Re: IPC::Run::time[r|out] vs our TAP tests

Michael Paquier <michael(at)paquier(dot)xyz> writes:
> On Thu, Apr 04, 2024 at 07:09:53PM -0400, Tom Lane wrote:
>> If you have time, that'd be great. What I suspect is that that
>> machine's readline isn't regurgitating the string verbatim but is
>> doing something fancy with backspaces or other control characters.

> I have been able to reproduce the failure manually and your patch is
> providing more information, indeed, as of:
> [10:21:44.017](0.002s) ok 83 - \r works
> [10:24:45.462](181.445s) # pump_until: timeout expired when searching
> for "(?^::\{\?VERBOSITY} )" with stream: "\echo :{?VERB^G^Mpostgres=#
> \echo :\{\?VERBOSITY\} "
> psql query timed out at
> /home/pgbuildfarm/git/postgres/src/bin/psql/../../../src/test/perl/PostgreSQL/Test/BackgroundPsql.pm
> line 281.

> This stuff is actually kind of funny on this host, "\echo :{?VERB\t"
> completes to something incorrect, as of:
> postgres=# \echo :\{\?VERBOSITY\}

Just to be clear: you see the extra backslashes if you try this
tab-completion manually?

> Attaching the log file, for reference. Now I can see that this uses
> libedit at 3.1-20181209, which is far from recent. I'd be OK to just
> remove libedit from the build to remove this noise, still I am
> wondering if 927332b95e77 got what it was trying to achieve actually
> right. Thoughts?

It kind of looks like a libedit bug, but maybe we should dig more
deeply. I felt itchy about 927332b95e77 removing '{' from the
WORD_BREAKS set, and wondered exactly how that would change readline's
behavior. But even if that somehow accounts for the extra backslash
before '{', it's not clear how it could lead to '?' and '}' also
getting backslashed.

regards, tom lane