On 11/21/05, jerry gay <[EMAIL PROTECTED]> wrote: > i've checked in a subset of Perl 5.9.2's regexp tests for PGE to chew > on. for now, i modified the stolen harness to emit PIR. the harness is > currently very ugly... that won't be for long, however, as i'll > refactor it soon. > it's been refactored, and is much less ugly now. the refactoring has allowed me to expand on the ability to parse perl match result vars like '$&', '$-[4]', '$1', etc. and that has allowed me to include even more of the 960 perl 5 tests, in fact, 800 of them. the final 160 are skipped by the harness because there are too many of them which cause parrot to crash, so i'll wait until the bugs are fixed before enabling those tests.
of the 800 tests the harness cares about, i decided to skip 290 tests; those which are designed to produce errors, expose perl bugs, those with syntax PGE doesn't understand (e.g. trailing modifiers,) and those which cause parrot to devour system resources just as vonnegut's Ice-Nine enveloped the earth. failing tests (there are approximately 155 of them) are currently marked todo, to prevent noise in parrot's test harness from PGE::P5Regexp, which is far from complete. i'm much happier seeing unexpected success messages rather than expected failures during development, so this should be the norm in the test suite. subtracting the skipped and todo tests from the total leaves about 360 passing tests from perl 5's test suite. thanks to patrick's work on PGE and P5Regexp, we now have a perl 5 regular expression parser capable of handling things like 'a:[b]:' =~ /([[:]+)/ nice work, patrick. ~jerry
