How much effort to put into supporting pre-DOM browsers?

[Richie]

I went through the past six months or so of entries
in c.l.javascript, and found a couple where people
had expressed opinions about the value of supporting
much older versions of Netscape and IE. The entries
included incidental mention of server logs showing
how many pages had been retrieved by such browsers.

I'd like to get some sort of communal variety of
opinions on how much effort it's worth to put in the
support, or in some cases, the "graceful degradation"
allowing partial capabilities to remain in a page
under the old browser.

My situation is that I have no access to server side
capabilities or CGI, have yet to buckle down and learn
Java, but have a fairly complex application I'd like
accessible via the web.

Are there links answering:

1. How many commercial sites still insist on full
compatibility? Allow significant degradation? Give
up on NN < 6? Ignore browsers with small market
shares?

2. How do developers feel about coding for browsers
older than, say, two years?

3. Are there statistics from a variety of sites
showing use by different browsers?

Thank you.

 

[Fred Oz]

Richie wrote:
[...]

3. Are there statistics from a variety of sites
showing use by different browsers?

At the risk of igniting the "browser stats are rubbish"
argument, here a site that quotes data from a number of
sources. You can see that the statistics vary greatly from site
to site, depending on the audience.

<URL:http://www.upsdell.com/BrowserNews/stat.htm>

Try this for radical stats:

<URL:http://www.mywebsite.force9.co.uk/>

Another good site is w3schools as they keep a history by month:

<URL:http://www.w3schools.com/browsers/browsers_stats.asp>

Just Google for "web browser statistics" and remember that if
the stats are just based on server logs then they are likely to
be unreliable. If there is no explanation of what they are
based on, they are no more use than hearsay.

 

[kaeli]

I went through the past six months or so of entries
in c.l.javascript, and found a couple where people
had expressed opinions about the value of supporting
much older versions of Netscape and IE. The entries
included incidental mention of server logs showing
how many pages had been retrieved by such browsers.

Browsers can spoof what the user agent is.
As can programs that aren't even browsers. I know this because I made one
myself using Java that has to spoof as a browser to do something.

I'd like to get some sort of communal variety of
opinions on how much effort it's worth to put in the
support, or in some cases, the "graceful degradation"
allowing partial capabilities to remain in a page
under the old browser.

Posting on Usenet automatically skews the opinions.
There are a LOT more people here (and on ciwas, etc) concerned with
supporting ALL possible user agents than with any other base population of
programmers. My opinion is that this is because there are a lot more people
who use *nix and Mac here, and they're all pretty tired of sites catering to
the most recent MSIE version. Plus, they are geared toward Internet-as-
Information, not internet (or intranet) apps.
I, for one, am pretty tired of sites not supporting Firefox. Good example:
Semantac download pages are written with vbscript, which is only supported by
MSIE. So I make the effort to have my own pages support as many browsers as
is feasible (that is, I have the time to do it and CAN do it). Now, by
'support', that may just mean it doesn't KILL netscrape 4.
Note that I use Norton. They don't require MSIE. Semantec (-ac?) lost my
business when their trial version didn't work with Firefox.

My situation is that I have no access to server side
capabilities or CGI, have yet to buckle down and learn
Java, but have a fairly complex application I'd like
accessible via the web.

Java is NOT Javascript.
If you're going to post here, you need to be very explicit on that. How to do
something with Java is vastly different, since it runs on the server (unless
we're talking applets). Plus there is such a thing as server-side javascript
if you're talking .NET.

Are there links answering:

I'm sure there are. And all of them are based on opinions or random checks
(statistics), not hard facts.

1. How many commercial sites still insist on full
compatibility?

Not a single one that I visit (that actually does anything useful) supports
NN4. Plain text with navigation notwithstanding.
All things get old. Even operating systems won't support old shit. You don't
have to, either.

Allow significant degradation? Give
up on NN < 6? Ignore browsers with small market
shares?

You can easily code your basic pages to be usable and navigable without
javascript and/or have your javascript do object detection so that it doesn't
totally crash old browsers. If an application has browser requirements, then
so be it. All software I have ever had has "minimum system requirements".
Even Microsoft doesn't support IE4 or Windows 95 anymore!
Keep in mind that there is a definite difference between web pages and a web
application.
(though IMO, if you're going to have an application, you should really be
using server-side scripting if for no other reason than it's really possible
to hack the hell out of anything that runs client-side)

2. How do developers feel about coding for browsers
older than, say, two years?

Personally? I just don't. My regular pages all work in all browsers to the
point where they can be read and navigated and nothing will crash if you view
them in IE3 or NN4.
My web apps all require MSIE 5.5+, NN6+, Firefox, Mozilla, Opera, or another
W3C DOM-supporting browser.
In my opinion, from vast amounts of surfing, most commercial sites have
browser requirements of some kind. Even my bank's site won't work in NN4. And
they state it right on the front page.
Some people have serious issues with this. But if you're talking a commercial
website, and you don't support certain browsers, people who use those
browsers won't be your clients (see my previous Semantec comment). So, it's a
financial decision. Weigh the costs of full development for browsers only a
small segment of the population uses. Most just say "no". Especially when the
other 80% or so of their clients are yammering for improvements made possible
by DOM browsers.

3. Are there statistics from a variety of sites
showing use by different browsers?

Sure. But don't think any are more than 75% accurate or so. I'd wager that a
good 90% of Opera users have it spoof as MSIE because it makes life a lot
easier.

Again, all of this has been MY OPINION. *grin*

--

 

[Michael Winter]

(e-mail address removed) enlightened us with...
[snip]

I'd like to get some sort of communal variety of opinions on how
much effort it's worth to put in the support, or in some cases,
the "graceful degradation" allowing partial capabilities to
remain in a page under the old browser.

As kaeli said, a site shouldn't crash any user agent, nor should a
site be unusable without the support of client-side scripting. Beyond
that, use your time wisely. If it's far too much hassle, or you don't
have the time to write explictly for NN4 or IE4 then don't. The site
will still work which is the main thing.

[snip]

Note that I use Norton. They don't require MSIE. Semantec (-ac?)
lost my business when their trial version didn't work with Firefox.

Last time I checked, Symantec (correct spelling ) produces the
Norton line of products. *Looks at CD* Yup.

[snip]

I'd wager that a good 90% of Opera users have it spoof as MSIE
because it makes life a lot easier.

Not to mention that it's the installation default.

Mike

 

[Dr John Stockton]

JRS: In article <[email protected]>, dated Tue, 8 Mar
2005 23:02:19, seen in Richie

I'd like to get some sort of communal variety of
opinions on how much effort it's worth to put in the
support, or in some cases, the "graceful degradation"
allowing partial capabilities to remain in a page
under the old browser.

Note that some old browsers that are still used a little can be
supported with only a little effort, whereas others perhaps still used
to an equal extent are considerably harder to support, at least for
pages using javascript only in moderation. (Those that use it
immoderately are likely to be about javascript or written by experts or
not worth visiting.)

For example, you may want to use getElementbyId ; it is not available in
MS IE 4, but can be provided by 3 lines of code :-

if (document.all && !document.getElementById) {
document.getElementById = function(id) {
return document.all[id] } }

( I don't say that the doc.all implementation exactly matches the native
one; merely that it generally suffices. )

OTOH, getElementbyId is IIRC not available in NS 4, and AIUI cannot
readily be provided there.


Also, consider your intended readership.

I once tried to visit a "sales" site of a PC company; the site was not
accessible with the browser I then had, but required the then-latest in
order to access sales information. They had not considered that those
most likely to want new computers include those who have only old
computers. Of course, in that case, explicit new-demo pages also make
sense, for those readers who have some access to new computers.

If a site offers a service to the public, there is (in the more
considerate countries) a legal obligation to make its services
accessible to the personally-disabled; there is at least a moral
obligation to do whatever reasonably can be done for the less
technically advanced, who may be financially less abled - charities, for
example.

 

[Richie]

Thank you for the replies!

A little background: I worked in VMS Development
for many years, so I have a good deal of experience
supporting older code. It's just that now I have
far fewer resources and time...

It was interesting and affirming to read through the
statistics, even understanding the limits of their
accuracy. And following links led to pages where I
saw examples of the consequences of limited support,
along with really bad ways of conveying that toward
one's audience.


from .2:

Java is NOT Javascript.

Apologies for my unclear writing. I use JavaScript
extensively. It's just that I never picked up Java
and therefore don't use it and the other technologies
you list for web applications.


from .2:

You can easily code your basic pages to be usable
and navigable without javascript and/or have your
javascript do object detection so that it doesn't
totally crash old browsers. If an application has
browser requirements, then so be it.

My regular pages all work in all browsers to the
point where they can be read and navigated and
nothing will crash if you view them in IE3 or NN4.
My web apps all require MSIE 5.5+, NN6+, Firefox,
Mozilla, Opera, or another W3C DOM-supporting browser.

For simple pages, yes, gentle degradation. For the
complex stuff, well, I guess I just have to accept
the limitations on my market or do the extra work.

-- Richie

 

[kaeli]

Last time I checked, Symantec (correct spelling ) produces the
Norton line of products. *Looks at CD* Yup.

Oops, my bad, I meant McAfee.
Symantec always sticks in my head as not being Norton for some odd reason.

I _really_ need to get more sleep. *sigh*

/ just finished rebuilding her computer after she killed it experimenting
again...

--
--
~kaeli~
"No matter what happens, somebody will find a way to take
it too seriously."
http://www.ipwebdesign.net/wildAtHeart
http://www.ipwebdesign.net/kaelisSpace