At 10:36 AM +0100 11/21/03, Willem Grooters wrote: >Two things: >1. What do you mean by "an excellent working Perl"? I did have tests >above 99% as well, but if chown and chmod don't work where I need them, >it's of no use at all.
You are correct that if you see a test failure in a feature you are going to depend on, then it's not the overall score that matters. But let's look at the two test failures Saku is reporting. One has to do with a timezone calculation when doing a stat() while one of the vmsish pragmas is in effect and daylight savings is not in force. Yes, there's something broken and I hope a fix will be forthcoming soon (not necessarily from me). The other failure is a non-reproducible one that has to do with sending a HUP signal to a subprocess. To have only these 2 issues out of about 75,000 tests on a platform that hasn't received extensive testing recently sounds like "an excellent working Perl" to me. >2. IF you _know_ which test fail on OpenVMS, mention them and, if >possible, tell why it fails (like the one you mentioned here). >Either fix the error of leave it out (and tell what features are missing >(and why)). Some tests do tell "not tested on platform" so it IS a >possibility. Yes, we do these things all the time. There are some tests that are skipped because they can't be expected to pass on a particular platform/configuration, and others marked TODO because there is a known bug that has yet to be addressed. I'm not at all suggesting that we sweep test failures under the rug. Saku's build reports confirm the timezone/stat() bug and also let us know that things are in pretty good shape on VAX except for the D_FLOAT default (or tests being unfriendly to D_FLOAT limitations). More reports are always welcome, and of course more fixes are always welcome. -- ________________________________________ Craig A. Berry mailto:[EMAIL PROTECTED] "... getting out of a sonnet is much more difficult than getting in." Brad Leithauser
