Re: Oldest `still relevant' browser?
Thomas 'PointedEars' Lahn
> Which, in your opinion, is the oldest browser which a modern web designer
> needs to be aware of?
You are opening up a can of worms here :-) Besides, it is a .misc question.
X-Post & Followup-To set.
> If it were, for instance, IE 5 then one could download that browser and
> work under the assumption that ``whatever renders correctly under IE 5 is
> acceptable".
No, for rendering differs considerably in later versions.
> So you could not use the pseudo-class :link for anything but
> <a> tags (since IE 5 ignores the :link class for all other tags, I
> think),
You could use it in the CSS code but it would not be applied there.
So you would probably want to separate those rulesets: Not
foo:link,
bar baz
{
...
}
but
foo:link
{
...
}
bar baz
{
...
}
> but your site would at least degrade gracefully. I would never want to
> design a web site that said: ``Best viewed in browser x", but standards
> can not be relied upon because browsers are a bit behind w3 in their
> implementation of the same.
In particular, one specific, yet the most distributed layout engine, MSHTML.
Others have pretty much catched up, are even ahead by now (Gecko and WebKit,
for example, currently implement some features that are only defined in
HTML5 and CSS3, which are mostly Working Drafts); whether that is a Good
Thing remains to be seen.
> What do you think? Is there some limiting case where people should just
> be told to ``get a new browser"?
It is perhaps a good idea to no longer test by default in browser versions
that have met their end-of-life, because they no longer receive security
updates, and encouraging users to continue using them would probably pose a
greater problem to them and others than if they upgraded/changed.
But it depends. If you have an important customer who uses an old browser
and you cannot convince them to upgrade/change, you will have no choice but
to change the site so that it looks reasonably well in that browser. I had
that case with IE 5 for Mac in 2007 when it had met its end-of-life in 2003.
You also need to be aware that some browsers and browser versions require
specific operating systems. For example, there is no IE 6+ for the Mac, no
IE 7+ for Windows 2000/Me/XP SP1, and there will be no IE 9+ for Windows
XP/Vista SP1. If you try to make users use at least a specific browser
version, you may be forcing them to do more than just download and install a
browser for free. Many users will not do that and may move on to the
competition instead.
I am tracking the product cycle of script-supporting applications, including
browsers, in the ECMAScript Support Matrix, which might be of use here (the
cross indicates end-of-life as announced by the vendor; CMIIW):
<http://PointedEars.de/es-matrix#version-info>
> Is there another reason why important parts of a web site should not rely
> on the most modern CSS, HTML, JS,
But do you really know what "JS" means? See above.
> etc.?
Yes, interoperability and, therefore, accessibility. But caring for the
oldest and developing exclusively for the newest ones are not exact
opposites.
HTH
PointedEars
--
Prototype.js was written by people who don't know javascript for people
who don't know javascript. People who don't know javascript are not
the best source of advice on designing systems that use javascript.
-- Richard Cornford, cljs, <f806at$ail$1$8300...@news.demon.co.uk>
Johan Förberg
> No, for rendering differs considerably in later versions.
I was unclear here. The specific example I was thinking of was writing a
pop-down menu without using JavaScript (or ECMAscript, or Jscript). This
could be done by givin a (visibility: visible) property to the (:hover)
pseudo-class of the drop-down <div>. This would not work in IE 5, since
it ignores this for all but <a> tags. Without having tested, I think the
resulting menu would be dysfunctional.
> In particular, one specific, yet the most distributed layout engine,
> MSHTML. Others have pretty much catched up, are even ahead by now (Gecko
> and WebKit, for example, currently implement some features that are only
> defined in HTML5 and CSS3, which are mostly Working Drafts); whether
> that is a Good Thing remains to be seen.
Yes, indeed. I will be anxious to see how the new <video> tag is received
by browser developers. Thanks to Youtube.com et al, video on the Internet
has become synonymous with diverse video player implementations depending
on flash. I think this relationship is very problematic, but it remains
to be seen if it will be remedied by HTML 5.
> But it depends. If you have an important customer who uses an old
> browser and you cannot convince them to upgrade/change, you will have no
> choice but to change the site so that it looks reasonably well in that
> browser. I had that case with IE 5 for Mac in 2007 when it had met its
> end-of-life in 2003.
Yes, this is mostly a discussion about IE users. People who have enough
"new and interesting!" in them to try other browsers probably do not
postpone updates for seven years. But still there are IE 5 users out
there.
> You also need to be aware that some browsers and browser versions
> require specific operating systems. For example, there is no IE 6+ for
> the Mac, no IE 7+ for Windows 2000/Me/XP SP1, and there will be no IE 9+
> for Windows XP/Vista SP1. If you try to make users use at least a
> specific browser version, you may be forcing them to do more than just
> download and install a browser for free. Many users will not do that
> and may move on to the competition instead.
The webpage which I take care of is not significant enough that people
will upgrade their browser to see it. What I am aiming for is a
functional and nice-looking layout in most browsers.
Thanks for your helpful responses!
Johan Förberg
Johan Förberg
> No, for rendering differs considerably in later versions.
I was unclear here. The specific example I was thinking of was writing a
pop-down menu without using JavaScript (or ECMAscript, or Jscript). This
could be done by givin a (visibility: visible) property to the (:hover)
pseudo-class of the drop-down <div>. This would not work in IE 5, since
it ignores this for all but <a> tags. Without having tested, I think the
resulting menu would be dysfunctional.
> In particular, one specific, yet the most distributed layout engine,
> MSHTML. Others have pretty much catched up, are even ahead by now (Gecko
> and WebKit, for example, currently implement some features that are only
> defined in HTML5 and CSS3, which are mostly Working Drafts); whether
> that is a Good Thing remains to be seen.
Yes, indeed. I will be anxious to see how the new <video> tag is received
by browser developers. Thanks to Youtube.com et al, video on the Internet
has become synonymous with diverse video player implementations depending
on flash. I think this relationship is very problematic, but it remains
to be seen if it will be remedied by HTML 5.
> But it depends. If you have an important customer who uses an old
> browser and you cannot convince them to upgrade/change, you will have no
> choice but to change the site so that it looks reasonably well in that
> browser. I had that case with IE 5 for Mac in 2007 when it had met its
> end-of-life in 2003.
Yes, this is mostly a discussion about IE users. People who have enough
"new and interesting!" in them to try other browsers probably do not
postpone updates for seven years. But still there are IE 5 users out
there.
> You also need to be aware that some browsers and browser versions
> require specific operating systems. For example, there is no IE 6+ for
> the Mac, no IE 7+ for Windows 2000/Me/XP SP1, and there will be no IE 9+
> for Windows XP/Vista SP1. If you try to make users use at least a
> specific browser version, you may be forcing them to do more than just
> download and install a browser for free. Many users will not do that
> and may move on to the competition instead.
The webpage which I take care of is not significant enough that people
Thomas Mlynarczyk
> I was unclear here. The specific example I was thinking of was writing a
> pop-down menu without using JavaScript (or ECMAscript, or Jscript). This
> could be done by givin a (visibility: visible) property to the (:hover)
> pseudo-class of the drop-down <div>. This would not work in IE 5, since
> it ignores this for all but <a> tags. Without having tested, I think the
> resulting menu would be dysfunctional.
There is a workaround for older versions of IE adding :hover support for
any element:
http://www.xs4all.nl/~peterned/csshover.html
(The site doesn't seem to mention IE5 anymore, but if I remember
correctly, the script should also work with that dinosaur.)
Greetings,
Thomas
--
Ce n'est pas parce qu'ils sont nombreux à avoir tort qu'ils ont raison!
(Coluche)
Thomas 'PointedEars' Lahn
> Thomas 'PointedEars' Lahn wrote:
>> No, for rendering differs considerably in later versions.
>
> I was unclear here. The specific example I was thinking of was writing a
> pop-down menu without using JavaScript (or ECMAscript, or Jscript). This
> could be done by givin a (visibility: visible) property to the (:hover)
> pseudo-class of the drop-down <div>. This would not work in IE 5, since
> it ignores this for all but <a> tags. Without having tested, I think the
> resulting menu would be dysfunctional.
There is no good reason to not use scripting here in a way that degrades
gracefully. In particular, there is no good reason not to use ordinary
links here. So the menu would not drop down, but you could still navigate
without scripting.
>> But it depends. If you have an important customer who uses an old
>> browser and you cannot convince them to upgrade/change, you will have no
>> choice but to change the site so that it looks reasonably well in that
>> browser. I had that case with IE 5 for Mac in 2007 when it had met its
>> end-of-life in 2003.
>
> Yes, this is mostly a discussion about IE users. People who have enough
> "new and interesting!" in them to try other browsers probably do not
> postpone updates for seven years. But still there are IE 5 users out
> there.
Although that is difficult to see, given relatively realistic statistics
from gs.statcounter.com or netapplications.com, and its end-of-life in
2005-06. Therefore, I would not test in or jump through hoops for IE 5.x
unless someone really important told me to. Client-side script libraries
maybe staying the exception, although I am considering to drop JScript 5.x
support there, too.
> Thanks for your helpful responses!
You are welcome.
PointedEars
Thomas 'PointedEars' Lahn
> Johan Förberg schrieb:
>> I was unclear here. The specific example I was thinking of was writing a
>> pop-down menu without using JavaScript (or ECMAscript, or Jscript). This
>> could be done by givin a (visibility: visible) property to the (:hover)
>> pseudo-class of the drop-down <div>. This would not work in IE 5, since
>> it ignores this for all but <a> tags. Without having tested, I think the
>> resulting menu would be dysfunctional.
>
> There is a workaround for older versions of IE adding :hover support for
> any element:
> http://www.xs4all.nl/~peterned/csshover.html
> (The site doesn't seem to mention IE5 anymore, but if I remember
> correctly, the script should also work with that dinosaur.)
You do not want to use that script unchanged. All that pointless UA-
sniffing ... use Conditional Comments to exclude supporting versions
instead.
PointedEars
Thomas Mlynarczyk
>> There is a workaround for older versions of IE adding :hover support for
>> any element:
>> http://www.xs4all.nl/~peterned/csshover.html
> You do not want to use that script unchanged. All that pointless UA-
> sniffing ... use Conditional Comments to exclude supporting versions
> instead.
You're right, that's how I do it.