Re: [Fwd: MobileOK Validator Issues]

Le lundi 10 mars 2008 à 11:27 -0400, Sean Owen a écrit :
> There is a slightly larger problem here, that we would have a problem
> on any DTD that we don't have a copy of, and there could be many. In
> this case we should not fail, as we do now. I can change that.

FWIW, I remember that some people were not happy with the idea of
downloading unknown DTDs from the Web; I personally would much prefer
this to the current situation where we simply abort when encountering
this. Maybe this should be a configurable option?

I think ideally, we would do as follow:
 * if we know the SYSTEM ID and have it in cache, we use the cached
version
 * if we don't know the SYSTEM ID, but there is a PUBLIC ID that matches
a well-known SYSTEM ID in cache, we use the cached version
 * if we don't know the SYSTEM ID, there is no PUBLIC ID or we can't
relate it to a known SYSTEM ID, we either download the DTD or FAIL
(depending on the configuration option)

> And then I just remove the unused HTML 4 DTDs from the checker.

Makes sense; still, the question remains: how do we deal with documents
in non-XML versions of HTML? We can't validate them with our existing
infrastructure (and from what I've heard, I don't think there is a good
Java SGML Validator available).

The problem is: if we get a valid HTML 4.01 document, at this time we
would still say it fails on CONTENT_FORMAT_SUPPORT-4.

(theoretically speaking, it is doable to create an HTML 4.01 document
that is mobileOK, but the checker would not admit it as of today).

I can add a test case if that helps.

Dom

Received on Tuesday, 11 March 2008 17:02:50 UTC