GRBs retractions - Param or value?

[Alasdair Allan]

Roy Williams wrote:
> Most interesting to astronomers, I think, is documentation of what  
> a parameter actually means, and we have that right from the start  
> with the Description tag. While some technical people are most  
> interested in an Xschema description, I think this is of less  
> interest to astronomers, and would in reality be a big roadblock to  
> entry.

I think we're in the classic computer science vs. astronomy argument  
that has dogged the VO throughout it's entire life. Most of the  
people who are going to have to sit down and implement the creation  
(and parsing of) VOEvent messages aren't computer scientists or  
"real" programmers (defined by for the sake of argument as someone  
who, when they've written a recursive routine actually knows they've  
done it).

For the most part they're astronomers who will view the messages as  
text, plain ASCII text. If 1 in 10 end users actually use an XML  
parser to rip a VOEvent message apart I'll be really surprised. Even  
those astronomers who will use an XML parser to create or parse the  
message will, again for the most part, not be comfortable with  
schema. I'd lay money on the fact that the only astronomers to verify  
the messages against the schema will be those whose parsers do it for  
them behind their back and don't tell them. Having to create entirely  
new schema every time you want to add the equivalent of a <Param> to  
a message would be seen a serious stumbling block. In reality what  
would happen is that people would start making tags up without  
creating the schema to go with them and you'd end up in a situation  
far, far, far, worse than having <Param> tags with UCDs attached to  
them. Although admittedly we're in the situation even right now where  
I'm to "make up" a UCD-like entity to tag my <Param> with, at least  
that's better than making up an entire element like <tau_max>, the  
consequences of which I hate to think about. Don't fool yourselves,  
it'd happen. At least at the moment you can verify the message  
against the schema, I doubt you'd be able to do that a couple of  
years down the line if we went down the alternative route. Need I  
mention FITS headers? Let alone the ESO "extensions" to FITS headers?

Al.