Created attachment 178514 [details] Update to version 4.3 I noticed there was a new version of GNU sed and I was bored ... Pet portlint (add license), and remove some unused cruft while here.
Maintainer timed out.
Please also fix the two INSTALL_DATA commands so they are not muted, as pointed out by portlint -C when you commit.
Created attachment 179139 [details] Update to version 4.3 Fixed muted INSTALL_foo commands (from portlint)
(In reply to Xin LI from comment #2) I am not a ports committer so I should send this back to the pool.
I failed build on 10.3-i386, last lines of log : depbase=`echo sed/sed_sed-utils.o | sed 's|[^/]*$|.deps/&|;s|\.[^.]*$||'` && \ cc -std=gnu11 -DHAVE_CONFIG_H -I. -I. -I./lib -I./lib -I./sed -DLOCALEDIR=\"/usr/local/share/locale\" -DLIBICONV_PLUG -I/usr/local/include -O2 -pipe -DLIBICONV_PLUG -fstack-protector -fno-strict-aliasing -MT sed/sed_sed-utils.o -MD -MP -MF $depbase.Tpo -c -o sed/sed_sed-utils.o `test -f 'sed/utils.c' || echo './'`sed/utils.c; \ if test $? = 0; then mv -f $depbase.Tpo $depbase.Po; \ else rm -f $depbase.Tpo; false; fi rm -f lib/libsed.a ar cr lib/libsed.a lib/copy-acl.o lib/set-acl.o lib/acl-errno-valid.o lib/acl-internal.o lib/get-permissions.o lib/set-permissions.o lib/c-ctype.o lib/c-strcasecmp.o lib/c-strncasecmp.o lib/close-stream.o lib/closeout.o lib/dfa.o lib/localeinfo.o lib/dirname-lgpl.o lib/basename-lgpl.o lib/stripslash.o lib/exitfail.o lib/getprogname.o lib/hard-locale.o lib/localcharset.o lib/glthread/lock.o lib/malloca.o lib/progname.o lib/qcopy-acl.o lib/qset-acl.o lib/quotearg.o lib/se-context.o lib/se-selinux.o lib/tempname.o lib/glthread/threadlib.o lib/unistd.o lib/version-etc.o lib/version-etc-fsf.o lib/wctype-h.o lib/xmalloc.o lib/xalloc-die.o lib/acl_entries.o lib/canonicalize-lgpl.o lib/error.o lib/fpending.o lib/fwriting.o lib/getopt.o lib/getopt1.o lib/obstack.o lib/regex.o lib/secure_getenv.o lib/strverscmp.o ranlib lib/libsed.a rm -f sed/libver.a ar cr sed/libver.a sed/version.o ranlib sed/libver.a cc -std=gnu11 -O2 -pipe -DLIBICONV_PLUG -fstack-protector -fno-strict-aliasing -fstack-protector -o sed/sed sed/sed_sed-compile.o sed/sed_sed-execute.o sed/sed_sed-mbcs.o sed/sed_sed-regexp.o sed/sed_sed-sed.o sed/sed_sed-utils.o sed/libver.a lib/libsed.a /usr/local/lib/libintl.so -Wl,-rpath -Wl,/usr/local/lib /bin/mkdir -p doc rm -rf doc/sed.1 doc/sed.1-t ./build-aux/help2man \ --name 'stream editor for filtering and transforming text' \ -p sed --include ./doc/sed.x \ -o doc/sed.1-t sed/sed \ && chmod a-w doc/sed.1-t \ && mv doc/sed.1-t doc/sed.1 env: perl: No such file or directory gmake[3]: *** [Makefile:5775: doc/sed.1] Error 127 gmake[3]: Leaving directory '/wrkdirs/usr/ports/textproc/gsed/work/sed-4.3' gmake[2]: *** [Makefile:3024: all-recursive] Error 1 gmake[2]: Leaving directory '/wrkdirs/usr/ports/textproc/gsed/work/sed-4.3' gmake[1]: *** [Makefile:2156: all] Error 2 gmake[1]: Leaving directory '/wrkdirs/usr/ports/textproc/gsed/work/sed-4.3' ===> Compilation failed unexpectedly. Try to set MAKE_JOBS_UNSAFE=yes and rebuild before reporting the failure to the maintainer.
(In reply to Wen Heping from comment #5) Ugh ... add the port maintainer to the bug report. Maybe he is familiar with the issue.
Created attachment 180628 [details] Update to version 4.4 Thanks Pedro, I'll lift it up from here. Please see updated patch (to version 4.4). It is based on yours plus additional improvements and cleanups. I guess an exp-run is in order now so we can be sure it does not break any consumers.
- Why not use the test framework (TEST_TARGET=check-TESTS for instance) - Is is normal that most regression tests are now failing? they were all succeeding with previous version
(In reply to Antoine Brodin from comment #8) Funny, tests seem to run fine with bsd make but not with gmake
(In reply to Alexey Dokuchaev from comment #7) Thanks for picking it up, I really have no time for it. I am wondering why you need an exp-run: very few ports depend on gsed, and BSD sed has become much more compatible since FreeBSD 11. Poudriere should suffice.
Well I'm glad that I asked for an exp-run (my quick grepping revealed more than just a few ports, so I thought it would be a good idea): Antoine already pointed out two important issues. Regardless of whether exp-run will show any regressions, I'd like to fix those issues before the patch hits the tree (that is, employ TEST_TARGET and either switch to BSD make if that's sufficient or fix the test suite so it passes fine with GNU make).
any progress on the regression tests?
back to the team
batch change of PRs untouched in 2018 marked "in progress" back to open.
This has taken forever an upstream is already on version 4.5. I'll let someone else take a fresh look.
Created attachment 201237 [details] Update to version 4.7 OK, let's close this Gestalt. Update to the latest version 4.7, now all tests pass identically when built against GNU or BSD make (GNU make is not required though).
Reopen and request another exp-run.
On 11.2 amd64 there is 1 new failure: http://package22.nyi.freebsd.org/data/112amd64-default-PR215768/2019-01-19_19h43m54s/logs/errors/pgbadger-10.2.log
A commit references this bug: Author: danfe Date: Mon Jan 21 13:35:03 UTC 2019 New revision: 490862 URL: https://svnweb.freebsd.org/changeset/ports/490862 Log: Unbreak the build against upcoming GNU sed version 4.7 and actually fix the pgBadger.pod generation. Unlike GNU make, BSD make does not define $< variable for explicit rules, but earlier versions of GNU sed did not complain about missing filename in r/R/w/W commands. PR: 215768 (exp-run for gsed 4.7) Changes: head/databases/pgbadger/Makefile
Thanks Antoine. Once you believe we're good now, could you reassign PR to me so I can commit the update?
The only new failure was pgbadger
A commit references this bug: Author: danfe Date: Mon Jan 21 14:37:47 UTC 2019 New revision: 490864 URL: https://svnweb.freebsd.org/changeset/ports/490864 Log: Update GNU sed to version 4.7. PR: 215768 Submitted by: pfg Exp-run by: antoine Changes: head/textproc/gsed/Makefile head/textproc/gsed/distinfo head/textproc/gsed/pkg-plist
The port was updated as of ports r490864, thanks everyone involved and sorry it took so long!
(In reply to Alexey Dokuchaev from comment #23) Thank you for making this happen!