Re: sources of kawa documentation

["Hoehle, Joerg-Cyril" <Joerg-Cyril dot Hoehle at t-systems dot com>]

Hi,

Per Bothner answered:
> print-as-xml is gone.  It was quite buggy.
> Instead we now have the --output-format option, which is
> both nice and documented.
Hmm, a command line option is not what I was looking for. I'll try to explain:

I thought of using BRL, XML-generation, HTML generation, HTTP query and form response, DB query, possibly Common Lisp all from one instance of a JVM. All of that is in Kawa (except DB?) so it seems natural to use that. I'd have Scheme glue code combine the whole stuff and make it work.

I don't understand how command lines would help me with that.
My idea:
o have some BRL to generate HTML, using Scheme functions defined elsewhere in a Scheme file.
o have some HTML generated without BRL
o Have separate code define HTTP action handlers, output HTML via one of the above methods.
o Have some Scheme code generate XML - or have a BRL for XML? I don't need XML as response to a HTTP query, I wish to put XML into a database (how weird these days are).


> Well, there is some documentation in this chapter:
> http://www.gnu.org/software/kawa/XML-tools.html

I knew about it, but with "Functions for generating HTTP responses", I did not see how to output that stuff to something. It says:
"response-header key value
   Create the response header key: value in the HTTP response.
 [...] when you actually "print" its value to the response output stream. 
Where is that HTTP response object and associated output stream that's being talked about?

>  It should be possible to implement
> print-as-xml's functionality using the --output-format framework.
> In fact I'm trying to come up with how best to do chains of
> output filters (i.e. combining consumers).
Nice, but hopefully as chains inside Scheme, not as separate UNIX programs?
Indeed, I like flow-based programming!

> > Other areas of interest to me: consumers concept,
> Well, it's an implementation-level issue.  I.e. it's not
> really supposed to be visible at the Scheme level.
I hope it will. In Common Lisp, I'm missing the flexibility that I have in Python of defining arbitrary classes that implement parts of the stream protocol, e.g. only write(), read(), read-line() etc. as needed. The Gray Streams proposal doesn't catch it IMHO, and the more recent simple stream proposal may, but I did not understand the low-level device issues.

Do I confuse things here?


> Well. this is implied, though I suggest:  (invoke x 'getClass).
> This is just a special case of invoke, as getClass is a standard
> Java method.
Ok, I remember reading the "name mangling" section.
I know little about Java and its thousands of classes.

BTW, I'll try to proof-read kawa.texinfo and edit typos as time allows.
Hmm, didn't ispell provide an easy checker for TeX(info?) at some time?

Regards,
	Jorg Hohle.