Bogofilter-1.1.0 - new current release

[Matthias Andree]

David Relson <relson at osagesoftware.com> writes:

> On Sun, 13 Aug 2006 15:33:21 +0200
> Matthias Andree wrote:
>
>> On Sun, 13 Aug 2006, Matthias Andree wrote:
>> 
>> > > FWIW, I've got /usr/bin/onsgmls installed (from OpenSP-1.5-8mdk),
>> > > not nsgmls.  Any tips?
>> > 
>> > grep -v ac_cv_path_NSGMLS config.cache >cc.$$ ; mv cc.$$
>> > config.cache ./config.status --recheck && ./config.status
>> > make check
>> 
>> This is to say: configure figures out if you've got OpenSP (onsgmls)
>> or the older nsgmls - if you still get complaints, remove the
>> offending line from config.cache and re-run configure so it figures
>> the right one.
>
> Hi Matthias,
>
> ./configure detected /usr/bin/onsgmls and doc/Makefile is using it.  A
> bit of experimentation with the contents of /etc/sgml/catalog shows
> that results of "make check" vary considerably depending on what's
> in it.

The distributions' reliability in creating a usable catalog for SGML or
XML validation is, to put it mildly, suboptimal.

I find myself de- and reinstalling RPMs on SUSE Linux regularly because
packages had been installed, but not properly recorded in the
catalog. I'm not sure if SUSE's YaST installs packages in the wrong
order to reduce the amount of CD changes for CD users so that the
catalog maintenance utility comes to late, if they install
with --noscripts or equivalent to defeat the registrations, I haven't
checked that - but I am not aware of other ways to spoil the catalog.

> Here's what I _had_ (a day or so back):
>
>    CATALOG "/etc/sgml/dsssl.cat"
>    CATALOG "/etc/sgml/sgml-docbook-3.0.cat"
>    CATALOG "/etc/sgml/sgml-docbook-3.1.cat"
>    CATALOG "/etc/sgml/sgml-docbook-4.1.cat"
>    CATALOG "/etc/sgml/xml-docbook-4.1.2.cat"
>    CATALOG "/etc/sgml/xml-docbook-4.2.cat"
>    CATALOG "/etc/sgml/xml-docbook-4.3.cat"
>
> Using this, "make check" generates:
>
>    *** Not validating HTML, nsgmls/HTML DTDs improperly installed ***
>
> Removing the sgml-docbook lines gives the desired results:

That's interesting. Can you pull out the nsgmls/HTML smoke test from
doc/Makefile.am and obtain its output with sgml-docbook*cat installed
and listed? Perhaps there's some implicit fallback, but I'm not sure
about that, I know too little about the OpenSP internals. Docbook
catalogs shouldn't be interfering with HTML, perhaps onsgmls on your
system is confused because it lists some SGML catalogs, but not HTML?

So, what happens if you remove and reinstall the HTML DTD RPM? Does that
cause additional lines to show up in your catalog?

Here's my catalog, for reference:

$ cat /etc/sgml/catalog
CATALOG "/usr/share/sgml/CATALOG.opensp"
CATALOG "/usr/share/sgml/CATALOG.html"
CATALOG "/usr/share/sgml/CATALOG.html-3.2"
CATALOG "/usr/share/sgml/CATALOG.html-4.0"
CATALOG "/usr/share/sgml/CATALOG.html-4.01"
CATALOG "/usr/share/sgml/CATALOG.iso_ent"
CATALOG "/usr/share/sgml/CATALOG.mathml-2.0"
CATALOG "/usr/share/sgml/CATALOG.svg-1.0"
CATALOG "/usr/share/sgml/CATALOG.svg-1.1"
CATALOG "/usr/share/sgml/CATALOG.xhtml-1"
CATALOG "/usr/share/sgml/CATALOG.xhtml-basic10"
CATALOG "/usr/share/sgml/CATALOG.xhtml-1-modularization"
CATALOG "/usr/share/sgml/CATALOG.xhtml-1.1"
CATALOG "/usr/share/sgml/CATALOG.xmlcharent"
CATALOG "/usr/share/sgml/CATALOG.openjade"
CATALOG "/usr/share/sgml/CATALOG.tei_4"
CATALOG "/usr/share/sgml/CATALOG.tei_4sgml"
CATALOG "/usr/share/sgml/CATALOG.docbook_3"
CATALOG "/usr/share/sgml/CATALOG.docbook-xml-slides"
CATALOG "/usr/share/sgml/CATALOG.docbook-xml-website"
CATALOG "/usr/share/sgml/CATALOG.docbook-simple-1.0"
CATALOG "/opt/kde3/share/apps/ksgmltools2/customization/CATALOG.kdelibs3"
CATALOG "/usr/share/sgml/CATALOG.docbook_4"
CATALOG "/usr/share/sgml/CATALOG.db41xml"
CATALOG "/usr/share/sgml/CATALOG.db42xml"
CATALOG "/usr/share/sgml/CATALOG.db43xml"
CATALOG "/usr/share/sgml/CATALOG.db44xml"
CATALOG "/usr/share/sgml/CATALOG.docbook-dsssl-stylesheets"

With this detail information:

$ cat /usr/share/sgml/CATALOG.html-4.01
OVERRIDE YES

PUBLIC "-//W3C//DTD HTML 4.01//EN" /usr/share/sgml/html/dtd/4.01/strict.dtd
PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" /usr/share/sgml/html/dtd/4.01/loose.dtd
PUBLIC "-//W3C//DTD HTML 4.01 Frameset//EN" /usr/share/sgml/html/dtd/4.01/frameset.dtd
PUBLIC "-//W3C//ENTITIES Latin1//EN//HTML" /usr/share/sgml/html/dtd/4.01/HTMLlat1.ent
PUBLIC "-//W3C//ENTITIES Special//EN//HTML" /usr/share/sgml/html/dtd/4.01/HTMLspecial.ent
PUBLIC "-//W3C//ENTITIES Symbols//EN//HTML" /usr/share/sgml/html/dtd/4.01/HTMLsymbol.ent
PUBLIC "-//W3C//ENTITIES Latin 1//EN//HTML" "/usr/share/sgml/html/dtd/4.01/HTMLlat1.ent"
DTDDECL "-//W3C//DTD HTML 4.01//EN" "/usr/share/sgml/html/dtd/4.01/HTML4.decl"
DTDDECL "-//W3C//DTD HTML 4.01 Transitional//EN" "/usr/share/sgml/html/dtd/4.01/HTML4.decl"
DTDDECL "-//W3C//DTD HTML 4.01 Frameset//EN" "/usr/share/sgml/html/dtd/4.01/HTML4.decl"

> A minimal configuration (just first line; neither sgml-docbook nor
> xml-docbook lines) also produces the desired results.

In doubt, you can pass the catalogs in an environment variable, too. I
don't know its exact name off-hand though.

-- 
Matthias Andree