[uf-discuss] xdmp profiles not enough for parsing?

[Tantek Ç elik]

On 11/16/05 8:24 AM, "Dr. Ernie Prabhakar" <drernie at opendarwin.org> wrote:

> Hi Tantek,
> 
> On Nov 16, 2005, at 7:57 AM, Tantek Çelik wrote:
>> The conclusion was, in practice, complete automatic generic
>> parsability is
>> futile, and thus not worth pursuing, in XMDP or any other schema like
>> language.
> 
> I'll buy that, but that doesn't mean that we can't do *better*, right?

Actually, by trying to do better, we might do worse.  Especially when it's
an endless pursuit.

Mark's famously quoted email about the Atom process explained it quite well.
I'm sure the folks who want to go from 80% to 81%, then 82% etc. just wanted
to "do better".


> I like David's suggestion that we try to identify what additional
> semantics _might_ make it *easier* to write the current generation of
> parsers.   It may not be completely general, but that doesn't mean it
> wouldn't be useful, right?

These additional semantics are already noted in the particular
specifications.

The problem is that such a "might" list is never-ending.


> I agree that we'll probably never have enough for a completely
> general automatic parser, but hopefully we can at least make it
> easier on human parser-writers...

But how much easier?  Is it worth the effort?

Rather than wasting time incrementally making something more generic, why
not spend the time coding specific parsers?  I bet you'll get more done in
less time that way.

Tantek