Re: [Bug] XHTML Strict Minimal document

Bjoern Hoehrmann wrote:

> * "Karl Dubost" <karl@w3.org> wrote:
> | At 16:39 +0200 24/10/00, Bjoern Hoehrmann wrote:
> | >No. This does not validate against one of the 3 DTDs as required above. A
> | >title element is required in the DTD. This is not expressed in the
> | >recommendation text, but thats not important, conformance to the DTD is
> | >required.
> |
> | Ok thanks ;-) Bjoern for the clarification. It's why I think we
> | should complete the validator with help messages.
> 
> Agreed. I think we should think about how to implement these error messages.
> Liam Quinn patched SP [1] to generate more helpful error messages for his
> validator; I'm currently translating these error messages into german. Terje
> said several times that he has to do something about these error messages to.
> Maybe this is the right time to discuss this issue.

Ok. I've been thinking about this for a while and there are AFAICT 3
options:

1. Trap SP's messages and print our own.
  - Horrible kludge! SP's error messages follow no known pattern and
    would require some truly awfull regex matching with few if any
    possibilities of optimizations.

2. Use Liam's hacked copy of SP with it's improved error messages.
  - Works well, but leaves us with a divergent copy of SP which makes
    it harder to upgrade. OTOH SP doesn't get upgraded that often.

3. Use OpenSP[0] and trap it's output to add our own error messages.
  - This is more work, and requires switching parser, but is the most
    elegant solution IMO. OpenSP outputs message _numbers_[2] which
    should be easy to trap and look up our customized error messages.
    This can be implemented with a minimal amount of change to "check".

I've been meaning to look into option #3 because it can be done with a
minor patch if OpenSP works the way I think it works[3], but I've been
putting it off while I'm working on some more elaborate changes[4]. I'm
pretty swamped right now, but I'll try to make time to look into it this
weekend. If no unforseen problems crop up it should be possible have
working code in very little time. Adding decent error messages would be
what takes time, but could OTOH be more easily delegated (Björn and Karl
just volunteered ;D) to cluefull users.

Switching to lq-nsgmls (Liam's version) should be about as quick if
anyone wants to look into that (Gerald, weren't you already considering
this?). My concerns with this approach is that SP has to be recompiled
every time you want to change a message and only one language can be
used at a time (AFAIK, Liam?), plus it leaves us with a divergen version
of SP. These may not be insurmountable problems and gives us the benefit
of a pre-made solution (least amount of work).


[0] - A sub-project of the OpenJade project. IIRC they are the new
"official"
      maintainers of SP as James seems to be trying to pass the torch.
      <URL:http://openjade.sourceforge.net/>

[1] - NMF

[2] - At least, I /think/ that's what the docs say, but I haven't tested
it.
      It wouldn't be the first time I've missed what SP's docs /really/
say.

[3] - See [2]. :-)

[4] - The so far entirely vapourware support for HTML Templates, i18n,
XML
      and text output, and ";imgonly". If I don't have beta-quality code
      available for testing the next time I mention this project in
public,
      twack me over the head....

      Hard! :-(

Received on Tuesday, 24 October 2000 13:22:38 UTC