Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-07

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:00 idiosyncrat_ joined #marpa
00:32 lwa joined #marpa
01:03 lwa joined #marpa
01:27 ronsavage joined #marpa
01:51 ronsavage joined #marpa
02:26 idiosyncrat_ A new version of Marpa::R3 is ready on the trial branch of the Github repo
02:27 idiosyncrat_ It include some build refactorings, including the use of ExtUtils::Install for copying libmarpa, and a few other cleanups
02:27 idiosyncrat_ It tests OK on my box.
02:28 idiosyncrat_ Checking for regression under Windows or elsewhere is appreciated!
02:58 lwa joined #marpa
03:49 rns idiosyncrat: re http://irclog.perlgeek.de/m​arpa/2016-04-07#i_12298624
03:49 rns cygwin: make releng ok, a nit: it started asking
03:49 rns gzip: Marpa-R3-1.001_000.tar.gz already exists; do you wish to overwrite (y or n)?
03:49 rns if if Marpa-R3-1.001_000.tar.gz already exists in the cpan dir.
03:50 idiosyncrat_ Yes. we need to fix that
03:50 rns windows:
03:50 rns "C:\perl-5.20.1\bin\perl.exe" -MExtUtils::Command -e rm_rf -- engine/perl_ac_build
03:50 rns "C:\perl-5.20.1\bin\perl.exe" -MExtUtils::Install -e  'install([ from_to => {@ARGV}, skip => q{engine/cf/INSTALL.SKIP}, dir_mode => '\''755'\'' ]);'  -- engine/read_only engine/perl_ac_build
03:50 rns The system cannot find the path specified.
03:50 rns NMAKE : fatal error U1077: '"C:\perl-5.20.1\bin\perl.exe"' : return code '0x1' Stop.
03:50 idiosyncrat_ I was overriding the COMPRESS macro to use "gzip --best -f" ...
03:50 rns ok
03:51 idiosyncrat_ but decided that's not portable as it overrides the native EU::MM facilities for detecting the right COMPRESS
03:52 idiosyncrat_ re the "The system cannot find the path specified" ...
03:52 rns re windows: I noted that engine/perl_ac_build isn't in fact created and added $(MKPATH) engine/perl_ac_build before $(LIBMARPA_INSTALL) but that didn't help
03:52 idiosyncrat_ Is the problem the hard-coded slashes in the file name for INSTALL.SKIP?
03:54 idiosyncrat_ And re "gzip: Marpa-R3-1.001_000.tar.gz already exists; do you wish to overwrite (y or n)?" -- do we want to consider it a feature that it refuse to overwrite an existing Marpa-R3-1.001_000.tar.gz?
03:55 idiosyncrat_ Apparenlty the EU::MM authors do consider this a feature.
03:55 ronsavage joined #marpa
03:56 rns re INSTALL.SKIP: I must confess that I ran make releng just after https://github.com/jeffreykegler/Marpa--R3/co​mmit/1c8b216ffa7783ded7aaebbf1c9a5efaaa3b7e95 when there was no INSTALL.SKIP and had the same results.
03:57 rns Perhaps it's just something wrong with my (very dated) windows xp sp3, perhaps Jean-Damien's tests will shed more light on this.
03:57 idiosyncrat_ OK.  Let me know.
03:59 rns Well, I hope jddurand will be abe able to run make releng on the recent trial branch and telel the results here.
04:00 rns re gzip: If it's a feature, I'm fine with it.
04:10 rns re windows changed single quotes to double in LIBMARPA_INSTALL, i.e.
04:10 rns "install([ from_to => {@ARGV}, skip => q{engine/cf/INSTALL.SKIP}, dir_mode => '$(PERM_DIR)' ]);" --
04:11 rns and it progressed to
04:11 rns "C:\perl-5.20.1\bin\perl.exe" -MExtUtils::Command -e chmod -- 755 libmarpa.lib Cannot write stamp-h1:Permission denied at C:/perl-5.20.1/lib/ExtUtils/Command.pm line 184.
04:11 rns after         "C:\perl-5.20.1\bin\perl.exe" -MExtUtils::Command -e rm_rf -- libmarpa.lib         lib -ltcg -out:libmarpa.lib marpa.obj marpa_ami.obj marpa_avl.obj marpa_codes.obj m arpa_obs.obj marpa_tavl.obj
04:12 ronsavage joined #marpa
04:17 rns Ok, with
04:17 rns cd engine/perl_ac_build && $(MAKE) && $(TOUCH) stamp-h1
04:17 rns changed to
04:17 rns cd engine/perl_ac_build && $(MAKE)
04:17 rns $(RM_RF) engine/perl_ac_build/stamp-h1
04:17 rns $(TOUCH) engine/perl_ac_build/stamp-h1
04:18 rns it runs all tests ok.
04:32 jdurand joined #marpa
04:34 jdurand hello - make releng seems to loop on my box, compiling, testing, compiling, testing and so on
04:34 jdurand this is on the trial branch
04:34 jdurand and on debian gnu linux
04:35 jdurand Ah... not exactly, it compiled/tested at least three times - and finally returned
04:35 rns yep, it tests several builds.
04:36 rns idiosyncrat: http://irclog.perlgeek.de/m​arpa/2016-04-07#i_12298807 and http://irclog.perlgeek.de/m​arpa/2016-04-07#i_12298822 are in https://github.com/jeffreykegler/Marpa--R3/pull/6
04:37 rns fixed winfows build for me.
04:37 jdurand rns: ok - so it seems to ran ok - Jeffrey I hope pastebin will like that I paste there the full output
04:37 idiosyncrat_ OK, good, because I've been having that problem on my box and it's costing me time :-)
04:38 idiosyncrat_ rns: Thanks
04:38 shadowpaste0 "jdurand" at 217.168.150.38 pasted "make releng Marpa--R3 trial branch" (2531 lines) at http://fpaste.scsys.co.uk/509747
04:39 rns :)) Absolutely welcome, hope it runs ok under unixes.
04:52 jdurand Re gzip: Marpa-R3-1.001_000.tar.gz already exists confirmed on a second round
04:53 shadowpaste0 "jdurand" at 217.168.150.38 pasted "2nd round with MARPA_USE_PERL_AUTOCONF=1 make releng Marpa--R3 trial branch on debian" (2045 lines) at http://fpaste.scsys.co.uk/509748
04:57 ronsavage left #marpa
05:07 rns jdurand: Cool. BTW, make release seems to be sufficient -- it runs full_test which runs perl_ac_disttest which uses MARPA_USE_PERL_AUTOCONF=1 -- https://github.com/jeffreykegler/M​arpa--R3/blob/master/Makefile#L62
05:08 rns so is make releng isn't broken it's ok overall
05:08 rns s/so is/so if/
05:30 lwa joined #marpa
05:32 idiosyncrat_ Good night!
05:49 lucs x
07:20 maybekoo2 joined #marpa
10:55 kaare_ joined #marpa
12:52 maybekoo2 joined #marpa
13:35 Cheery joined #marpa
16:18 idiosyncrat_ joined #marpa
17:31 idiosyncrat_ I have just pushed a new version of the 'trial' branch.
17:32 idiosyncrat_ It turns out there are special macros for customizing the contents of Makefile's in EU::MM, targeted at things like portably doing "cd xyz && $(MAKE)" and portably doing Perl one-liners
17:33 idiosyncrat_ I changed over to these, and have tested them on my Debian setup, but haven't tested that they actually work as intended under Windows.
17:33 idiosyncrat_ Your help is appreciated!
17:35 idiosyncrat_ rns: re http://irclog.perlgeek.de/m​arpa/2016-04-05#i_12290893
17:36 idiosyncrat_ I've converted the show_ahms() NAIF grammar method and show_earley_sets(), show_and_nodes(), show_or_nodes() NAIF recognizer methods ...
17:36 idiosyncrat_ to have SLIF grammar and SLIF recognizer versions.
17:37 idiosyncrat_ I did a lot of refactoring to move logic down to the THIF, so it could be shared.
17:38 idiosyncrat_ I tested that that refactoring does not break the NAIF calls, but have *not* tested the SLIF calls -- I hope a new sl_jirotka.t and a new sl_catalan.t will do that.
17:38 idiosyncrat_ Thanks!
18:02 rns idiosyncrat: Great, thanks, I'll give sl_jirotka.t and sl_catalan.t a try this weekend. WIll let you know.
18:38 jdurand joined #marpa
18:39 jdurand Re http://irclog.perlgeek.de/m​arpa/2016-04-07#i_12302097 - make releng ok on Debian as well, except the "gzip: Marpa-R3-1.001_000.tar.gz already exists; do you wish to overwrite (y or n)?" - is it possible to add a .PHONY target to remove it systematically before the creation?
18:40 jdurand I have also a virtualized solaris box running perl-5.10 if I remember well - which minmum perl version will require Marpa::R3 ?
18:44 jdurand AFK - will backlog
18:47 idiosyncrat_ Marpa::R3 is already set to only work with Perl 5.10.1 or greater.
18:47 idiosyncrat_ It won't support 5.10.0
18:48 idiosyncrat_ re http://irclog.perlgeek.de/m​arpa/2016-04-07#i_12302378
18:48 idiosyncrat_ rns also noticed this, as did I and I initially considered it a bug to be fixed ...
18:50 idiosyncrat_ but my thinking is that EU::MM has made the decision to refuse to silently overwrite a distribution file, and I should follow EU::MM's choice.
18:51 idiosyncrat_ The only question is whether I should override this in the 'releng' target, and here again, I wonder -- yes, the question is usually a nuisance, but might it not be a lifesave at some point?  Perhaps in development?
18:52 idiosyncrat_ The choices in EU::MM do embody *a lot* of experience.
20:55 koo7 joined #marpa
22:49 idiosyncrat_ joined #marpa

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary