xml-dev - Re: Lark

[Peter@ursus.demon.co.uk (Peter Murray-Rust)]

Tim,
	Thanks very much for the latest Lark.  I have run it on a 
medium-sized file 20Kb and a few hundred nodes and it performs fine.

Some very minor comments for distribution:

(a) It would be really useful to have it as a package - all you have to
do is add 
package lark;

at the head of each file.  This means that the compiled classes can be located 
in standard libraries, etc.  At present I have
/myclasslib/pmr/sgml/*.class
/myclasslib/NXP/*.class
/myclasslib/NXP/Catalog/*.class
etc.
and it would be valuable to have
/myclasslib/lark/*.class

Secondly it means that it's easier to distribute classes in a robust fashion.
If there is a clear API then developers can subclass rather than hack the 
code - this is what I'd like to aim towards myself, so I'm happy to treat
lark and NXP as black boxen.  So, at the least, this could be done for 
Lark and Namer.

The problem in packages come when:
	there is some internal that people want to access.  This results from
		an insufficiently developed API
	there is some complex dependency between classes.  If you have
		A importing B
		B importing A
		then something is probably wrong.  (It's also difficult
		to compile unless you do them simultaneously.
I have about 10 packages in JUMBO, which took some sorting out.  I believe
that they have to be arranged as a DAG - I'm sure there is years of theory
about this.  Wherever I had trouble forcing them into a DAG it revealed itself
as a design fault :-)

(b) It still doesn't like the valid construction (prod. [32])
<!DOCTYPE CML>

it requires ExternalID.  [Unfortunately if I create a file like 
<!DOCTYPE CML SYSTEM "cml.dtd">
 and give it to NXP, NXP insists on *validating it* :-)]

-------------------
The unpacked files have ^M at record ends (this isn't a problem for me) and
some are missing and EOL and EOF.  Again not a problem.
Also it might be helpful if the files were packaged under a directory
such as V088/ (giving V088/Lark.class) so that when unpacked there was no
confusion between versions.

	P.

-- 
Peter Murray-Rust, domestic net connection
Virtual School of Molecular Sciences
http://www.vsms.nottingham.ac.uk/

xml-dev: A list for W3C XML Developers
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To unsubscribe, send to majordomo@ic.ac.uk the following message;
unsubscribe xml-dev
List coordinator, Henry Rzepa (rzepa@ic.ac.uk)