Hi, I wrote a script that parses our semi converted rules.t file from pugs to parrot.
This script can grok all the constructs in use right now, and emit
equivalent perl 5 code.
Given another few hours of work I can make the perl emission code
into a Test::Base backend to retain portability.
The tests themselves originated from perl 5's test suite.
The reason I'm bringing this up is that now that I've gone over the
file to iron out the rough edges, I don't think this test is worth
it. The reasons for this are:
a) the tests are approximate and lossy - the notion of failure
is unclear in the test that check that something doesn't match
b) it's hard to convert linearized $1, $2, $3 etc into the
multidimensional format, and although i've fixed many of these
manually i'm sure there are more in there. This gets very nasty
when these are backreferences inside the matching part of the
pattern.
c) not all the tests were converted, and the script written to
convert them is now lost, afaict
I'd rather spend my time extracting useful test cases, that really
demonstrate missing features or problems in PGE, and rewriting them
in a clearer, more accurate manner.
Then I think it's wise to use something akin to QuickCheck to
generate patterns and input strings, because it seems like that is
what most of the perl 5 test suite is. If we do this from scratch we
still have the advantage of being more accurate, and covering more
aspects of PGE's design.
Please let me know what you'd like me to do next
--
() Yuval Kogman <[EMAIL PROTECTED]> 0xEBD27418 perl hacker &
/\ kung foo master: /me does not drink tibetian laxative tea: neeyah!
pgpBVP4MbOu93.pgp
Description: PGP signature
