thought we were talking about an HTML document plus some CSS?

The CSS2.1 specification defines the syntax and semantics of
CSS2.1, nothing more, nothing less.

Indeed, and part of the "semantics" of CSS is where things go on the
screen, i.e. how they are rendered.

 

[John Dunlop]

Ben C:

Indeed, and part of the "semantics" of CSS is where things go on the
screen, i.e. how they are rendered.

Yes, the semantics of *CSS*. However, the rendering is ultimately up
to the user-agent. Even if all the rule sets in your author stylesheet
are honoured in full by a conforming user-agent, and you know that they
have been, you still can't reliably predict the rendering because
user-agents have idiosyncrasies.

And that's a big if. I've already given a multitude of instances where
that condition wouldn't be met, and knowing whether that condition is
met or not is not feasible.

 

[Andy Dingley]

You can safely say, for example, that a conforming UA will not place
right floats to the left of left floats (in the same block formatting
context). If that isn't an expectation about how a document will be
rendered, then I don't know what is.

This is perhaps a safe statement to say, but it's also an obscure case
and so is largely unhelpful. As it's only possible safely to say these
pathological cases with any certainty, and it's not possible to be so
precise about the many less definite cases, then it's not a
particularly useful operation to spend considerable time defining the
small subset of CSS rendering that it's possible to define so
forthrightly. Even if it were done, it still wouldn't be of much help
as it leaves far too much unsaid.

The fact that the W3C could be definitively prescriptive about a few
cases doesn't mean that they have been, that they should, or that they
could be equally adamant about the majority of rendering.


Is there really much issue over multiple interpretations of the CSS
spec anyway, as opposed to mis-interpretations (IE box model) that can
be cleared up by more careful study of the spec alone? We have a real
problem with lack of clarity and the second problem, but not (IMHO) the
first.

 

[Ben C]

This is perhaps a safe statement to say, but it's also an obscure case
and so is largely unhelpful.

It was just an example. But it's not particularly obscure, it's part of
the definition of right floats-- that they go to the right.

As it's only possible safely to say these pathological cases with any
certainty, and it's not possible to be so precise about the many less
definite cases, then it's not a particularly useful operation to spend
considerable time defining the small subset of CSS rendering that it's
possible to define so forthrightly. Even if it were done, it still
wouldn't be of much help as it leaves far too much unsaid.

I don't understand. CSS does specify quite forthrightly all manner of
aspects of rendering, and you can have detailed expectations about what
conforming UAs will do.

This applies both to fluid layouts and to pixel exact layouts, both of
which are possible in CSS (although you don't have pixel-exact control
over fonts).

Is there really much issue over multiple interpretations of the CSS
spec anyway, as opposed to mis-interpretations (IE box model) that can
be cleared up by more careful study of the spec alone?

I don't think so. I think as you say, the problems are mostly just
non-conformance. The spec is generally not ambiguous.

 

[Andy Dingley]

I don't think so. I think as you say, the problems are mostly just
non-conformance. The spec is generally not ambiguous.

The spec is never ambiguous, but it is thoroughly and deliberately
abstract. It's specific for definition of a CSS feaature (if not always
clear) but it can't define how this should finally be rendered because
it doesn't know the browser capabilities. Nor should it try to define
this, because the whole "platform independent" aspect of the web is
crucial.