Re: Design Principles

Jonas Sicking On 09-05-27 01.49:
> On Tue, May 26, 2009 at 2:56 PM, Leif Halvard Silli <lhs@malform.no> wrote:
>   
>> Kornel On 09-05-26 16.06:
>>     
>>> On 26 May 2009, at 12:18, Anne van Kesteren wrote:
>>>       
>>>>> 2.4. Pave the Cowpaths
>>>>>    = this to me also supports building on existing profile related
>>>>> authoring practises such as microformats. Or is it only those
>>>>> microformatters that do /not/ use @profile that represent a cowpath?
>>>>>           
>>>> While microformats claim to need a profile attribute in practice they do
>>>> not use it I believe for consuming etc.
>>>>         
>> The microformats points to  /HTML 4/ which says that one can establish
>> profile page URIs and  link to them via @profile. The HTML 4 spec here
>> presents /more/ than a mere attribute specification, it presents a "sub
>> specification system". Where is the "claim" in this?
>>
>> Do you suggest that authors establish profile pages for describing meta data
>> profiles, but that actually /using/ these URIs in order to inform what
>> conventions that are being followed  should be prohibited? This seems very
>> contrary to what the Web is about.
>>
>> So the question is still /what/ in this is it that represents a cow path?
>>   * HTML 4 /has/ a method for defining meta data profiles:  A single
>>     web page that represents the profile. Do we need to change that
>>     cowpath?
>>   * HTML 4 [snip] {uses a} URI, the most common cowpath of all, for
>>     pointing to the profile that is being used. Do we need another
>>     cowpath than a URI?
>>   * When you describe how some _User Agents_ do not use the URI for
>>     anything, then I think you are stretching the cowpath concept,
>>     cowpaths do not pertain to what User Agents do.
>>     
>
> I believe that the term 'cow path' refers to something that has a lot
> of usage. I.e. something that people use or do a lot.
>
> So I definitely agree that the fact that UAs don't use the profile
> attribute does not make it a non-cow-path.
>
> The question is instead, do pages use the profile attribute.
>   

That is only one of the questions (see above).  Also note that the 
design principle talks about "consider cowpaths instead of inventing 
something new". It doesn't say "consider if something is a cowpath, and 
if it isn't, then consider dropping the feature".


>>> Indeed. When developing hCard validator (conformance checker) I've noticed
>>> that few pages use profile attribute, and there are no hCard processors that
>>> use it, so I haven't made that a requirement. I'm even tempted to remove all
>>> warnings about it.
>>>       
> [snip]
>   
>>> I've also found pages that use wrong profile — most often XFN. As far as I
>>> understand that would mean that hCards should not be processed on such page,
>>> and profile support in that case would be harmful.
>>>       
>
> Based on this very small data sample it does not seem like @profile is
> heavily used for hCard pages. Thus, if this data is representative,
> there would be no cow path.
>   
There are also many web pages that doesn't use a doctype, btw.

There is the cowpath that different microformat 
authors/specifiers/editors uses the profile mechanism by defining 
profile pages - aka "authoritative page or URI that represents the 
specification, and possible also tell how to transform it".

When I read about microformats, I do get the impression that the 
profiles concept was one thing that helped justify the effort.

> We can also guess, based on the fact that it seems to make
> little-to-no difference if the profile attribute is present, that many
> people don't put the profile in, either out of forgetting, not
> knowing, or simply not caring. However this is just a guess and actual
> data is of course much better.
>   
Data: There aren't, I think, many books on microformats. But the one 
very well known doesn't mention @profile a single time, as far as I can 
tell.
-- 
leif halvard silli

Received on Wednesday, 27 May 2009 01:54:07 UTC