[uf-discuss] hAtom draft

[David Janes -- BlogMatrix]

Ryan King wrote:
> On Dec 4, 2005, at 9:15 AM, David Janes -- BlogMatrix wrote:
>> Ryan King wrote:
>>> 4. Why do we prefer <h#> over class="title" for entry titles?
>>
>> See my earlier note. I'd really appreciate if you or Tantek got back 
>> to me here: my understanding is that we'd always prefer appropriate 
>> XHTML constructs.
> 
> Yes, I'd say we should prefer the appropriate html construct.
> 
> In this particular case, though, I'm afraid using <h#> is a bit brittle- 
> this is coming from helping triage support requests coming into 
> Technorati about us not indexing their blog properly. For this 
> particular element I would prefer:
> 
> 1. an explicit classname (most people are using a classname already, no?)
> 2. fallback to <h#>
> 
> I think the explicit declaration should be preferred, but this is just a 
> suggestion. I know that other xhtml-syndication efforts have used <h#> 
> for entry titles, but I'm not sure of their success. Anyone with 
> experience here, please speak up.

I'm going to go with your suggestion. I've actually been doing lots of 
playing with parsing Microformats using Python, DOMs, and so forth and 
I'm getting a better sense of what practically works.

> 
>>> 5. "Entry Permalinks MUST be absolute URIs". Why? We have well 
>>> established rules for relative urls.
>>
>> I could lower this to SHOULD; feedback would be appreciated.
> 
> I think requiring absolute URIs is a bit too high a hurdle, not not 
> quite neccessary.

I'm going to change this to SHOULD. There, done.

> 
>> However, what I'm trying to accomplish is to let "rel-bookmark" 
>> provide byte comparable strings for providing "the best location for 
>> this resource".
> 
> Like I said, the rules for transforming relative URIs to absolute ones 
> are pretty well established, so any consumer should be able to take care 
> of this for themselves. I think this is just a case where we need to 
> optimize for the publisher over the consumer.

I was reading a blog post yesterday about how much misery atom:base and 
relative URIs are causing. Can't find it, ah well.

> 
>> The problem with relative URIs is that readers at 
>> "http://instapundit.com" and at "http://www.instapundit.com" will come 
>> up with two different sets of Entry Permalinks that are actually 
>> representing the same resources.
>>
>> This even gets uglier with LiveJournal. I do recognize this may be an 
>> attempt at some mild social engineering on my part.
> 
> FWIW, there has been some (offline and on-) discussion about a 
> rel-canonical microformat. Maybe hAtom should defer this problem (*it 
> is* bigger than just atom/blogs).

Fair enough. Maybe it'll be a role model.

> 
>>> 6. quote:
>>>> there can be at most 1 Entry in an XHTML document without an Entry 
>>>> Permalink; the Entry Permalink of this Entry is the URI of the page
>>>> This rule is needed for media pages (i.e. a news article on 
>>>> cnn.com). There is some ugliness of with this because the URI could 
>>>> be non-canonical."
>>> I'm not sure I follow this and don't see anything on the 
>>> brainstorming page about it.
>>
>> It's in the blog-post-examples [1]. I'd like to make in practical for 
>> organizations such as CNN to markup pages such as [2] in hAtom without 
>> requiring them rewriting the way they do pages.
> 
> So the use-case is a "document with one entry"? Is this really worth 
> making a general rule about?
> 
>> ...
>> It's all great -- bring it on. I'm back in fighting shape :-)
> 
> Great.

A few more changes have gone in. I've documented a list [1] for people 
tracking the proposal. I've also started collecting practical advice on 
templates, CSS and so forth [2]. Contributions from WP people and so 
forth would be appreciated.

> 
> -ryan
> 

Regards, etc...
David
http://www.blogmatrix.com

[1] http://microformats.org/wiki/hatom#Recent_Changes
[2] http://microformats.org/wiki/hatom#Hints_and_Tips