Michael G Schwern <schw...@pobox.com> writes: > Steffen Schwigon wrote: >> Somewhat related to Andy Lesters problem some days ago about the TAP >> of "prove --verbose" should not be touched. >> >> Is it possible to make the output "idempotent", i.e., that I can take >> the output of "prove --verbose" and parse it again with same results? >> >> In particular the finally added "ok" line should not be printed, or >> only with --normalize. > > ... > >> Skipping the "ok" line lets it PASS again and again. >> I need it when I archive TAP output and re-evaluate it later. > > Trying to parse the mix of TAP and prove output is never going to work right. > You're risking that none of the prove output will be valid TAP. > Even if that "ok" line is removed something else will mess it up > later.
Canonically you are right. On the other side I wouldn't be so pessimistic about prove's output. It's *the* tool that knows best about TAP. Why shouldn't it be clever enough to not generate TAP itself. It already nearly is. And the latest --verbose/--normalize discussion already targets a similar goal. Anyway, I like the later idea of using --archive, maybe with some finetuning. I now go back and meditate about it. Thanks. Kind regards, Steffen -- Steffen Schwigon <s...@renormalist.net>
