Re: [platform-testers] groff 1.23.0.rc3 on GNU/Hurd and NetBSD

[G. Branden Robinson]

Hi Bruno,

At 2023-03-06T01:54:44+0100, Bruno Haible wrote:
> On GNU/Hurd and NetBSD 9.0, the build succeeds,

Excellent news!

My conntections to FSF France's NetBSD host have timed out for me every
time I've tried for the past few weeks.

> and there is 1 test failure:

They're different in each case.  Looking at the Hurd first:

FAIL: src/roff/groff/tests/initialization_is_quiet.sh
=====================================================

testing "-m cs"; COMPAT=""
FAIL src/roff/groff/tests/initialization_is_quiet.sh (exit status: 1)

No output.  Hard to say.  (I have only myself to blame.)

Here's the relevant piece of the test script.

set -e

# Regression-test Savannah #60874.
#
# groff should start up in any supported locale, in compatibility mode
# or not, without producing diagnostics.

# Keep preconv from being run.
unset GROFF_ENCODING

for COMPAT in "" -C
do
  for LOCALE in cs de en fr it ja sv zh
  do
    echo "testing \"-m $LOCALE\"; COMPAT=\"$COMPAT\""
    OUTPUT=$("$groff" -ww -m $LOCALE "$COMPAT" < /dev/null 2>&1)
    echo "$OUTPUT"
    test -z "$OUTPUT"
  done
done

My testing (and shell variable naming) philosophy has changed since I
wrote the above; I'd probably take out the "set -e" but still force the
test to fail if unsetting $GROFF_ENCODING failed (even if it is null,
its presence forces preconv(1) to run; this is intentional, I think--no
value simulates the argumentless groff -k option, otherwise the
argumentful -K option).

It's hard to say what is failing.  groff itself, echo(1), or test(1).

NetBSD has the same failure.

FAIL: src/roff/groff/tests/initialization_is_quiet.sh
=====================================================

testing "-m cs"; COMPAT=""
FAIL src/roff/groff/tests/initialization_is_quiet.sh (exit status: 1)

Again, shell access to a Hurd or NetBSD environment (or a canned QEMU
disk image of either or both I can run, since fortunately they're Free
Software) could be helpful.

If that's not possible I can send you a modified version of this test
script, and if that fixes the problem (unlikely) or communicates more
information, I will commit the change.

I also see a wart in
src/devices/grotty/tests/basic_latin_glyphs_map_correctly.sh that I can
fix.

printf "\\ " >&2

Thank you for uncovering these problems with the rigor of my tests.

Regards,
Branden

signature.asc
Description: PGP signature