Re: sgml-validate

[Allan Gottlieb <gottlieb@nyu.edu>]

On Sat, Feb 12 2011, William F. Hammond wrote:

>>> When I tried M-x sgml-validate, it offers to run
>>> nsgmls -s myfile.html
>>>
>>> I don't have nsgmls so looked around and found
>>> opensp providing onsgmls and a gentoo bug report claimed
>>> a symlink nsgmls -->  onsgmls.
>>>
>>> Emerged opensp and now have onsgmls.
>>>
>>> But onsgmls -s myfile.html
>>> just hangs and uses no CPU time.
>>>
>>> onsgmls claims it will validate "the SGML document whose document entity is
>>> specified by the system identifiers SYSID..." so perhaps my error is in
>>> supplying just a file name.  What is needed for a valid SYSID?
>
>> AFAIU that's rather an sgml-issue.
>>
>> Seems the command onsgmls fails, which should happen too for the very
>> reasons also from the command line outside Emacs.
>>
>> BTW I use `xmllint' for validation, may be it's ok for your purposes too.
>
> onsgmls is a very reliable tool.  In general, however, it requires
> complicated command lines.

Agreed.  I have given up since I was not able to figure out how to have
it "not validate" and xmllint has that ability.  The validation adds
minutes to the test so needs to be used sparingly.

> AIUI "xmllint" works only on xml files, and, therefore, probably won't
> work with most extant html files, even those self-identified as
> the xml form of html, which commonly are not "well-formed xml".
> (Xml well-formedness is easily and quickly checked using James Clark's
> "xmlwf" that is included with "expat" distributions.)

Thank you very much for this tip.  I ran xmlwf and my html passed!  This
was made possible in large part by sgml-xml-mode.  I believe xmllint
also validates, but xmlwf has the plus of speed and not commenting about
entities (see below).

> If your html is dependably well-formed xml, then take your questions
> to comp.text.xml.  Otherwise maybe take them to
> comp.infosystems.www.authoring.html unless you really, really want a
> full understanding of the sgml background for html in which case ask
> in the rather quiet group comp.text.sgml.

Actually I am now doing OK thanks to the help I received here.
The only annoyance is that xmllint keeps telling me that
≥ et al are not valid entities.

> Furthermore, be aware that the proposal for future html5 served as
> "text/html" falls outside of all classical sgml/xml validation
> paradigms.

Thanks for the warning.

allan