p5-shell-parser: sed: RE error: illegal byte sequence

[amadeus24]

Last lines of build:

:debug:extract chowned /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_perl_p5-shell-parser/p5.12-shell-parser/work to macports
:debug:extract euid/egid changed to: 503/501
:debug:extract Privilege de-escalation not attempted as not running as root.
:debug:patch patch phase started at Tue Nov 20 20:43:34 CET 2012
:debug:patch Executing org.macports.patch (p5.12-shell-parser)
:debug:patch Privilege de-escalation not attempted as not running as root.
:debug:configure configure phase started at Tue Nov 20 20:43:34 CET 2012
:notice:configure --->  Configuring p5.12-shell-parser
:debug:configure Using compiler 'Mac OS X clang'
:debug:configure Executing org.macports.configure (p5.12-shell-parser)
:debug:configure Environment: CPATH='/opt/local/include' CFLAGS='-pipe -O2' CPPFLAGS='-I/opt/local/include' CXXFLAGS='-pipe -O2' LIBRARY_PATH='/opt/local/lib' MACOSX_DEPLOYMENT_TARGET='10.8' CXX='/usr/bin/clang++' CC_PRINT_OPTIONS_FILE='/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_perl_p5-shell-parser/p5.12-shell-parser/work/.CC_PRINT_OPTIONS' F90FLAGS='-pipe -O2' LDFLAGS='-L/opt/local/lib' FCFLAGS='-pipe -O2' OBJC='/usr/bin/clang' INSTALL='/usr/bin/install -c' PERL_AUTOINSTALL='--skipdeps' OBJCFLAGS='-pipe -O2' FFLAGS='-pipe -O2' CC_PRINT_OPTIONS='YES' CC='/usr/bin/clang'
:debug:configure Assembled command: 'cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_perl_p5-shell-parser/p5.12-shell-parser/work/Shell-Parser-0.04" && /opt/local/bin/perl5.12 Makefile.PL INSTALLDIRS=vendor CC="/usr/bin/clang" LD="/usr/bin/clang"'
:debug:configure Executing command line:  cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_perl_p5-shell-parser/p5.12-shell-parser/work/Shell-Parser-0.04" && /opt/local/bin/perl5.12 Makefile.PL INSTALLDIRS=vendor CC="/usr/bin/clang" LD="/usr/bin/clang" 
:info:configure Checking if your kit is complete...
:info:configure Looks good
:info:configure Writing Makefile for Shell::Parser
:debug:configure Executing proc-post-org.macports.configure-configure-0
:info:configure sed: RE error: illegal byte sequence
:info:configure Command failed: find /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_tarballs_ports_perl_p5-shell-parser/p5.12-shell-parser/work/Shell-Parser-0.04 -name Makefile -type f -print0 | xargs -0 /usr/bin/sed -i "" '/^CCFLAGS *=/s/$/ /' ;
:info:configure Exit code: 1
:error:configure org.macports.configure for port p5.12-shell-parser returned: command execution failed
:debug:configure Error code: NONE
:debug:configure Backtrace: command execution failed
    while executing
"$post $targetname"
:info:configure Warning: targets not executed for p5.12-shell-parser: org.macports.activate org.macports.configure org.macports.build org.macports.destroot org.macports.install
:error:configure Failed to install p5.12-shell-parser
:debug:configure Registry error: p5.12-shell-parser not registered as installed & active.
    invoked from within
"registry_active ${subport}"
    invoked from within
"$workername eval registry_active \${subport}"

[ryandesign (Ryan Carsten Schmidt)]

A common error seen on Mountain Lion, fixed by setting LANG to C at the appropriate point. In this case, the appropriate point will have to be the post-configure procedure in the perl5 portgroup.

[jmroot (Joshua Root)]

Firstly, it should be LC_CTYPE, not LANG; and secondly, setting it to C will produce incorrect results unless the non-UTF8 characters are confined to comments. You need to use the actual character encoding of the file (or convert it from that to UTF-8 before running sed).

[ryandesign (Ryan Carsten Schmidt)]

Replying to jmr@…:

Firstly, it should be LC_CTYPE, not LANG;

Oops, I didn't realize that. As these issues have come up, I've been setting LANG in the ports. This has worked, but LANG sets several environment variables in addition to LC_CTYPE. Not sure what adverse implications that might be having.

and secondly, setting it to C will produce incorrect results unless the non-UTF8 characters are confined to comments. You need to use the actual character encoding of the file (or convert it from that to UTF-8 before running sed).

​Jeremy Huddleston Sequoia said using the C locale was fine, and that's what we've been doing for months already, since the release of Mountain Lion, in dozens of ports. If this was wrong, then there's a bit of cleanup to be done. Perhaps a recipe in PortfileRecipes would be good too, in either case.

Note that in the case of this specific ticket, we're talking about making a change to the perl5 portgroup, which does not know what encoding the file being changed is using. Or are you proposing that, instead of changing the perl5 portgroup, p5-shell-parser should change the file's encoding before the portgroup's code runs? That could be done. Looks like p5-shell-parser's Makefile is just ISO-8859-1. Maybe we want a proc in base for converting encodings so we don't have to call iconv with system manually.

[jeremyhu (Jeremy Huddleston Sequoia)]

Replying to ryandesign@…:

Replying to jmr@…:

and secondly, setting it to C will produce incorrect results unless the non-UTF8 characters are confined to comments. You need to use the actual character encoding of the file (or convert it from that to UTF-8 before running sed).

​Jeremy Huddleston Sequoia said using the C locale was fine, and that's what we've been doing for months already, since the release of Mountain Lion, in dozens of ports. If this was wrong, then there's a bit of cleanup to be done. Perhaps a recipe in PortfileRecipes would be good too, in either case.

If the reinplace is actually matching/changing what is written in that other locale, then it can cause a problem. If you want to just skip over it and continue as sed has done in the past, then LC_CTYPE=C is ok. The best solution would be to fix the file's encoding to UTF-8 and submit the change upstream, but that's not always a possibility as some files contain multiple encodings in embedded byte-streams on purpose (and thus you need to use the C LC_TYPE to deal with them).

[cooljeanius (Eric Gallager)]

Cc Me!

[danielluke (Daniel J. Luke)]

this should be resolved now (r123725)

[danielluke (Daniel J. Luke)]

Replying to dluke@…:

this should be resolved now (r123725)

oops, I mean r123724 :)