cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Crossley <>
Subject Re: [PROPOSAL] build-time XML validation via RELAX NG
Date Fri, 01 Nov 2002 04:46:32 GMT
Colin Paul Adams wrote:
> >>>>> "David" == David Crossley writes:
>     David> I notice that the RNG grammar is catering for the
>     David> deliberately broken lint/sitemap.xmap 
> Actually, it's not deliberately broken - the invalid bits are
> commented out.

Ah, i understand now.

> What I've been doing is then to uncomment them one-by-one to
> test the
> power of the grammar to reject invalid constructs.

That is a clever approach.

> Those that it fails
> to reject, I have added as comments to the grammar file,
> to provoke thoughts on what to tackle next.

Another way to assist this process would be to exclude
lint/sitemap.xmap from the "validate" target and have a
new target "validate-lint-sitemap" which only processes
the lint one. From time-to-time we could expose certain
difficult cases via CVS mods to lint/sitemap.xmap so that
more people can see the validation messages.

I followed some leads and found the old CVS history of
the lint sitemap. There are some useful CVS comments from
Vadim that might help.

> However, I finding it quite hard to see how to tackle these cases. I
> can't work out what the rules ought to be. They are certainly more
> complex than the comments against these invalid cases suggest.

Is this the best approach for clarification of all
sitemap constructs? Perhaps should we just start at the
top and carefully go through each component in turn,
discussing its parameters and context.

This would be a great opportunity to update the xdocs
and javadocs as we go. It may take longer, but the
people involved would learn a lot and Cocoon gets polish.


To unsubscribe, e-mail:
For additional commands, email:

View raw message