At 7:17 PM -0800 on 3/8/00, Alain Farmer wrote:
>Alain: The debate rages on! Are CGI programs that
>don't do HTTP headers by definition 'broken'? Despite
>the fact that they have been in use for several years
>already without any complaints from my thousands of
>'captive' clients?
Yes. The HTTP/1.0 standard, HTTP/1.1 proposed standard, and the
HTTP/0.9 standard practice ALL requires a certain set of headers. When
you don't otherwise pass headers, the document itself becomes a
header(!).
Actually, the cache Erol's is running is broken too; a cache is
required to pass any headers it does not understand -- in this case,
the webpage.
The reason that the web browsers display this is another requirement:
Be conservative in what you send, liberal in what you accept.
A CGI should send the most standards-compliant header it can. A browser
should accpet and deal with the worst mess concievable.
>
>> Adrian: We could definitely do that.
>> There will be a way to incorporate them
>> into the new list archive
>> that we will set up too.
>
>Alain: How about a mySQL database ?
That could be scripted, too.
>
>> Adrian: We have much to sort out yet. I will
>> continue my experiments with SourceForge.
>
>Alain: It's not that important but I was wondering if
>anyone was aware of the fact that I was proposing this
>move to sourceForge weeks ago ... in 3 installments.
>It was put on the back-burner because we didn't want
>to rush into anything. What a difference a few weeks
>makes.
I think the argument was a let's wait and see one, and investigate
alternatives. Sourceforge is still growing strong -- and is better than
ever. And no alternatives have popped up. I concede, you were right in
the first place.
