Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-12

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

All times shown according to UTC.

Time Nick Message
03:06 ronsavage joined #marpa
04:29 rns idiosyncrat: re http://irclog.perlgeek.de/m​arpa/2016-04-11#i_12320111 -- all valid points, got it.
04:30 rns re http://irclog.perlgeek.de/m​arpa/2016-04-11#i_12320119 -- ok, will take it
04:32 shadowpaste0 "rns" at 217.168.150.38 pasted "Marpa::R2 build error on darwin reproduced under cygwin" (19 lines) at http://fpaste.scsys.co.uk/510304
04:36 rns idiosyncrat: build failure under darwin can be related to parallel make -- they use make -j4 -j4
04:37 rns perhaps we can use .NOTPARALLEL: to prevent this failure
04:50 ronsavage joined #marpa
04:57 idiosyncrat_ rns: re http://irclog.perlgeek.de/m​arpa/2016-04-12#i_12321634
04:58 idiosyncrat_ Apparently .NOTPARALLEL is GNU make specific, and there's no convenient way to detect GNU make
04:59 idiosyncrat_ Maybe if we made the change Darwin specifc -- add '.NOTPARALLEL' to the Makefile postamble only for Darwin.
05:01 rns Yes. BTW, do you see that failure under linux with make -j4 -j4 releng?
05:08 idiosyncrat_ Just tried it.  Yes, it fails with "-j4 -j4"
05:10 idiosyncrat_ Getting late here.  Will backlog.
05:10 idiosyncrat_ Good night!
05:11 rns Good night!
05:18 teatime hrm, curious: what is -j4 -j4 supposed to do?
05:28 rns teatime: good question. :) 4 threads on each of two processors, if any? :) Unfortunately, it fails with make -j4 too.
05:31 rns Oh, it means -j4 -- ‘If there is more than one ‘-j’ option, the last one is effective.’ -- http://www.gnu.org/software/make/​manual/make.html#Options-Summary
05:37 teatime ah, word.  I was expecting something fancy like you suggested :)
05:37 teatime that's at least for GNU make, I guess.
05:43 rns It looked, well, impressive, even for GNU make. :)
05:45 rns idiosyncrat: on BSDs in most (didn't look into all, but most) cases the failure is
05:45 rns cp engine/gnu_ac_build/.libs/libmarpa.a xs/libmarpa.a
05:45 rns cp: engine/gnu_ac_build/.libs/libmarpa.a: No such file or directory
05:45 rns *** [xs/libmarpa.a] Error code 1
05:46 teatime it seems like if -j (positive int) fails, most likely you lack a depdencency map in your makefile that should exist?
05:47 teatime it seems like Duration could even 'is Rat', but that could suck for future additions/changes, and/or make me people feel icky?  not sure.
05:51 rns Good point -- seems to be marpa_codes.h isn't copied to the build dir as it should.
05:51 teatime sorry last message was a mis-chan :)
05:52 rns :) Well, it looked like that. No prob if you ask me, just fun. :)
06:19 shadowpaste0 "rns" at 217.168.150.38 pasted "make sure all libmarpa files are install()'ed before other recipes execute" (15 lines) at http://fpaste.scsys.co.uk/510305
06:19 rns FWIW, with the above patch make -j4 releng works for me under cygwin.
06:48 rns Looks like .NOTPARALLEL is supported by *BSD -- https://www.freebsd.org/cgi/man.cgi?make -- http://man.openbsd.org/OpenBSD-current/man1/make.1 -- http://netbsd.gw.com/cgi-bin/​man-cgi?make++NetBSD-current
08:26 rns idiosyncrat: re -- http://irclog.perlgeek.de/m​arpa/2016-04-11#i_12320119 -- https://github.com/jeffreykegler/Marpa--R3/pull/12 -- let me know if I can help with conversion of other tests into SLIF.
10:32 kaare_ joined #marpa
12:10 koo7 joined #marpa
16:20 idiosyncrat_ joined #marpa
16:21 idiosyncrat_ rns: If my makefile dependencies could be changed to support paralllized makes, that would be best.
16:24 idiosyncrat_ Re .NOTPARALLEL and BSD -- if it's not a POSIX make feature, I'd like it turned on a OS-by-OS basis, rather than just adding it and hoping there are no side-effects in those makes which don't support it.  Side-effects are unlikely, to be sure, but I think it's bad practice to introduce non-portable stuff into the general release just because we're almost certain to get away with it.
16:27 idiosyncrat_ rns: Two more tests conversion could do are randal.t -> sl_randal.t; and minus.t -> sl_minus.t
16:28 idiosyncrat_ Your idea of leaving the tests broken was an excellent one.
16:29 idiosyncrat_ For show_rules() and show_symbols(), and in general when you leave a test to me to get working, leave the old, failing expected output in the example -- it can be helpful.
16:29 idiosyncrat_ Thanks!
21:47 idiosyncrat_ joined #marpa

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