Perl 6 - the future is here, just unevenly distributed

IRC log for #perl11, 2014-04-05

| Channels | #perl11 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:24 bulk88 here is the reason why you can't include sstream after perl's headers, but this is win32 perl specific, Will_the_Chill http://paste.scsys.co.uk/341106
00:25 bulk88 http://perl5.git.perl.org/perl.git/blob/HEAD:/XSUB.h#l528
00:26 bulk88 its part of NO_XSLOCKS
00:26 bulk88 but I thought I turned them off
00:26 bulk88 or what that in 03.t, not 04.t I did that
00:26 bulk88 *was that in
01:08 Will_the_Chill bulk88: so I'm counting at least 3 issues we found with you since last night...
01:08 Will_the_Chill 1.  spaces in file names (Reini says he fixed)
01:09 Will_the_Chill 2.  compiler flags including NO_XSLOCKS
01:09 Will_the_Chill 3.  move inline headers after C/C++ headers
01:09 Will_the_Chill are those the only 3???
01:38 bulk88 NO_XSLOCKS and move headers are sorta  related
01:38 bulk88 if NO_XSLOCKS is universally added, moving the headers might not be necessery
01:39 bulk88 you see what XSUB.h does
01:39 bulk88 spaces in file name is a different independent issue
02:18 Will_the_Chill bulk88: okay so am I missing any other issues that are outstanding with your side of things?
02:44 bulk88 thats it for now, I bet there will be more problems once I get past 04.t
02:44 bulk88 03.t segvs/exceptions due to lack of NO_XSLCKS
02:45 bulk88 04.t has missing -D CCFLAGs defines
02:45 bulk88 and XSLOCKS/sstream syntax error (same issue) problem
02:46 bulk88 I showed you the .i file earlier tofay of why setbug method syntax errored in sstream
02:46 bulk88 *setbuf
02:47 Will_the_Chill can't we easily fix the compiler flags issues in this file?
02:47 Will_the_Chill https://github.com/wbraswell/rperl/blob/master/lib/RPerl/Inline.pm
02:47 Will_the_Chill I bet we can get all of these issues solved tonight
02:47 Will_the_Chill and Reini says he fixed some of the problems already
02:47 bulk88 http://paste.scsys.co.uk/341193
02:47 Will_the_Chill oh, he says he fixed the space-in-filename issue in Inline
02:48 Will_the_Chill what is -DRRRRRRRRR ?
02:51 bulk88 http://paste.scsys.co.uk/341196
02:51 bulk88 meaningless I added as proof that my change went through
02:51 bulk88 your instructions say -Ilib not the traditional -Tblib
02:51 bulk88 *-Iblib
02:51 bulk88 so at first I edited the blib and nothing was happening
02:52 bulk88 eventually I edited the lib copy and the change took effect
02:52 Will_the_Chill you are talking about the instructions from my install_notes.txt file?
02:52 bulk88 yes
02:52 Will_the_Chill yes Reini now has everything in blib due to `make test` finally being implemented
02:52 bulk88 when I run dmake, the blib copy is used, your manual instructions,  well use lib
02:52 Will_the_Chill but we still need to be able to use the manual test methods for when make may fail
02:53 Will_the_Chill so I must modify my install_notes.txt to include both lib/ and blib/ I think, yes?
02:53 bulk88 also if something goes wrong and the build dir makefile is not regened by EUMM/Inline I can see that then with that bogus define
02:53 bulk88 no, 1 or the other
02:53 bulk88 blib is an ignored dir by git
02:54 bulk88 it contains what is eventually copied to your isntalled perl
02:54 bulk88 lib contained tracked git files
02:54 Will_the_Chill well Reini now has blib activated due to `make test` so what can I do?
02:54 Will_the_Chill yes well you are telling me to choose one or the other and I don't know how I can only choose one?
02:54 bulk88 its a policy/sdanity choice, not a technical choice
02:54 Will_the_Chill I need my manual tests to still work with just lib/
02:55 Will_the_Chill but now Reini has `make test` working with blib/
02:55 bulk88 if your .pm files are generated by makefile.PL, sometimes they are, then they would be written into blib, not lib
02:55 Will_the_Chill what is the best practices for this case?
02:55 bulk88 I guess leave them the way they are
02:55 bulk88 you wont have a blib until you run make atleast once
02:55 bulk88 so if you do just makefile.pl, then perl -Ilib /t/01.t
02:56 Will_the_Chill so the real point is that we either trust make or we don't trust make
02:56 bulk88 that will work ATM
02:56 Will_the_Chill if we trust make, then we make everything use blib/
02:56 Will_the_Chill if we don't trust make, we can keep using lib/, but then what's the point of having make at all if we don't trust it?
02:56 bulk88 why dont we trust make?
02:56 Will_the_Chill I'm just paranoid because RPerl is so crufty
02:57 Will_the_Chill adding another layer of automation between me and the code just means another layer of possible weirdness
02:57 Will_the_Chill when there are already 17 layers of weirdness
02:57 Will_the_Chill is blib/ the new standard I assume?
02:58 bulk88 remember these are policy choices,  you can ISA override EUMM and have the makefile put the files staged for copying to installed perl intoa  folder called "perlsucksgousepython", its a policy/sanity choice, not a technbical choice
02:58 bulk88 blib is the EUMM default, Module Build does something different I dont care to remember
02:58 Will_the_Chill yes I realize it is policy, I am asking from that perspective
02:58 bulk88 you use EUMM, so blib is the standard
02:59 Will_the_Chill can we think of any situations where somebody would have Perl but not make?
02:59 bulk88 but in order to use blib, the instructions have to say to run make once
02:59 bulk88 if you dont have make, you cant install 90% of CPAN modules
03:00 bulk88 there is a tool on CPAN for pure perl modules, so you dont need make, but few use it
03:00 Will_the_Chill okay, then we'll go with blib/
03:00 Will_the_Chill I'll update the install_notes.txt now
03:00 bulk88 if you dont have make, you dont have a c compiler, and whats the point of rperl testing then?
03:00 Will_the_Chill right, that was the point I was looking for
03:00 Will_the_Chill so we go with blib/
03:00 bulk88 change them to blib
03:00 bulk88 and incldue to run make atleast once before trying individualk runs
03:01 bulk88 your dirty unwashed testers might not know what -I or blib are
03:01 bulk88 since they aren't C people
03:02 Will_the_Chill roger that, doing it now
03:02 Will_the_Chill meanwhile, please tell me exactly what to do to RPerl/Inline.pm
03:02 Will_the_Chill not the -DRRRR thing, but the real flags?
03:04 bulk88 -DNO_XSLOCKS has to be on the command line to every call of g++/gcc
03:04 bulk88 also you can't wipe out CCFLAGS, you can only cat onto it
03:04 bulk88 IDK where exactly to get the correct copy of CCFLAGS from
03:05 bulk88 %Config is the easy choice,  but IDK if its the right choice
03:05 bulk88 notice I was useing CCFALGSEX
03:05 Will_the_Chill well if you could help me put together that solution, that'd be great
03:08 bulk88 I'm not going to research Inline's API or figure out its internal APIs for you, its your project and your it's designer, my changes are random unorganized hacks not fit for RTM
03:08 bulk88 notice I used 2 different ways http://paste.scsys.co.uk/341207
03:08 Will_the_Chill oh I didn't realize it was related to Inline's internals
03:09 bulk88 IDK which is best or anything
03:09 Will_the_Chill I thought it was just windows-specific compiler flags
03:09 Will_the_Chill I thought it was just some compiler flags you could tell me to put in there
03:09 bulk88 they are windows specific, but you have to figure out how to get them to g++ through your software stack
03:10 bulk88 you will also have to make -Wno-reserved-user-defined-literal -Wno-literal-suffix GCC version specific, since those 2 dont exist in 4.6.3
03:10 bulk88 and are giving me warnings
03:10 bulk88 eventually you will also have to add -DPERL_NO_GET_CONTEXT for speed
03:10 Will_the_Chill yes we know about those warnings, Reini had to add those flags to make it work for Macintosh I think
03:11 bulk88 or #define PERL_NO_GET_CONTEXT before perl.h
03:11 Will_the_Chill so can I start by just adding -DNO_XSLOCKS -fpermissive to the RPerl/Inline.pm compiler flags?
03:11 Will_the_Chill then go on to get a copy of the existing CCFLAGS and merge?
03:11 bulk88 -fpermissive  is hacking from yesterday ignore it
03:11 bulk88 it fixed nothing
03:12 bulk88 PERL_NO_GET_CONTEXT is something you can test on *nix if you have an ithread perl
03:12 bulk88 the XSUB.h/libc token overrides are win32 specific since PERL_IMPLICIT_SYS doesn't work on anytihng but win32
03:13 bulk88 PERL_IMPLICIT_CXT means every perl function starts with my_perl param, it is required for ithreads on all OSes
03:14 Will_the_Chill okay you're kinda losing me a bit
03:14 Will_the_Chill :)
03:14 Will_the_Chill so I only need -DNO_XSLOCKS ???
03:14 bulk88 PERL_IMPLICIT_SYS is a OS virtualizer, it creates multiple pwds in 1 process for example, it only works on Win32 and is used for psuedofork
03:14 bulk88 ATM yes
03:14 bulk88 I'll retry your github rperl after you figure out how to get in the right way
03:15 bulk88 PERL_IMPLICIT_SYS also creates multiple copies of env vars in 1 process
03:15 bulk88 to do that, you have to hook libc all over the place obviously
03:16 Will_the_Chill okay I think I've found all the need-to-be-changed occurrences of lib/ to blib/
03:16 Will_the_Chill running tests
03:19 Will_the_Chill okay EUMM issue
03:19 Will_the_Chill it is not copying over lib/RPerl/HelperFunctions.cpp into blib/
03:20 Will_the_Chill but it IS copying over some other .cpp files from lib/ to blib/
03:20 Will_the_Chill why?
03:20 Will_the_Chill I've never used EUMM before  :/
03:21 Will_the_Chill OH WAIT it is getting copied over
03:21 Will_the_Chill but somehow it is not being found...  :(
03:21 Will_the_Chill okay this is weird
03:22 Will_the_Chill OHHHHHH I see
03:22 Will_the_Chill I have to change everything from lib/ to blib/lib/
03:22 Will_the_Chill what I did was change from lib/ to blib/
03:22 Will_the_Chill am I correct here?
03:22 bulk88 yes
03:22 bulk88 i thought of 1 reason to not swtich things
03:22 Will_the_Chill oh?
03:22 bulk88 and its lazyness :D
03:23 Will_the_Chill haha well I'm already halfway changed and I want to just get it changed!
03:23 Will_the_Chill :D
03:23 Will_the_Chill okay, trying again!
03:23 bulk88 if you edit /lib, you are editing the git version, if you edit blib, you have to copy back, and rememebr not to make clean or u loose ur changes
03:23 bulk88 if you edit /lib, the changes wont appear in /blib until your "make" again
03:24 bulk88 ....     /lib is faster to develop with
03:25 bulk88 i'd leave the blib  vs lbi aloen right now
03:25 bulk88 *lib
03:25 bulk88 -DNO_XSLOXKs and the  not dropping CCFLAGs  are more improtant
03:26 * bulk88 doesn't actually see what he types before he sends, so spelling mistakes are huge
03:27 Will_the_Chill already committed to changing from lib to blib/lib
03:27 Will_the_Chill and yes I will ONLY edit lib/ files, then do make
03:32 Will_the_Chill yay tests pass with blib/lib!
03:37 travis-ci [travis-ci] RPerl Commit By Will Braswell. The build passed. http://travis-ci.org/wbraswell/rperl/builds/22319318
03:38 Will_the_Chill DOUBLE YAY!  Travis passed too!  :D
03:38 Will_the_Chill bulk88: now onto -DNO_XSLOCKS
03:50 Will_the_Chill okay seems to compile fine on my machine with -DNO_XSLOCKS
03:53 Will_the_Chill bulk88: am trying to figure out how to get cflags...
03:53 Will_the_Chill http://stackoverflow.com/questions/8740443/how-to-change-default-cflags-for-makemaker
03:54 Will_the_Chill I don't see cflags here?
03:54 Will_the_Chill https://metacpan.org/pod/distribution/perl/configpm
04:07 bulk88 joined #perl11
04:13 bulk88 reading
04:16 Will_the_Chill bulk88:
04:16 Will_the_Chill perl -e 'use Config qw(myconfig); print myconfig();'
04:16 Will_the_Chill gives
04:16 Will_the_Chill Compiler:
04:16 Will_the_Chill cc='cc', ccflags ='-D_REENTRANT -D_GNU_SOURCE -DDEBIAN -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64',
04:16 Will_the_Chill optimize='-O2 -g',
04:16 Will_the_Chill cppflags='-D_REENTRANT -D_GNU_SOURCE -DDEBIAN -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include'
04:16 Will_the_Chill I've got that far at least.  :)
04:17 bulk88 use no, use $Config{ccflags}
04:17 bulk88 *no, use $
04:17 bulk88 the problem is, it wont work on activeperl
04:17 bulk88 maybe forget about AP for right now, and lets try just strawberry
04:18 bulk88 AP's %Config will be set up for Visual C, but AP includes some .pm files, which doa some slicing and dicing, and integrate into EUMM, allow GCC for work
04:19 bulk88 even though perl516.dll was compiled with VC
04:19 Will_the_Chill I tried ccflags but it said it wasn't exported by Config
04:19 Will_the_Chill OH it is cppflags
04:19 Will_the_Chill OR NOT
04:20 Will_the_Chill okay it says cppflags is not exported by Config
04:20 Will_the_Chill dang
04:22 bulk88 http://perldoc.perl.org/Config.html
04:24 Will_the_Chill perl -e 'use Config qw(ccflags); print config_ccflags();'
04:24 Will_the_Chill "ccflags" is not exported by the Config module
04:24 Will_the_Chill what am I doing wrong?
04:36 Will_the_Chill also
04:36 Will_the_Chill perl -e 'use Config qw(config_ccflags); print config_ccflags();'
04:36 Will_the_Chill "config_ccflags" is not exported by the Config module
04:37 Will_the_Chill oh dang okay I see
04:38 Will_the_Chill perl -e 'use Config;  print $Config{ccflags};'
04:38 Will_the_Chill -D_REENTRANT -D_GNU_SOURCE -DDEBIAN -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64
04:39 Will_the_Chill also
04:39 Will_the_Chill perl -e 'use Config;  print $Config{cppflags} . "\n";'
04:39 Will_the_Chill -D_REENTRANT -D_GNU_SOURCE -DDEBIAN -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include
04:39 Will_the_Chill bulk88: which to use???
04:39 Will_the_Chill ccflags or cppflags?
04:43 Will_the_Chill testing now with ccflags
04:45 Will_the_Chill yay it works with ccflags!
04:47 Will_the_Chill bulk88: pushed now to github, give it a whirl!  :D
04:48 bulk88 IDK ccflags vs cppflags
04:48 bulk88 ill try in maybe 30 mins
04:49 Will_the_Chill okay well when you try it on your system, you can let me know if we should switch from ccflags to cppflags
04:49 Will_the_Chill thanks!  :)
04:50 Will_the_Chill also, I'm trying to figure out how to tell EUMM to use Reini's new pre-release Inline which fixes some of the stupid warnings and the space-in-filename issue
04:52 rurban joined #perl11
04:54 travis-ci [travis-ci] RPerl Commit By Will Braswell. The build passed. http://travis-ci.org/wbraswell/rperl/builds/22321240
04:55 Will_the_Chill rurban: hey!
04:55 Will_the_Chill Unparsable version 'Inline-0.54_01' for prerequisite Inline at Makefile.PL line 5.
04:55 Will_the_Chill Unparsable version 'RURBAN/Inline-0.54_01' for prerequisite Inline at Makefile.PL line 5.
04:55 Will_the_Chill Unparsable version 'RURBAN/Inline-0.54_01.tar.gz' for prerequisite Inline at Makefile.PL line 5.
04:56 Will_the_Chill how do I tell EUMM to install your new 0.54_0X
04:56 Will_the_Chill ???
05:14 Will_the_Chill okay I think this must be impossible with EUMM, I will just use cpanm instead
05:27 bulk88 still busy
05:28 Will_the_Chill bulk88: roger
05:28 Will_the_Chill I manually installed Reini's new Inline-0.54_02
05:28 Will_the_Chill I am testing it now
05:28 Will_the_Chill okay, passes on my machine, but still emits lots of the annoying warnings
05:32 bulk88 ill install the reini unauth inline ok?
05:32 Will_the_Chill yes
05:32 Will_the_Chill cpanm RURBAN/Inline-0.54_02.tar.gz
05:33 bulk88 afk for 15 mins
05:34 Will_the_Chill roger
05:37 travis-ci [travis-ci] RPerl Commit By Will Braswell. The build passed. http://travis-ci.org/wbraswell/rperl/builds/22322507
05:39 Will_the_Chill Okay so far I am not at all annoyed by travis bot telling me that my builds are passing.  :D
05:40 Will_the_Chill bulk88: I was unable to figure out a way to make EUMM install Reini's Inline v0.54_02, so I just made a cpanm command for it in the install notes, now in the INSTALL file
05:40 Will_the_Chill if you know how to make EUMM do this it would be good, so that travis can run automated tests using v0.54_02
05:44 Will_the_Chill so your tests will tell us if -DNO_XSLOCKS solves the issue so we don't need to move inline headers to be after C headers?
05:50 rurban EUMM cannot PREREQ unauthorized releases
05:51 rurban But you can add to lib/ again
05:53 rurban The missing $Inline::CPP::ntype really gets annoying
05:58 bulk88 EUMM doesn't install or download anything
05:59 bulk88 EUMM creates meta files which specify prereqs, a separate tool like cpan shell then processes and follows
05:59 bulk88 then
06:02 Will_the_Chill rurban: how to fix the ntype annoying error?
06:07 rurban Haven't looked yet. It's in CPP
06:07 Will_the_Chill okay
06:07 rurban Tomorrow if Mirjam let's me
06:08 Will_the_Chill *going to eat*
06:22 bulk88 back in 1-2 hrs
07:02 Will_the_Chill early morning for me, going to sleep now, catch up tomorrow!  :)
07:08 rurban joined #perl11
08:08 rurban joined #perl11
08:23 acanfora joined #perl11
08:59 basiliscos joined #perl11
09:02 bulk88 http://paste.scsys.co.uk/341421 Inline still fails miserably with path with spaces
09:03 bulk88 hmm no wtc
09:09 rurban joined #perl11
09:33 bulk88 http://paste.scsys.co.uk/341435 , 03.t still crashes same as before http://paste.scsys.co.uk/341441
09:38 bulk88 here is 04.t http://paste.scsys.co.uk/341445
09:39 bulk88 nothing changed anywhere, im using hust your git's bead
09:39 bulk88 *just
10:10 rurban joined #perl11
11:11 rurban joined #perl11
12:12 rurban joined #perl11
13:12 rurban joined #perl11
13:59 rurban joined #perl11
15:32 mirjam joined #perl11
18:27 Will_the_Chill joined #perl11
18:40 basiliscos Will_the_Chill: Hello!
18:40 basiliscos Will_the_Chill: newest rperl passes tests well, except some warning: http://bpaste.net/show/198018/
18:41 basiliscos instructions for installation are perfect, and no error I've met
19:47 rurban joined #perl11
19:55 bulk88 http://paste.scsys.co.uk/341435 , 03.t still crashes same as before http://paste.scsys.co.uk/341441 ,  here is 04.t http://paste.scsys.co.uk/341445 , nothing in src code changed anywhere, im using just your git's head
19:55 bulk88 http://paste.scsys.co.uk/341421 Inline still fails miserably with path with spaces
20:00 rurban joined #perl11
21:01 rurban joined #perl11
22:02 rurban joined #perl11
23:03 rurban joined #perl11
23:05 rurban1 joined #perl11
23:24 Will_the_Chill bulk88: looking
23:25 Will_the_Chill basiliscos: yes you can ignore those warnings, we are working to get them fixed, as long as the tests pass then it is working correctly!
23:28 Will_the_Chill bulk88: your code is dying because the code must have write permission to it's own directories:
23:28 Will_the_Chill # died: Can't move rperltypes_mode.h input file to rperltypes_mode.h.orig: Permission denied, dying at blib/lib/rperltypes.pm line 118, <$TYPES_H_FILEHANDLE_IN> line 9.
23:30 Will_the_Chill rurban_: bulk88 has shown that Inline is still failing on folders-with-spaces-in-their-name     http://paste.scsys.co.uk/341421

| Channels | #perl11 index | Today | | Search | Google Search | Plain-Text | summary