Apparent improper handling of SIGPIPE

[p6rt]

Migrated from rt.perl.org#131608 (status was 'open')

Searchable as RT131608$

[p6rt]

From frettled@gmail.com

According to a brief mail discussion on perl6-users (https://www.nntp.perl.org/group/perl.perl6.users/2017/06/msg3905.html and subsequent messages), Rakudo seems to improperly handle piping with e.g. head, and therefore most likely any other kind of external program that would close a pipe.

For the sake of bug record keeping, here's the relevant snippet of the discussion.

On Fri, Jun 16, 2017 at 12​:13 PM, Elizabeth Mattijsen wrote​:

Failed to write bytes to filehandle​: Broken pipe in block \ at -e line 1

So it would appear this got fixed by the synchronous IO refactor, and will be available in the 2017.06 release, which is scheduled for tomorrow.

This appears to be only a partial fix with an unexpected outcome.

awk, Perl 5, Ruby, etc. do not generate errors for similar use of pipes, and instead use exit status 0.

PS​: I have not installed the latest Rakudo myself, this just caught my interest, and I was asked to file a rakudobug about it.

-- Jan

[p6rt]

From @Leont

On Tue, 20 Jun 2017 06​:12​:54 -0700, jani wrote​:

According to a brief mail discussion on perl6-users (https://www.nntp.perl.org/group/perl.perl6.users/2017/06/msg3905.html and subsequent messages), Rakudo seems to improperly handle piping with e.g. head, and therefore most likely any other kind of external program that would close a pipe.

For the sake of bug record keeping, here's the relevant snippet of the discussion.

The problem with the unix default SIGPIPE handling is that while it's very useful for command-line applications, it's most annoying for pretty much anything else (in particular networking applications).

Regardless of the default, this needs an easy to use switch to toggle the behavior.

Leon

[p6rt]

The RT System itself - Status changed from 'new' to 'open'