atoms (was Re: [Negotiation] How to get rid of a Content-Encoding: header?)

[Dean Gaudet <dg...@arctic.org>]

On Sun, 29 Aug 1999, Life is hard, and then you die. wrote:

> No, that header is generated by mod_mime. What happens is that mod_mime
> goes through the extensions and sets the content_type, content_encoding,
> and content_language fields in the request_rec. Because it doesn't set
> the headers directly, you can't unset them with a Header directive.

One of the reasons for having them in the request_rec is because
manipulating them in the headers table is not as fast.  We've talked about
a solution to this before -- using "atoms"... which lets us make the
tables faster by mapping strings such as "Content-Encoding" to integers at
compile-time, so that lookups/etc. are faster at run-time. 

This would help many of the headers, but doesn't help the multi-valued
headers... such as "Vary"...

at any rate, yeah it'd be cool if there were fewer special purpose fields
in request_rec, and we used tables for more things... 

Dean

Re: atoms (was Re: [Negotiation] How to get rid of a Content-Encoding: header?)

[Rodent of Unusual Size <Ke...@Golux.Com>]

Manoj Kasichainula wrote:
> 
> Ben Hyde (I'm pretty sure I've got the right Ben this time!) pointed
> me to some work he did a while back at
> <http://zap.ne.mediaone.net/p/p/symbols/>. Is this good enough?

This symbolisation might also make easier the idea of a
registry of error message text, which could ease I18N significantly.
-- 
#ken    P-)}

Ken Coar                    <http://Web.Golux.Com/coar/>
Apache Software Foundation  <http://www.apache.org/>
"Apache Server for Dummies" <http://ASFD.MeepZor.Com/>

Re: atoms (was Re: [Negotiation] How to get rid of a Content-Encoding: header?)

[Manoj Kasichainula <ma...@io.com>]

On Tue, Sep 07, 1999 at 03:47:13PM -0700, Dean Gaudet wrote:
> We've talked about a solution to this before -- using "atoms"...
> which lets us make the tables faster by mapping strings such as
> "Content-Encoding" to integers at compile-time, so that lookups/etc.
> are faster at run-time. 

Ben Hyde (I'm pretty sure I've got the right Ben this time!) pointed
me to some work he did a while back at
<http://zap.ne.mediaone.net/p/p/symbols/>. Is this good enough?

-- 
Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/

Re: atoms (was Re: [Negotiation] How to get rid of a Content-Encoding: header?)

[Manoj Kasichainula <ma...@io.com>]

On Wed, Sep 08, 1999 at 02:42:45PM -0700, Greg Stein wrote:
> Manoj Kasichainula wrote:
> > What's an "ADT"?
> 
> Abstract Data Type.

Ahh. I just haven't seen the term abbreviated. After high school, we
just called them "objects".

-- 
Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/

Re: atoms (was Re: [Negotiation] How to get rid of a Content-Encoding: header?)

[Greg Stein <gs...@lyra.org>]

Manoj Kasichainula wrote:
> 
> On Tue, Sep 07, 1999 at 03:55:54PM -0700, Roy T. Fielding wrote:
> > Yes, I'd be really happy if all the header manipulation stuff was
> > replaced with an efficient ADT
> 
> What's an "ADT"?

Abstract Data Type.

Think "encapsulation" or "modularity".

Cheers,
-g

--
Greg Stein, http://www.lyra.org/

Re: atoms (was Re: [Negotiation] How to get rid of a Content-Encoding: header?)

[Manoj Kasichainula <ma...@io.com>]

On Tue, Sep 07, 1999 at 03:55:54PM -0700, Roy T. Fielding wrote:
> Yes, I'd be really happy if all the header manipulation stuff was
> replaced with an efficient ADT

What's an "ADT"?

-- 
Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/