Re: Errors in MathML DTD

On Sat, 19 Apr 2003, David Carlisle wrote:

> > Nope.  Fetching the MathML DTD + modules by HTTP is too big an overhead
> > and will slow the server unacceptably.
> 
> OK (I thought that the files might be available locally to the validator
> machine from the cvs repository sitting behind the w3.org web site,
> but if you say not, I believe you)

I don't know if they could be accessed without HTTP (eg by NFS), but
even if they can it's network traffic (not to mention the problems
of dealing with any NFS problems gracefully from a web POV).

We're also dealing with more than one validator.  validator.w3.org
is at MIT; qa-dev.w3.org at Keio, and I'm of course concerned
with valet.webthing.com in the UK - which has a much faster validator
than the W3C one but no privileged connectivity to w3.org,
making the issue of local copies more critical!

> > It would be more helpful if you could notify us whenever the DTDs are
> > updated.
> 
> I can certainly do that.

Thanks!

> > I'm wondering what would be the implication of
> > setting up a cron job to fetch the whole lot weekly or something?
> It wouldn't do any harm, although normally this is a relatively rare
> event, there were a few changes recently as part of the process of
> preparing mathml 2,0 2nd edition we went back over the mathml comments
> list archive and reviewed all the comments on the dtd.

OK.  Round tuits permitting, I may well hack something up for this.

-- 
Nick Kew

Received on Saturday, 19 April 2003 15:04:24 UTC