Re: Minutes: GEO telecon 20031022

Hello Richard,

Just some comments, no on the accuracy of the minutes,
but continuing the discussion.

At 12:37 03/10/23 +0100, Richard Ishida wrote:

>Charset Declaration
>We discussed the first technique in section 3.2
>http://www.w3.org/International/geo/html-tech/#ri20030218.131104811
>which involved a discussion about transcoding by servers.
>We agreed in principle that we should recommend that people should NOT
>declare the encoding using the HTTP header, but rely instead on
>declarations inside the document.

I'm really not sure about this. There is nothing bad at all with
having the right HTTP header. Recommending that people do NOT
use it clearly sends the wrong message.


>This is because of the difficultly
>the average content developer

I think that talking about the 'average content developer'
is potentially dangerous. Most of us like to think that we and
the people around us are 'average'. Yet the Web is extremely
diverse. There are large companies that translate manuals
into dozens of languages, and on the other side, there are
many personal Web sites where nobody ever thinks about
translation. And there are many other examples.


>has in managing the server information and
>because it carries a stronger likelihood of being correct if the
>encoding is changed from time to time.

Well, I agree in certain scenarios. But in others (as an example
when editing in a plain text editor), the information in a file
is probably not very reliable, either.
There are also other scenarios, for example a company with
clearly established policies (all content in languages a, b, c
in encoding foo,...) that are implemented through the use
of appropriate tools,...


>However, where a document is
>transcoded the server should send the charset info in the HTTP header -
>this will override the in-document declaration.

Well, yes, but usually, these transcoders don't look inside the
document to find the encoding in the first place.


>There may be
>exceptional cases, such as Tim Bray's site, where he only uses server
>settings because he has complete control over his pages and environment.

This case is exceptional, but in some ways, it's the way of the future
(everything in UTF-8). And in some ways, every Web site is an exception.


>We need to get more feedback on the list and from other people about how
>prevalent transcoding is these days, and who does it.
>ACTION: Tex, look at Google's cache to see how that works vis a vis
>transcoding


Regards,    Martin.

Received on Thursday, 23 October 2003 16:17:08 UTC