Re: [OpenJade-devel] Re: Any C++ programmers around? (was: Unix --> NT (source code stuff))

On 17.03.01 at 05:28, Adam Di Carlo <> wrote:

>>We're currently futzing around with SGML_CATALOG_FILES [...]
>Well, if you build with the right settings, you don't need to do this.

The CGI frontend gets downloaded and used with locally built versions of
SP. I can't control how people choose to build it -- especially since some
build it for other uses and not exclusively for use with the CGI frontend!
-- without jumping through some fairly major hoops.

>>SP_ENCODING are "magical" in that they are necessary to enable XML mode.
>Hmm.  I find that XML mode works pretty well without those.  But maybe
>I haven't looked deeply enough.  Can someone enlighten me.

According to the docs from the jclark days you _must_ set SP_CHARSET_FIXED
to "YES" (or Boolean "True", I suspect) and SP_ENCODING to "UTF-8" to
enable XML mode.

>Oh, I pointed out in my last email that DTDDECL in your SGML open
>catalogs will fix this. We ship a lot of DTDs in Debian this way (but
>I had to hack Jade for Debian so Jade wouldn't bitch about the Jade
>non-supported directive).

What, Debian doesn't use OpenJade? :-)

>Perhaps I should give an example:
>PUBLIC  "-//W3C//DTD XHTML 1.0 Strict//EN"        "xhtml1-strict.dtd"
>DTDDECL "-//W3C//DTD XHTML 1.0 Strict//EN"        "xhtml1.dcl"

Oh, cool! I didn't know you could do that. This might obviate the need for
a command line switch. You wouldn't happen to support Namespace Delegation
in Catalogs too? :-)

BTW, why doesn't the "url" Storage Manager try to use a relative URI for
resolving System Identifiers like the "osfile" Storage Manager?

>>>> * Blue Sky: A Perl (XS) Module Interface
>Yes, I agree it would be pretty easy and quite nice.  Even just
>getting the entity mgmt function from libsp (or libosp) would be
>pretty sweet.

"libosp.xs" Wanted! Payment in Beer and Undying Gratitude! :-)

>>>> * Blue Sky: Configurable Error Format
>>Well, the reason I'm so gung ho on switching to OpenSP is that it has a
>>switch "-n" that outputs message numbers ("relevant clauses") with error
>>messages: [...]
>Doesn't that solve this wishlist item, then?

It solves most of _my_ problems, certainly -- :-) -- but I was thinking
more along the lines of giving a printf()-ish format string for what the
output should look like. On second thought it may not be all that

Oh, wait, now I remember! I wanted it because sometimes osngmls outputs a
different number of "fields" instead of leaving empty the meaningless (in
context) or unavailable fields. Makes it a PITA to write a generic parser
for the error output. Lesse, I think it's when some kind of non-SGML
related error occurs; since there is no line/column number these fields are
omitted from the output. Probably a couple of other cases too.

BTW, there is no distinction between an SGML error and an application error
in this output. Is this intentional? File Not Found could be argued to be

>I use 'onsgmls -gues file' and that tells me containing element.

Oooo... Nice! :-)

That may cover it.

>I think I saw that email come in but I haven't had time to look yet.

If you get the time I'd dearly like to know what's going on with that.

>FYI, did anyone but me notice that XHTML 1.0 is not clean according to
>onsgmls -Wall?  'xhtml1-strict.dtd' contains bogus stuff like an
>entity 'FrameTarget' which is defined but never used.

As a wild guess it's there to reserve the name for the future.

>I emailed the address given for comments, but have gotten no feedback.

Try (public list).

>I wish the spec committes would at least try to validate their DTDs
>fully before publishing the spec...

Actually, they probably validated...

...using <URL:>; so since it doesn't set -Wall
   they wouldn't have noticed either. :-(

Now Playing: "Things I've Seen" - Spooks

Received on Saturday, 17 March 2001 11:50:51 UTC