Where can I get a good list of browser market shares

On the other hand, if you make a competent WWW page, why would any WWW
browser not be able to browse it competently? If you believe that
MSIE needs special treatment, maybe it indicates that you don't
believe that MSIE is capable of browsing WWW pages? (A lot of pages
give the impression that their authors suspected the same.)

If by that you mean IE5 doesn't follow the standards, then yes! Its a
browser we have to support, so find ways around the problem. Now, that
could be by using a different sytlesheet, which I did orininally, or by
taking advantage of even more bugs to cancel out the original bug!
I am actually playing around with the box-model hack to get the styles right
for IE5 / IE5.5, but i'll still use the browser detection for Konqueror
(which doesn't seem to follow any rules of css) and for a very small bug in
Opera where it can't handle the bottom: XXpx; attribute properly. If the
browser isn't one of those I know to have problems though, the site will
display as the original code which works on IE6, Mozilla, new Netscapes and
a few others.

Opera gives its name as MSIE 6 because it means that websites
which don't have very good browser checking will let it view the same
code
as IE does, which in general will be compatible.
As far as I understood it, the original reason for Opera pretending to
be MSIE was because of stupid web deezyners who checked for NN4 or
MSIE and told everyone else to get lost.

Thats sort of what I was getting at. I was referring to checking for
specific browsers and assuming anything else couldn't handle css so reverted
to a cut down layout, but could of course be the complete rejection of the
browser too.

which do want to do something specific for Opera can check for the Opera
string though and work with that.
I.e they make themselves part of the problem, instead of helping to
solve it.

Well of course! We can't make a standards compliant website if none of the
big browsers follow the standards! However, very few sites are so modified
to work on the old browsers that the designer forgets to make it work on new
fully standards compliant browsers too, so if we're just being specific for
old browsers with known problems then thats not hurting the future.

David

On Fri, Jul 11, David Walker inscribed on the eternal scroll:

On the other hand, if you make a competent WWW page, why would any WWW
browser not be able to browse it competently? If you believe that
MSIE needs special treatment, maybe it indicates that you don't
believe that MSIE is capable of browsing WWW pages? (A lot of pages
give the impression that their authors suspected the same.)

If by that you mean IE5 doesn't follow the standards, then yes!

Well, IE does it more ostentatiously than most, in as much as it has
the cheek to document some of its deliberate flouting of mandatory
requirements; but when push comes to shove, all browsers have
shortcomings of one sort or another. So follow principles of graceful
fallback. None of the actual readers care whether the page is
pixel-exactly what the designer had in their head: anyway they'll be
using a different dpi, different window width and height, different
colour depth, display gamma etc. etc. than what the designer was
using: all they care is that the page is above their threshold of
visual appeal (of course that threshold differs from reader to reader,
but there's nothing you can do about that, nor indeed does there seem
any point in trying), and that they can access the page's content,
whatever it may be.

Its a browser we have to support,

It's doing just fine anyway: it doesn't need your support! Let it get
on with the job.

so find ways around the problem.

"The" problem?

Now, that could be by using a different sytlesheet, which I did
orininally, or by taking advantage of even more bugs to cancel out
the original bug!

OK, I'll go along with the latter plan, if you wish to put in the
extra work.

As far as I understood it, the original reason for Opera pretending to
be MSIE was because of stupid web deezyners who checked for NN4 or
MSIE and told everyone else to get lost.

Thats sort of what I was getting at. I was referring to checking for
specific browsers and assuming anything else couldn't handle css so reverted
to a cut down layout, but could of course be the complete rejection of the
browser too.

If I don't know the browser, then I'll give it the benefit of the
doubt. Either it does what the specifications imply it should (i.e
ignore any CSS properties which it cannot support correctly), or its
user will have already needed to learn how to turn its broken CSS
support off on umpteen otherwise-unusable web pages, is my hunch.

which do want to do something specific for Opera can check for the Opera
string though and work with that.
I.e they make themselves part of the problem, instead of helping to
solve it.

Well of course! We can't make a standards compliant website if none of the
big browsers follow the standards!

There's nothing stopping us from designing for graceful fallback,
though. So what if the margin _is_ 15px out at a particular window
size? The readers neither know nor care.

fallback. None of the actual readers care whether the page is
pixel-exactly what the designer had in their head: anyway they'll be
using a different dpi, different window width and height, different
colour depth, display gamma etc. etc. than what the designer was
using: all they care is that the page is above their threshold of
visual appeal (of course that threshold differs from reader to reader,
but there's nothing you can do about that, nor indeed does there seem
any point in trying), and that they can access the page's content,
whatever it may be.

Thats why I've removed a few images for some browsers. They were only
corner images, but the browsers were unable to place them correctly, so it
was easier to just remove them.

Its a browser we have to support,
It's doing just fine anyway: it doesn't need your support! Let it get
on with the job.

If the site doesn't work in a browser that 40% of the world uses then I
think it is quite a major problem, and needs to be resolved...

"The" problem?

Ummmm... "'The' problem" of having so many problems! D

OK, I'll go along with the latter plan, if you wish to put in the
extra work.

Yeah - I think i've actually given up on that plan now. It makes the
stylesheet into a mess of repeated code and junk. With the separate
stylesheets I plan to allow users with problems to set a different
stylesheet, or to remove the stylesheet altogether and give them text only.
That bit will come once I get the site generated by PHP.

If I don't know the browser, then I'll give it the benefit of the
doubt. Either it does what the specifications imply it should (i.e
ignore any CSS properties which it cannot support correctly), or its
user will have already needed to learn how to turn its broken CSS
support off on umpteen otherwise-unusable web pages, is my hunch.

Well, its better than "if I don't know the browser I won't let it in". I
will be adding options for changing and disabling css at some point,
although I can't see it being an issue for the people who will be using the
site anyway.

There's nothing stopping us from designing for graceful fallback,
though. So what if the margin _is_ 15px out at a particular window
size? The readers neither know nor care.

A lot of the errors look very obvious, and for the browsers with ~1%+ share
(on any statistics i've seen) then i've tried to make sure the site will
work well in that, without obvious graphical problems. In some cases that
meant removing images, but in all the ones I tested it does now look right
(except for a bit of text wrapping in Konqueror).

David

On Fri, Jul 11, David Walker inscribed on the eternal scroll:

Its a browser we have to support,

It's doing just fine anyway: it doesn't need your support! Let it get
on with the job.

If the site doesn't work in a browser that 40% of the world uses then I
think it is quite a major problem, and needs to be resolved...

So we come right back to my earlier point: you really DON'T think
that MSIE is capable of browsing WWW pages!

There's nothing stopping us from designing for graceful fallback,
though. So what if the margin _is_ 15px out at a particular window
size? The readers neither know nor care.

A lot of the errors look very obvious,

My diagnosis would have to be that that's because you're not yet
designing for graceful fallback, but are still trying to do
pixel-exact placements etc.

My counsel would have to be that aiming for that would be a _much_
better investment in your time than fiddling around designing multiple
stylesheets and implementing cache-hostile server-side stylesheet
switching.

"Altamir" <altamir (AT) wp (DOT) pl> wrote

Jim Dabell <jim-usenet (AT) jimdabell (DOT) com> wrote

From that very page:
"You cannot - as a web developer - rely only on statistics. Statistics
can
often be misleading."

True. Like all of them. But as a web developer I must rely on
something beside my web server logs.

Why? What better measure of what software and settings your visitors use
than data collected from your visitors themselves? Each web site is going
to attract a different audience with different usage patterns. I say
customize your site based on what you find in your own server logs. The
trick is to make sure your logfile analyses are valid...

-David Safar

David Safar wrote:

"Altamir" <altamir (AT) wp (DOT) pl> wrote in message
news:6507ddef.0307100029.32d90dd (AT) posting (DOT) google.com...
Jim Dabell <jim-usenet (AT) jimdabell (DOT) com> wrote in message
news:<NkydnYbcgs3CuJGiRTvUqw (AT) giganews (DOT) com>...
From that very page:
"You cannot - as a web developer - rely only on statistics. Statistics
can often be misleading."

True. Like all of them. But as a web developer I must rely on
something beside my web server logs.

Why? What better measure of what software and settings your visitors use
than data collected from your visitors themselves? Each web site is going
to attract a different audience with different usage patterns. I say
customize your site based on what you find in your own server logs. The
trick is to make sure your logfile analyses are valid...

I'm interested in the fact that according to this
discussion there are people out their who can develop
sites instantly...generally it takes me a while, so I'm
interested in browser trends and new developments more
than market share

in this business it's no use basing anything on what
happened last month unless you also have a way of looking
ahead


--
eric
www.ericjarvis.co.uk
"Hey Lord don't ask me questions
There ain't no answer in me"

On Sat, Jul 12, David Safar inscribed on the eternal scroll:

True. Like all of them. But as a web developer I must rely on
something beside my web server logs.

Well, if I take the risk of commenting on this feature, then I don't
for a moment want to imply that _any_ kind of browser statistics can
be a proper substitute for well-engineered WWW pages, that can be
relied upon to convey the content to a wide range of client situations
(of which, the browser vendor and model might not be the most
important factors!).

Why? What better measure of what software and settings your visitors use
than data collected from your visitors themselves?

If your site is implacably hostile to a particular browser, say, or to
a particular browsing situation, then it'll be no surprise that those
situations will be under-represented by actual visitors, and this will
be reflected in the logs, which then could easily confirm the author's
misguided assumptions, leading to a vicious cycle.

Each web site is going
to attract a different audience with different usage patterns.

Is it? Most web sites are not about web browsers or about web
browsing situations. If a site about (let's say) basket weaving tells
one to get lost because (let's say) the visitor hasn't got cookies
enabled, then that's going to be one more non-repeat visitor that
won't show up again in their statistics, and will no doubt confirm
their assumption that demanding cookies to be enabled was a vital
component of basket weaving. Do you see any logic in that? Hint: I
don't. And you can repeat that for any other optional feature of
browsing situations, not forgetting of course that the typical
indexing robot also has images turned off, no cookie support, no
javascript, no flash, typically no frames either. Discriminate
against those at your peril:

http://www.google.com/search?q=%22yo...27t+support%22

I say customize your site based on what you find in your own server
logs.

I say make well-engineered web pages about the "topic of discourse".

The trick is to make sure your logfile analyses are valid...

That pretty-much rules out your approach, then. Jeff Goldberg has a
message for you.

In the final analysis, what you do with your own web site is your own
affair, subject to the limitations of prevailing legislation, so don't
let me stop you: some folks in business would be only too happy to
have competitors who deliberately send away potential customers.

"Brian" <brian (AT) wfcr (DOT) org.invalid-remove-this-part> wrote

David Safar wrote:
Why? What better measure of what software and settings your
visitors use than data collected from your visitors themselves?

Perhaps there is no better measure. But it does not follow that such
data gives you an accurate picture.

You are correct. However, it is the most accurate picture available, and
when one requires data, one would do well to use the most accurate data
available.

this thread is about browser usage. To customize a site based on
unreliable data is not very sound. How do you measure for faked
browser strings? How do you account for proxy servers?

Very simple: you don't. That's not how statistics work. When you don't
have data, you don't invent data to "account for" the margin of error. You
ensure that the data you have are as complete as possible, then acknowledge
the margin of error and move on.

The trick is to make sure your logfile analyses are valid...

That *is* the trick. Or would be, anyways. So, how does one "make
sure [her/his] logfile analyses are valid" with respect to browser
usage, the subject of this thread?

I'm open to suggestions on this point. Step number one, clearly, is to use
data for the demographic that can be shown to actually visit the site and/or
similar sites if such data are available (although it is unlikely that a
webmaster will have access to his competitors' logfiles in any legal
manner), not data for all entities with web browers. Step number two is not
to make claims that are unsupported by your data. This is really a much
simpler matter than most people realize. Usually it only requires the
alteration of a word or three. For example:

WRONG: 93% of all hits came from Internet Explorer.
RIGHT: 93% of all recorded hits came from Internet Explorer.

Easy.

Steps 3-n and determination of the value of n are left as exercises for the
reader.

-David Safar

On Sat, Jul 12, David Safar inscribed on the eternal scroll:

You are correct. However, it is the most accurate picture available, and
when one requires data, one would do well to use the most accurate data
available.

Not necessarily. When one knows from first principles that the data
will be biased, with no practical way to compensate for that nor even
to estimate it, then it may be more appropriate to look for solutions
that don't depend on having data.

Remember, you don't estimate TV viewing figures by measuring the
current in the transmitter antenna. You might suppose that method to
be slightly more effective on the web - I mean measuring hits at the
server, of course - but the results could turn out to be misleading,
and - what's worse - could be self-fulfilling.

Very simple: you don't. That's not how statistics work. When you don't
have data, you don't invent data to "account for" the margin of error.

That's right: you assume that the data prove nothing, unless and until
you can demonstrate otherwise by reliable methods of statistical
analysis. Technically this is called the 'null hypothesis'.

You ensure that the data you have are as complete as possible, then
acknowledge the margin of error and move on.

Not so, when the margin of error is precisely the issue.

If the errors are unknown and unknowable, then you can't meaningfully
use the data, you need to find a different approach. Merely
vehemently asserting that the data are the most accurate you can get
does not make them any more useful to a statistician.

WRONG: 93% of all hits came from Internet Explorer.
RIGHT: 93% of all recorded hits came from Internet Explorer.
No - that's still wrong. Try:
93% of all recorded hits came from agents identifying themselves
as Internet Explorer.
(And even the correctness of that statement would need to be reviewed in
the light of the exact method being used.)

Thats actually 93% of hits from agents with "MSIE" somewhere in the browser.
If you test for other browser names, they are almost always in the string
too and so you can put these down as the real browser if you test enough.

David