Noble web site
Discussion
zaktoo said:Are you using a browser from the Ark? I didn't know any browser that took attention of the content type when it's getting the first object at a URL???
Please tell my why on earth they set their content-type header to text/plain? Anyone from Noble browse these fora that can sort it out please?
J
zaktoo said:I didn't say that, I said a browser that works out the content for the *first* object get.
Only a foolish browser ignores content-type headers as a matter of fact.
If you actually telnet to port 80 and GET "/" you'll find that it's actually a broken HTTP1.1 "/" redirect implementation issue that was cronic some time ago in the hosting field. Microsoft, being a sensible company, obviously wishes their browser to overcome classic server misconfigurations and hence actually inspects the contents *as well as* the mime type and makes a decision based on that.
Other browsers that slavishly follow the HTTP spec will obviously break when they find websites that aren't configured correctly.
You tell me which is better - to overcome "classic" server misconfigurations and give the user a better expierience, or slavishly follow a specification singing "lalala" when things break????
Half our life is spent at work getting customers to configure their FM server web servers correct (the unix people are far worse it seems in this respect) for their types.
Unfortuantly it seems the (major UK) ISP hosting the noble website is a classic "ASP on UNIX" issue where they haven't configured their webserver correctly, and hence isn't a "design" issue but a problem with the hosting provider.
I'll send the factory some helpful pointers (we only provide the email and connectivity for the factory) that will hopefully help you sort it out for all you non-MS people.
J
joust said:
If you actually telnet to port 80 and GET "/" you'll find that it's actually a broken HTTP1.1 "/" redirect implementation issue that was cronic some time ago in the hosting field. Microsoft, being a sensible company, obviously wishes their browser to overcome classic server misconfigurations and hence actually inspects the contents *as well as* the mime type and makes a decision based on that.
Joust, I hate to be difficult... :P
My browsers (Mozilla, Firfox, Lynx, Netscape 4.7) have so far all failed to display the html from the redirected page because the content from that page is served as text/plain, and rendered accordingly. All 4 coped perfectly with the redirect from "/". As far as I can see, the redirect, whilst not being good way to do it, is not actually deviating from the spec at that point. It seems to be valid from the headers right through to the closing html tag.
joust said:
Other browsers that slavishly follow the HTTP spec will obviously break when they find websites that aren't configured correctly.
To be fair, when the server instructs a resource to be treated as X, it isn't really on to second-guess it and treat it as Y. In my humble opinion anyway.
joust said:
You tell me which is better - to overcome "classic" server misconfigurations and give the user a better expierience, or slavishly follow a specification singing "lalala" when things break????
I think that depends entirely on what the misconfiguration is, and whether or not it is clear that it is a misconfiguration or intentionally so set.
joust said:
Half our life is spent at work getting customers to configure their FM server web servers correct (the unix people are far worse it seems in this respect) for their types.
Umm... what is an "FM server"? Just curious
joust said:
Unfortuantly it seems the (major UK) ISP hosting the noble website is a classic "ASP on UNIX" issue where they haven't configured their webserver correctly, and hence isn't a "design" issue but a problem with the hosting provider.
Absolutely. However, whoever did do the Noble website really did themselves a disservice by not checking it in at least a few different types of browser - they surely would have found & corrected this fault pretty quickly had they done so... Running sites through the W3 validation service is also a good way to pick up faults too; I find it indespensible for my web work really.
joust said:
I'll send the factory some helpful pointers (we only provide the email and connectivity for the factory) that will hopefully help you sort it out for all you non-MS people.
Thanks Joust, appreciate that!
Ciao
Zak
zaktoo said:
My browsers (Mozilla, Firfox, Lynx, Netscape 4.7) have so far all failed to display the html from the redirected page because the content from that page is served as text/plain, and rendered accordingly. All 4 coped perfectly with the redirect from "/". As far as I can see, the redirect, whilst not being good way to do it, is not actually deviating from the spec at that point. It seems to be valid from the headers right through to the closing html tag.
No - the problem is to do with apache source code derived servers and configs.
To get the asp executed you need to do a re-write rule, it's that that breaks the MIME type
zaktoo said:
To be fair, when the server instructs a resource to be treated as X, it isn't really on to second-guess it and treat it as Y. In my humble opinion anyway.
No different than a program slavishly expecting that a .mpg file is actually a MPEG2 file (it could be MPEG4 or even DivX). You should always inspect the contents to see if they agree (try renaming a .mpg as a .avi and see which programs cope with it).
We don't live in a perfect world, and any good HCI graduate will tell you that useability is the first concern, standards following the second. IMHO its fundamentals of software engineering.
zaktoo said:
I think that depends entirely on what the misconfiguration is, and whether or not it is clear that it is a misconfiguration or intentionally so set.
Why? Who cares if it's a mistake or intentional. The fact is that MS copes with it, Netscape et. all don't. That dissadvantages the users of those packagaes, and hence breaks the fundamentals of good Human Computer Interface design!
zaktoo said:
Umm... what is an "FM server"? Just curious
Facility Managed Server - where customers have total control over a server installed in our datacentre(s).
zaktoo said:
Absolutely. However, whoever did do the Noble website really did themselves a disservice by not checking it in at least a few different types of browser - they surely would have found & corrected this fault pretty quickly had they done so... Running sites through the W3 validation service is also a good way to pick up faults too; I find it indespensible for my web work really.
True - but it doesn't excuse either the lack of HCI issues being designed into those other browsers either in my humble opinion.
J
>> Edited by joust on Friday 14th May 07:13
Hmm.
On one hand it it handy that MS completely ignores MIME headers and guesses the content, but on other it sucks...
1) it allows web designers to incorrectly design web pages - they assume that because it works in IE and everyone has IE then it's correct.
2) it allows sysadmins to incorrectly configure their webservers (as in this case)
3)if you try and do some dynamic content delivery e.g. serve a dynamic PDF from within say PHP or ColdFusion - even though you can force the MIME type within the code to beingt application/PDF IE can still assume the page to be PHP or CF etc. (this is only really a problem on older IE on Macs).
IE also allows lots of other HTML sins as well - which if not tested for in other browsers causes problems.
IMHO (and experience) the W3C standards are perfectly flexible enough to meet anyones HCI requirements. And codeing correct web pages and setting up servers correctly is not that difficult to get right.
Microsoft allowing poor design and web server config issues to be ignored simply allows for the propagation of more design and server errors.
Amusingly - whoever designed this website for Noble had Netscape in mind as the javascript code for the DHTML menus checks for DOM model. Shame it was never tested on the live server
Anyway - should be an easy fix to get the server right.
On one hand it it handy that MS completely ignores MIME headers and guesses the content, but on other it sucks...
1) it allows web designers to incorrectly design web pages - they assume that because it works in IE and everyone has IE then it's correct.
2) it allows sysadmins to incorrectly configure their webservers (as in this case)
3)if you try and do some dynamic content delivery e.g. serve a dynamic PDF from within say PHP or ColdFusion - even though you can force the MIME type within the code to beingt application/PDF IE can still assume the page to be PHP or CF etc. (this is only really a problem on older IE on Macs).
IE also allows lots of other HTML sins as well - which if not tested for in other browsers causes problems.
IMHO (and experience) the W3C standards are perfectly flexible enough to meet anyones HCI requirements. And codeing correct web pages and setting up servers correctly is not that difficult to get right.
Microsoft allowing poor design and web server config issues to be ignored simply allows for the propagation of more design and server errors.
Amusingly - whoever designed this website for Noble had Netscape in mind as the javascript code for the DHTML menus checks for DOM model. Shame it was never tested on the live server
Anyway - should be an easy fix to get the server right.
From our website, ahem;
"The HTK website has been designed and coded to meet strict W3C standards and WAI Accessibility guidelines. Our web experts can help you to address these issues on your website."
My web guy tells me that we had one of (if not the first) WAI-AAA site in the World! But then web guys do tend to think that evey line of code they write is worthy of a Nobel prize... no offence to all you web guys out there!
Anyway, no real point for the post other than to say "nana, na-nana" and a big "boo-sucks" to whoever is responsible.
And don't get me started on software engineering...
"The HTK website has been designed and coded to meet strict W3C standards and WAI Accessibility guidelines. Our web experts can help you to address these issues on your website."
My web guy tells me that we had one of (if not the first) WAI-AAA site in the World! But then web guys do tend to think that evey line of code they write is worthy of a Nobel prize... no offence to all you web guys out there!
Anyway, no real point for the post other than to say "nana, na-nana" and a big "boo-sucks" to whoever is responsible.
And don't get me started on software engineering...
Which is the greater evil?
Taking the "pure" stand and then having 100's of millions of users around the world not being able to view a small but significant number of sites (due to crap sysadmins / crap programmers)
or
Working around the "real" world to give the best possible expierence to the user.
Personally I'd vote for the second everytime - particularly if 70% of the world is using your product.
However, as you rightly point out it can lead to a virtuitous circle - but that's life I suppose.
Of course others may/will take a different view.
J
Taking the "pure" stand and then having 100's of millions of users around the world not being able to view a small but significant number of sites (due to crap sysadmins / crap programmers)
or
Working around the "real" world to give the best possible expierence to the user.
Personally I'd vote for the second everytime - particularly if 70% of the world is using your product.
However, as you rightly point out it can lead to a virtuitous circle - but that's life I suppose.
Of course others may/will take a different view.
J
joust said:
From our website, ahem;
"The HTK website has been designed and coded to meet strict W3C standards and WAI Accessibility guidelines. Our web experts can help you to address these issues on your website."
Are you advertising...?????
Ted! Mods!
J
Didn't even cross my mind!
joust said:
Personally I'd vote for the second everytime - particularly if 70% of the world is using your product.
J
I agree - but sadly for our target audience (we're an acedemic STM publisher) IE usage is less than 45% of the browsers used. We've only just been able to drop support for v3 browser FFS!
So we have to (at least try to) get the standards and servers right, which for the most part I thnk we do.
FunkyGibbon said:Ah - the world of academia
I agree - but sadly for our target audience (we're an acedemic STM publisher) IE usage is less than 45% of the browsers used. We've only just been able to drop support for v3 browser FFS!
So we have to (at least try to) get the standards and servers right, which for the most part I thnk we do.
J
Gassing Station | Noble | Top of Page | What's New | My Stuff