[uf-discuss] HTTP profile header proposal

[Al Gilman]

At 12:20 PM -0400 6/23/06, Chris Casciano wrote:
>On Jun 23, 2006, at 11:58 AM, Scott Reynen wrote:
>
>>Via Danny Ayers' blog [1], Mark Nottingham's proposal to add 
>>profiles to HTTP headers seems very relevant to microformats:
>>
>>http://www.ietf.org/internet-drafts/draft-nottingham-http-link-header-00.txt
>>
>>It would allow for HEAD requests to identify pages containing 
>>microformats without downloading the full document.
>>
>>Peace,
>>Scott
>>
>>[1] http://dannyayers.com/2006/06/23/return-of-the-http
>>
>
>the question remains with all profile discussion -- does their lack 
>of accuracy overcome their usefulness
>
>how does a blog package (as an example) know if there is a hreview 
>or hcalendar data in one of the individual blog postd displayed on a 
>given page to accurately represent what profiles are to be used / 
>sent in the headers? Failing the ability to do just that, do you 
>ignore profile inclusion or just include all profiles all the time?

You send it all the time.

At least my impression at present is that if there is an @profile in
the html:head then the author thought about it and cares about it.
It's not that popular yet to be much mis-used. So it's worth sharing.
Once it succeeds, of course, the road to bloat is clear.  Thank God
for Moore's Law.

I think your problem is one of precision, not accuracy. Accuracy
problems are what we have had trouble with concerning Content-Type:
not being in line with the actual data in the entity body / resource
representation. Here the header only gives a rough summary of what
you want to know in more detail.

Besides, the way this works when it works is that when you are
looking at a blog page, the author is either using or not using the
markup practices that come with the account, and you have already
processed all those metadata pages and you know whether this profile
works for you or not by opaque URI-matching in the @profile string.
Practice is much more 'lumpy' than the language we create to describe
it. It tends to fall into a few recognizable patterns, and you will
rapidly figure out which work for you.

Once you decide that your filter expression for profiles is going
to be heuristics, not science, there are reasonable results to be had.

In the end, would you rather be fetching a corpus of pages where
5% of them have stuff you can use, or 0.05%?

[I speak as a fool.]

Al

>
>sorry for not adding anything new to the discussion, just felt the 
>desire to re-present the question.
>
>--
>[ Chris Casciano ]
>[ chris at placenamehere.com ] [ http://placenamehere.com ]
>
>_______________________________________________
>microformats-discuss mailing list
>microformats-discuss at microformats.org
>http://microformats.org/mailman/listinfo/microformats-discuss