Page 5 of 5
Re: New Ai Ideas
Posted: 08 Jan 2009, 19:54
by sli
Tad offtopic, but is there anything like Python's doctest for Perl? What it does is scan source docstrings looking for test that looks like interactive interpreter sessions in the comments and converts them into unit tests (read: fucking awesome). This would let one test out the entire API quickly and easily.
Obviously a Perl one would be different since it doesn't use an interactive interpreter, but the idea is there.
Re: New Ai Ideas
Posted: 09 Jan 2009, 08:41
by isieo
kLabMouse wrote:
Almost none of those doc's have nothing to do with "AI 2008", they describe old design.
The new one, is that "Star Ship" model with some additions to it.
I'm thinking, about adjusting auto documentation, so there will be better description about API.
yes we need something equivalent to the old documentation.
Re: New Ai Ideas
Posted: 10 Jan 2009, 09:55
by kali
sli wrote:Tad offtopic, but is there anything like Python's doctest for Perl? What it does is scan source docstrings looking for test that looks like interactive interpreter sessions in the comments and converts them into unit tests (read: fucking awesome). This would let one test out the entire API quickly and easily.
Obviously a Perl one would be different since it doesn't use an interactive interpreter, but the idea is there.
You can run createdoc.pl from src/doc/
Code: Select all
#!/usr/bin/env perl
# Documentation extractor. Extracts documentation from comments from .pm files and put them in HTML files.
But all it does is extract comments, so it's not exactly the one you're looking for (just the doc, not the test part).
Re: New Ai Ideas
Posted: 10 Jan 2009, 21:14
by Technology
python's doctest for perl:
Test::Snippet
(also take a look at "see also")
Re: New Ai Ideas
Posted: 24 Apr 2013, 14:12
by Technology
See following sections in this pdf:
http://www.valvesoftware.com/publicatio ... _booth.pdf
"Moving through the Environment" -> looks like kore's current drunken walking.
"Path Optimization" -> "Following path directly looks robotic"
"Reactive Path Following" -> Looks like what we want kore to do. (+ the followed path is close to the one with path optimization, but not as much bot-like)
The "Intention subsystem" looks interesting as well.