JamisBuck said:
I completely agree with Andargor... in fact, that's what I was ambitiously trying to do before I "fell off the face of the Earth."

The treasure generator is more or less there, although it uses a non-standard file format. The NPC generator, town generator, and dungeon generator, though, could all benefit from a complete rewrite.
Yeah, I took a look at the Basilisk project (within which is the treasure generator), and I have to admit that I don't feel as comfortable updating it. Mainly because I would have to learn how the thing works, and the data files are proprietary.
I would be willing to parse the PCGen files for you and output the files in your format, since the relevant info is there (and my parse-fu is strong

). But to ensure it works properly, etc., I would leave that up to you (sorry).
JamisBuck said:
Perhaps a viable option would be to just use PCGen's data files... once they move to XML, that will be a more plausible solution, but (and Andargor and I have talked about this briefly) it would be nicer if new classes/spells/items/skills/etc. could be completely represented in the data, instead of requiring portions of the program itself to be modified. *sigh*
I have to disagree with using the PCGen data files when they are in XML.
With all due (and well deserved) respect to the PCGen people, I have looked into this as well, but have concluded that they would be of limited use. Having worked extensively with PCGen data files, and helping them update to 3.5, I have consistently been frustrated with one fact: PCGen blurs the line between data and code.
Basically, if there is a "tag" change, the Java and LST code has to be modified. That is why there is a new version every two weeks and is hard to follow, and old data files need to be recoded, and why 3.5 is such a pain to redo (yes, redo from scratch). Heck, in 3-6 months, the NPC generator will have to be updated again to reflect LST changes for export purposes.
So in essence, LST is code that people other than Code Monkeys can change. It's a custom scripting language that changes often.
And the XML format they are working on does not change this, since they are just embedding LST "code" into tags.
But I am working with others on a generic way to represent the data sets that is independent of the tool or engine used. The concept is to transform this immuable, never changing data (except for, say, errata) to fit the needs of tools using such things as XSLT, XPath and XQuery. An XML Database, really. (See the "Suite Interoperability" thread here, and the d20-XML group for those discussions, there is even a nice diagram that I made to illustrate this)
This would be what is required to ensure that another recode of the tools is not necessary for a good long time.
JamisBuck said:
If I keep at this, I'll talk myself into working on it. Guess I better stop.
Keep talking!
Andargor