Just for Firefox?

[Eric B. Bednarz]

from my logs -- just this week:

Well, I can recognize most of the bots (at times, it can be entertaining
to be google-bot, to have fun with idiots, who sniff UA strings [for
some *real* fun, you can even get idiots to block google-bot]).
Besides that, there are proxies (and then, a 'normal' network has *nix
where the work is done, windows with office in the administration and an
iMac at the check-in -- which might mean WYGIWSF [what you get is who
shot first]).

 

[SpaceGirl]

SpaceGirl wrote:

[User-agent strings]

They dont lie usually. They only lie if your too fecking stupid to write
a script that can processes a basic line of text looking for
combinations of matching words. It's not rocket science.

Mine always lies Saying that though, I am not the average user. At
the moment, my browser sends "Dylan/1.0 [en] (Windows NT 5.1; U)" as the
user-agent string.

Today my UA string is:
"Fnorkzilla/5.0 (Windows; Z; Windows NT 5.0a; rv:0.0.9) Wacko/20041231
Fnork/0.9"

Who know what it will be tomorrow.

Yeah, and you're SO representative of Joe public. Just how many people
in the REAL WORLD do you think does that? So, say 90% or more of people
are using IE, so dont fake their UA string. Of the 10% of other people,
probably only a tiny fraction of people would actually bother to do it
anyway (because... why bother? It's something web developers and techie
do, and we may like to THINK we rule the world, there's not actually
billions of us ).

--


x theSpaceGirl (miranda)

# lead designer @ http://www.dhnewmedia.com #
# remove NO SPAM to email, or use form on website #

 

[SpaceGirl]

SpaceGirl wrote:




You make it sound unusual!

Here are some of my visitors
from my logs -- just this week:

Googlebot, msnbot, unknown, psbot, Ilirovatore, TAMU_CS_IRL_CRAWLER,
NutchCVS,Yahoo,W3C_Validator,sohu,SurveyBot,PEERbot.www.peerbot.com,
SearchByUsa,appie,ia_archiver,GeneaSeek,Wget,Bookmark,Amfibibot,mywebcollege,
zerxspid,Faxbot,NaverBot,Baiduspider,Missigua Locator, Ace Explorer,
Gaisbot,eStyleSearch4,Java,sna,WEP Search,
webcollage,Jigsaw,MobileExplorer,Jetbot,IP+Works!,http:,SOFT411 Directory,
NG, Talkro Web,TurnitinBot,MJ12bot,Vagabondo,FavOog,Ultraseek, pipeLiner,
wbdbot,Robozilla,Links SQL, Scooter, NPBot,PHP version
tracker,Y!OASIS,LWP::Simple,Zeus52864WebsterPro,DoctorHTML,PortHuronLabs,
lwp,GoForIT.com, Verizon Superpages Web Crawler, -DIE, SonyEricksson700i.

Even excluding the bots:
Can you say you know how many of *these* are faked?
Can you say whch are even actual web browsers?

I know you can write an agent with a few lines of Perl that will identify
itself as anything you want. *Anything.*

Sometimes it makes sense to test the environment for what a browser will DO.
It doesn't make sense to test for what a browser IS.

What percentage of your logs do these represent? I tiny percentage, I
would imagine, unless you have some really obscure site target at
techies. Plus, most of those are bots. Pretty easy to write a catch all
for something that does look like a real browser.

Anyway - depends what you are using it for. We use SS detection for
modifying the stylesheets we present, and sometimes modifying the
content slightly before it is sent out. Really for the benefit of IE on
the Mac, which is just a bloody weird browser whatever way you look at
it. We simplify our sites for it

--


x theSpaceGirl (miranda)

# lead designer @ http://www.dhnewmedia.com #
# remove NO SPAM to email, or use form on website #

 

[SpaceGirl]

On Sat, 04 Dec 2004 13:33:20 +0000, Toby Inkster

[...]

Safari is basically honest, but includes the magic string "Gecko" in its
user-agent.

Which I never really understood because it's not a Gecko browser.

Hyatt said that he did this so that browser sniffers not capable of
distinguishing Safari would treat Safari like Gecko.

Only crap browser sniffers then.

--


x theSpaceGirl (miranda)

# lead designer @ http://www.dhnewmedia.com #
# remove NO SPAM to email, or use form on website #

 

[NOXwebmasterx]

What percentage of your logs do these represent?

About 5%. Significant.

I tiny percentage, I
would imagine, unless you have some really obscure site target at
techies.

Absolutely not.

Plus, most of those are bots.

In your subjective opinion. We don't know even how many that identify
themselves as bots are faked. 'WE' just don't know.

Pretty easy to write a catch all
for something that does look like a real browser.

Anyway - depends what you are using it for. We use SS detection for
modifying the stylesheets we present, and sometimes modifying the
content slightly before it is sent out. Really for the benefit of IE on
the Mac, which is just a bloody weird browser whatever way you look at
it. We simplify our sites for it

You perform kludges so that you can 'simplify' your site?
Wouldn't it just make more sense to design without kludges?
Do you *really* need to be delivering stuff that requires a bloodhound?
What worthwhile are you really adding to your visitor's experience?

 

[SpaceGirl]

About 5%. Significant.




Absolutely not.

Well, my logs differ. Of the 4 or five million visits we get on our main
server each month, it's about 95% IE, 3% others, and 2% bots (as far as
I can tell). We serve unmodified pages to IE and most other browsers
(except Mac IE), and one of two stylesheets depending on specific
browsers (which dont render 100% valid pages properly).

In your subjective opinion. We don't know even how many that identify
themselves as bots are faked. 'WE' just don't know.

Okay. This is the real world. How many of the 10 of thousands of
computers sold every month will have IE installed? Hmm? Take your head
out of the sand for God sake. We'd all LOVE for something other than MS
to be dominant, but hey, I'd like to win the lotto too pppppppp

You perform kludges so that you can 'simplify' your site?
Wouldn't it just make more sense to design without kludges?
Do you *really* need to be delivering stuff that requires a bloodhound?
What worthwhile are you really adding to your visitor's experience?

If you have ever tried combining Flash and layered content (css and div)
on Mac IE, you'd know what I meant. It's a real bitch It's much
quicker (development wise) to alter the page a little for that platform
(at least for the site in question). Sometimes rather than struggling to
get the same bit of CSS to work in all browsers, it saves a lot of
headache just to serve a DIFFERENT css file.

--


x theSpaceGirl (miranda)

# lead designer @ http://www.dhnewmedia.com #
# remove NO SPAM to email, or use form on website #

 

[C A Upsdell]

You make it sound unusual!
Sometimes it makes sense to test the environment for what a browser will
DO.
It doesn't make sense to test for what a browser IS.

I agree that it does sometimes make more sense to test for what a browser
will do. It makes even more sense to write code that is browser
independent. Unfortunately, this does not suffice for many things -- e.g.
how the browser controls indentation of LI's -- because there is no way for
a page to test for such things.

 

[C A Upsdell]

Only crap browser sniffers then.

Of which, alas, there are far too many. Little things like checking for
'Opera' before checking for 'MSIE', or checking for 'Safari' before checking
for 'Gecko/' (not 'Gecko'!) appears to be too much for many people.

 

[NOXwebmasterx]

How many of the 10 of thousands of
computers sold every month will have IE installed?

Doesn't mean you can ignore the others.

If you have ever tried combining Flash and layered content

Can't imagine ever wanting to do something that silly.

 

[kchayka]

It makes even more sense to write code that is browser
independent.

Yeah, OK...

Unfortunately, this does not suffice for many things --
e.g. how the browser controls indentation of LI's

Huh? You can do this in CSS without any hacks or browser sniffing at all.

-- because there is
no way for a page to test for such things.

There is no need to test for such a thing. If you think you need to do
browser sniffing for the likes of LI indentation, I can only imagine what
else you're doing that is also unnecessary.

 

[Kris]

If you have ever tried combining Flash and layered content (css and div)
on Mac IE, you'd know what I meant. It's a real bitch It's much
quicker (development wise) to alter the page a little for that platform
(at least for the site in question). Sometimes rather than struggling to
get the same bit of CSS to work in all browsers, it saves a lot of
headache just to serve a DIFFERENT css file.

Is that worth the effort for the percentage of visitors MacIE represents
in your logs?

 

[Kris]

Today my UA string is:
"Fnorkzilla/5.0 (Windows; Z; Windows NT 5.0a; rv:0.0.9) Wacko/20041231
Fnork/0.9"

Who know what it will be tomorrow.

Yeah, and you're SO representative of Joe public. Just how many people
in the REAL WORLD do you think does that? So, say 90% or more of people
are using IE, so dont fake their UA string. Of the 10% of other people,
probably only a tiny fraction of people would actually bother to do it
anyway (because... why bother? It's something web developers and techie
do, and we may like to THINK we rule the world, there's not actually
billions of us ).[/QUOTE]

There are thousands of proxy servers run by techies, through which
millions connect to your site.

 

[Frederic Banaszak]

Yeah, and you're SO representative of Joe public. Just how many people
in the REAL WORLD do you think does that? So, say 90% or more of people
are using IE, so dont fake their UA string. Of the 10% of other people,
probably only a tiny fraction of people would actually bother to do it
anyway (because... why bother? It's something web developers and techie
do, and we may like to THINK we rule the world, there's not actually
billions of us ).

I am the master of all I survey...

 

[Toby Inkster]

Anyway - depends what you are using it for. We use SS detection for
modifying the stylesheets we present, and sometimes modifying the
content slightly before it is sent out.

Done this before. It's a recipe for disaster IMHO. Some proxies seem to
cache CSS pages blindy ignoring caching headers.

 

[Toby Inkster]

Well, my logs differ. Of the 4 or five million visits we get on our main
server each month, it's about 95% IE, 3% others, and 2% bots (as far as
I can tell).

Stats for main site at work <URL:http://www.nctpregnancyandbabycare.com/>
(not exactly aimed at geeks) for November 2004. (Over 4 million hits.)

IE6/win 82.5%
IE5.5/win 5.1%
IE5.0/win 4.8%
IE5.x/mac 1.2%
Gecko-based 4.5%
Safari 1.4%
Opera 0.4%

The following also show up in the Top 20 browsers: Konqueror, Dillo, WebTV
and three different PDA-based browsers. We get almost 4% hits from Mac OS
and about 1% from other Unix (incl Linux). FWIW, relating to an earlier
discussion, we get ten times as many hits from Solaris as we do from Risc
OS.

Disclaimer: above site wasn't built by me, though I do some maintainance.
The code is an appalling, appalling state, but at least on most pages I've
replaced the "if your browser is not IE then it must be NN4" browser
detection, which has made parts of it more friendly to Gecko, Opera,
Safari and Konq.

On my personal site though (somewhat less trafficked), I get under 50%
Internet Explorer and over 10% Opera.

 

[Toby Inkster]

No it doesn't. It has the WORD Mozilla in the string, but also has the
words "Internet Explorer".

Except that according to the HTTP 1.1 spec (RFC 2616), anything in
parentheses is a comment and may be ignored. So this HTTP header:

User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)

is syntactically equivalent to this:

User-Agent: Mozilla/4.0

Thus it *LIES*.

Yes, also the words "Netscape" appear in there. Again, easy.

Does it?

User-Agent: Mozilla/4.7 [en] (WinNT; U)

I can't see the word "Netscape" in there anywhere. Perhaps you can.

Netscape >= 6.0 are a bit more honest. They say that they're Mozilla, but
then mention afterwards that they're Netscape as well. Kind of mixed
messages, but at least they tell the truth partly. Opera (default config),
Camino, Firefox and Safari are much the same in that respect -- mixed
messages.

Konqueror, I've checked again, lies outright like MSIE. The magic string
"Konq" is parenthetical.

The only browsers that are 100% honest are used by very small numbers of
people -- e.g. Lynx, Dillo et al.

 

[SpaceGirl]

Yeah, and you're SO representative of Joe public. Just how many people
in the REAL WORLD do you think does that? So, say 90% or more of people
are using IE, so dont fake their UA string. Of the 10% of other people,
probably only a tiny fraction of people would actually bother to do it
anyway (because... why bother? It's something web developers and techie
do, and we may like to THINK we rule the world, there's not actually
billions of us ).

There are thousands of proxy servers run by techies, through which
millions connect to your site.
[/QUOTE]

'millions'. yeah.

Anyway, if you THINK about it the answer is pretty simple; so long as
your script 'fails safe' then it's cool. If our script cant work out
what the browser visiting is, it just assumes IE (which is most likely).
And I would suspect that the target audience for the sort of sites we do
are NOT that likely to be using proxies anyway. Again, that is the realm
of techies - or corporate browsers.

--


x theSpaceGirl (miranda)

# lead designer @ http://www.dhnewmedia.com #
# remove NO SPAM to email, or use form on website #

 

[SpaceGirl]

Except that according to the HTTP 1.1 spec (RFC 2616), anything in
parentheses is a comment and may be ignored. So this HTTP header:

User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)

is syntactically equivalent to this:

User-Agent: Mozilla/4.0

Thus it *LIES*.

Mr pedantic pants. Okay, so it lies. But what the hell does that have to
do with browser detection? You can still tell it is IE.

Yes, also the words "Netscape" appear in there. Again, easy.

Does it?

User-Agent: Mozilla/4.7 [en] (WinNT; U)

Okay fair enough But, you can assume that version is NS.

I can't see the word "Netscape" in there anywhere. Perhaps you can.

I forgot my magic lenses today.

Netscape >= 6.0 are a bit more honest. They say that they're Mozilla, but
then mention afterwards that they're Netscape as well. Kind of mixed
messages, but at least they tell the truth partly. Opera (default config),
Camino, Firefox and Safari are much the same in that respect -- mixed
messages.

Not really. You can work out by reading the string what browser they are
most likely to be.

Konqueror, I've checked again, lies outright like MSIE. The magic string
"Konq" is parenthetical.

So what? We were talking about scripts that identify the browser. Just
because the spec says 'anything in () is a comment' doesn't mean you
cant read it!!! Dumbass The string is still sent to the server so I
can read it and in turn decide what the client is

The only browsers that are 100% honest are used by very small numbers of
people -- e.g. Lynx, Dillo et al.

....

--


x theSpaceGirl (miranda)

# lead designer @ http://www.dhnewmedia.com #
# remove NO SPAM to email, or use form on website #

 

[SpaceGirl]

Is that worth the effort for the percentage of visitors MacIE represents
in your logs?

Yes - because they represent a chunk of our core market.

Plus we dont IGNORE other users - we just present them with generic 100%
xhtml & css (with IE fixes) if we cant work out what they are. For
example, we use a lot of PNGs. These aren't supported by IE (properly)
without using IE specific (and totally invalid) CSS. I want to AVOID
sending deliberately broken code to browsers like FireFox just to keep
IE happy. So, our pages get modified.

--


x theSpaceGirl (miranda)

# lead designer @ http://www.dhnewmedia.com #
# remove NO SPAM to email, or use form on website #

 

[Jeffrey Silverman]

Well, my logs differ. Of the 4 or five million visits we get on our main
server each month, it's about 95% IE, 3% others, and 2% bots (as far as

My logs indicate only about 70-80% MSIE any flavor, 15-20% Netscape (well,
gecko-based) and then the rest.

Anyone else?