W3C home > Mailing lists > Public > www-style@w3.org > March 1998

Re: text/css vs application/x-pointplus.

From: Chris Lilley <chris@w3.org>
Date: Sun, 01 Mar 1998 10:06:39 +0100
Message-ID: <34F9251F.FDC8B6BD@w3.org>
To: Sue Jordan <sjacct@worldnet.att.net>
CC: "'W3C Style List'" <www-style@w3.org>
Sue Jordan wrote:
> 
> Failure of some ISPs to support the mime type text/css has been the
> subject of several discussions in the stylesheets newsgroup.  My
> attempt to rectify this with my ISP has resulted in this answer:
> 
> "Unfortunately, that MIME extension is already mapped to something
> else in the standard MIME.types file that comes with our servers.

> Until IANA sorts out which extension goes with which application, 

bogus. 

IANA registers MIME types; that registration can include common filename
extensions, Apple 4-letter filetype codes and suchlike info but the
definitive thing is the MIME type (eg, text/css) not the extension
(.css, whatever). *That is the point*. Some servers use filename
extensions to generate the MIME type; some use Apple codes; some use a
database, whatever. 

There is no "standard" for MIME.types - this is simply the name that
some servers (which use filename extensions) call a file which does the
extension to MIME mapping. What they mean is "this is what came out of
the box" - default for that server product, rather than standard for the
Web.

> we
> will maintain our MIME.types file as close to the manufacturer's
> recommendation as we can. 

The manufacturer in this case being ....

> So, for now, .css files remain as
> application/x-pointplus..."

Any MIME type which has x- is experimental, unregistered. (And
unregistrable).

> I find that IANA shows registration for neither text/css nor
> application/x-pointplus. 
> Is this merely an oversight? 

It is not an oversight.

The registration for text/css is in process. Because IANA drasticaly
revised the registration procedure to essentially allow two categories:

a) application/vnd.manufacturername.whatever
   (easy to register)
b) all the rest
   (the 'IETF tree', hard to register, for IETF standards only) 

This made it a bit difficult to actually register text/css which has
been in use for three years without registration. It could have been
registered as something else but this would have hurt interoperability.

W3C has succesfully concluded discussions with IETF application area
directorate and registration of text/css is now proceeding; there is an
internet draft defining text/css which points to the CSS1
Recommendation, and this internet draft will become an Informational RFC
in due course and then text/css will be added to the IANA registry.

The x- prefix is a historical artefact - it used to be used for
unregistered experimental types. This usage is now deprecated by IETF.
Any type containing x- cannot, by definition, be registered.

> If this issue
> is already known, are we simply waiting for resolution of this
> apparent conflict? 

There is no actual conflict. There is only one text/css. It is
unfortunate that there are existing files that use *.css as filenames,
but with only three characters there are bound to be conflicts. How many
different file types are called *.doc ?

> Is there anything, short of changing ISPs, that we
> can do to rectify this?

Find out what server product they are using, lobby that manufacturer to
update their MIME.types file. Actually I would be interested to hear
what servers come with text/css support as standard.

Ensure that Apache has out of the box support for text/css; if not, get
it added. That should cover near 50% of the Web, after upgrades.

Do a web crawl to find out how many URLs return a resource labelled
application/x-pointplus and how many return one labelled text/css. I
suspect this will give a suitable statistic to claim that there are far
more text/css; it may give a statistic to show that
application/x-pointplus is unused. If so, ask them to delete that line
in MIME.types

Use another filename extension. Here at W3C we have both .stl and .css
mapped to text/css. This depends on platform and whether you are limited
to 8.3 format. Remember there is nothing magic about .css - you can use
.w3ccss or .foo .cascade or anything, as long as it gets served up with
the correct MIME type. However, many authoring products assume .css and
it is thus wise to stick to that if possible.

As you note, changing ISP (with a letter to the chair of the ISP company
explaining why you are changing and why you will encourage others to do
likewise) may have some effect, as a last resort.

--
Chris
Received on Sunday, 1 March 1998 04:13:24 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:53:54 GMT