Lists Home |
Date Index |
email@example.com (Paul Prescod) writes:
>I would most often use attributes for columns of numbers. But yes, I
>can see this situation if you are using elements.
I prefer to use elements for content and attribute for metadata. I find
the arguments for "use attributes when you need unordered content"
unpersuasive and generally bogus, even damaging to markup practice.
>The issue is not that they are bothered. The issue is that the
>irrelevance of the order must now be documented and they must read the
>documentation to know what is going on. This also holds for everyone
>writing software that will process the information. They must also keep
>in mind that the ordering is under user control _and yet meaningless_.
>Otherwise they will fall into traps of depending on the ordering they
>saw in some particular input document or of ascribing meaning to the
>ordering -- even though that ordering could be messed up by an
In my cases - entering information from maps created in forests by a lot
of different people, or building forms to support data entry where
efficient measurement means changing the order of information on every
individual tree measured, I think the benefits far outweight the costs.
"Under human control _and yet meaningless_" is only a problem if you
assume that everything humans control must have meaning. Interleave
seems like a perfect means of combining the need for information with
the contingencies involved in collecting it from the world. The only
time an intermediary process can botch the order is if you ascribed the
order meaning. Interleave seems to make it clear that there is no
meaning, and documentation is a normal part of good markup process, so
Imposing an arbitrary order seems to me to give far too much credit to a
universe dominated by punch cards where every bit of order mattered. "I
am a human being. Please do not spindle, fold, or staple me." Or
Ring around the content, a pocket full of brackets
Errors, errors, all fall down!
http://simonstl.com -- http://monasticxml.org