Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2015-03-15

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

All times shown according to UTC.

Time Nick Message
03:20 mohawk well that was more epic than expected
03:20 opkick [pdl] mohawk2 created corebuild-improve (+3 new commits): http://git.io/ppXI
03:20 opkick pdl/corebuild-improve 7b9db8d Ed J: "core" target also needs Math and MatrixOps
03:20 opkick pdl/corebuild-improve 175e11b Ed J: Change Core.o dep to just make blib/arch, avoid always recompiling
03:20 opkick pdl/corebuild-improve d0799c8 Ed J: Work around parallel-make problem for "core" target
03:23 opkick [pdl] mohawk2 opened pull request #73: "core" target fixes (master...corebuild-improve) http://git.io/ppXR
03:48 opkick [pdl] coveralls comment on issue #73:
04:04 opkick [pdl] wchristian pushed 1 new commit to master: http://git.io/ppyt
04:04 opkick pdl/master 31266c2 Derek Lamb: Add names for all the tests in t/ops.t....
04:10 mohawk looks like derek doesn't like the idea of making branches?
04:27 travis-ci PDLPorters/pdl#114 (master - 31266c2 : Derek Lamb): The build passed.
04:27 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/8647162a42f6...31266c24d4ef
04:27 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/54419923
13:12 Mithaldu for commits that only add documentation that's fine
13:13 Mithaldu mohawk: btw, feel like hearing about a case where i spent two fucking days rewriting git history?
13:13 Mithaldu it's in a project that's fairly similarly named to this one
13:53 mohawk Mithaldu, i do - hit me :-)
13:56 opkick [pdl] mohawk2 comment on issue #73: Further pondering reminds me that the "core" target with the workaround only works for "dynamic". Likewise, the "coretest" only works with dynamic, although that's fixable with a recursive `$(MAKE) test TEST_FILES="t/core.t t/ops.t"`. http://git.io/pjaJ
13:57 opkick [pdl] mohawk2 comment on issue #73: (Although in "static", that recursive make would actually depend on "static" which would build the whole thing, thereby rendering the smaller build pointless). http://git.io/pjaB
14:00 Mithaldu mohawk: ever looked at SDL?
14:00 mohawk i have not - URL?
14:00 Mithaldu https://github.com/PerlGameDev/SDL/network
14:00 Mithaldu be sure to hit f5 after the first load
14:01 mohawk yeesh, i see a merge commit from you? ;-)
14:02 mohawk and wow, may last year(?) looks "complicated"
14:03 mohawk and before then, all very twisty-turny
14:04 mohawk this is *after* your history rewrite?
14:04 Mithaldu this is the after state
14:05 * mohawk weeps
14:05 Mithaldu before that it was ALL twisty-turny and "complicated"
14:07 Mithaldu sec, doing something else for a moment
14:07 mohawk ok
14:07 chm joined #pdl
14:09 chm wow....lot's of progress.  I'm working through commit messages and email.
14:09 chm Anyone with a quick summary of where we stand w.r.t. 2.008 feature freeze and release.
14:09 chm I've seen a lot of activity but not sure whether or which commits made it to master.
14:10 mohawk chm, good question
14:10 mohawk these are the open PRs: https://github.com/PDLPorters/​pdl/pulls?q=is%3Aopen+is%3Apr
14:11 mohawk they're mostly works in progress
14:11 mohawk i would like to see "changesfix" merged, since it makes PDL "changes" work right on metacpan
14:11 mohawk which i believe was the point of kmx's doc-mangling effort
14:12 mohawk changesfix doesn't have the second half of the change yet, to convert Release_Notes to the metacpan-standard YAML
14:13 mohawk that'll be a very easy one, but sivoais has to work on his thesis
14:14 mohawk chm, otherwise, in terms of PDL actual features for 2.008, look at: https://github.com/PDLPorters/pdl/issues?q=is%​3Aopen+is%3Aissue+milestone%3A%22PDL+v2.008%22
14:14 mohawk i think that one may well be fixed (sivoais can weigh in), so we may well be good to go
14:17 opkick [pdl] mohawk2 comment on issue #33: @zmughal - does 4b77820a0e783ab6d553e6fbf2f2a7bcb9b943ce fix this in your view? http://git.io/pj6O
14:17 Mithaldu chm: alright, so
14:17 chm mohawk, taking a look at the bugs list http://sourceforge.net/p/pdl/bugs/ shows all the priority fixes in
14:18 Mithaldu as you see they have a master branch and an experimental branch
14:18 mohawk Mithaldu, on pdl?
14:18 Mithaldu and in the past they merged back and forth between them, even did dev releases directly from the experiment branch
14:18 chm except for PDL::NiceSlice which I would like to move to the 2.009 release cycle to avoid further delay
14:18 Mithaldu sdl
14:18 mohawk ok
14:18 mohawk chm, i agree with that
14:19 chm mohawk: for today I'll review and update the imag2d patch if needed (it in there, just may need tweak/docs)
14:19 mohawk chm, i propose that subject to sivoais agreeing with this, that we feature freeze except for updating our Changes process
14:19 mohawk (and your imag2d tweaks)
14:19 mohawk what do you say?
14:19 chm I'll revert the PDL::NiceSlice engine to the 2.007 default
14:20 chm I think PDL::IO::HDF can be deferred to 2.009
14:20 mohawk chm, could you do that on a branch so everyone gets to look and see?
14:20 mohawk (that = the NiceSlice reversion)
14:22 chm Personally, I would like to push the Changes update to 2.009 as well since there are a number of user visible issues that could come up
14:22 mohawk chm, ok
14:22 chm One more release with the current Release_Notes and then move to the more standard perl module usage
14:22 mohawk sounds good
14:22 mohawk i would like 2.009 to happen a great deal sooner than 2.008 did
14:22 chm BTW, some of these things were done long before the perl "standard" existed in a more structured form.
14:23 mohawk chm, i am very aware of that :-)
14:23 chm I would as well.
14:23 mohawk the branching technique will be a way to achieve a hugely faster release cycle
14:23 chm Now that there are more developers active, it should be possible and reasonable to have more frequent releases while retaining quality...
14:23 mohawk quality is a technique thing
14:24 mohawk comes down to good tests, and good process
14:24 chm In this case, I mean that PDL users don't have things break constantly.
14:24 mohawk yes
14:24 mohawk checked by good tests :-)
14:24 chm Things need to change but for a key module, there needs to be evolution, not revolution
14:24 mohawk breakage = regression, testing = regression-prevention
14:24 mohawk chm, you're quoting me now :-)
14:25 chm The biggest source of breakage that arises is like this "lets just change this to wonderful new option b"
14:25 chm Voila!  Broken for existing users....
14:26 chm In the sense of being problematic for PDL users---trying to get work done...
14:26 mohawk that would be a sign of failure of regression tests
14:26 mohawk at the moment we only have 50.29% coverage
14:26 mohawk and we only even know that thanks to sivoais's efforts
14:26 mohawk (and Mithaldu setting up the sf->github mirror)
14:27 chm A regression is a different issue.  The example was where the developers were deliberately changing things because they could.
14:27 mohawk probably because there weren't tests putting a bound on the old behaviour?
14:28 mohawk it is absolutely not a different issue
14:28 chm All of their contributions have made a significant improvement in the development approach which will pay off going forward.
14:28 mohawk chm, is there any assistance i can give in helping you do the niceslice reversion on a branch?
14:28 chm Yes, the new tests are great!  It is valuable but takes time...
14:30 mohawk if you use a branch, we can make a github PR which will run travis over it
14:31 mohawk i can also test it on my win32 setup
14:31 mohawk and fix any problems before it hits master
14:32 mohawk by the way, i think by "new tests" you mean travis running the current tests; i am referring to the actual t/*.t not having tests that "document" (enforce) the current, expected, behaviour
14:32 chm mohawk:  the only change required is commenting out line 8 and uncommenting line 10 in PDL/Basic/SourceFilter/NiceSlice.pm
14:32 mohawk chm, still wants doing on a branch
14:33 chm I would appreciate the help with that as the back-emails include some things I need to learn about the new rebase work flow.
14:33 mohawk ok, i can help with that!
14:34 mohawk here are the git commands (i'm assuming your master is up to date with sf)
14:34 Mithaldu chm: feel free to also ask me whenever you need help
14:34 mohawk git checkout -b nicesliverevert master # branch off master even if you were on different branch
14:34 mohawk # now make your edits
14:34 mohawk # now commit them
14:35 mohawk git push --set-upstream WHATEVERYOUCALLSF nicesliverevert
14:36 Mithaldu chm: btw, you on windows?
14:36 mohawk # in the above, feel free to spell "niceslice" right rather than "niceslive" :-)
14:36 mohawk now, if someone merges something or otherwise pushes onto master while we're working on nicesliverevert
14:37 mohawk then what we do is this:
14:37 chm Mithaldu: I'm in cygwin running on win7
14:37 mohawk git fetch --all
14:37 mohawk git checkout master
14:37 mohawk git merge WHATEVERYOUCALLSF/master
14:38 mohawk git rebase -i master nicesliverevert # will also leave you checked out on that branch
14:38 chm mohawk:  Is WHATEVERYOUCALLSF the url in .git/config?
14:38 mohawk chm, do "git remote -v"
14:38 mohawk i have 3 remotes on my repo
14:38 Mithaldu chm: install and/or upgrade those: https://msysgit.github.io/ https://code.google.com/p/​tortoisegit/wiki/Download
14:38 mohawk this is the crucial bit:
14:38 Mithaldu tortoisegit's log view on a repository makes for much more intuitive manipulation
14:39 mohawk Mithaldu, can we finish me giving the command line version, THEN you give the GUI one?
14:39 mohawk let's minimise distraction :-)
14:39 Mithaldu i'm done already :P
14:39 mohawk for future reference ;-)
14:39 mohawk git push -f # updates sf with rebased branch
14:39 mohawk # done! branch now up to date again
14:44 opkick [pdl] wchristian created niceslicerevert (+1 new commit): http://git.io/pjyY
14:44 opkick pdl/niceslicerevert 27764a1 Chris Marshall: Make Filter::Util::Call the default engine for PDL::NiceSlice...
14:47 pdurbin mohawk: http://i.imgur.com/XFQLB.jpg ;)
14:49 mohawk chm, nice - could you now also open a pull request?
14:49 mohawk that'll trigger the travis build testing
14:49 mohawk pdurbin, ha ha!
14:52 chm mohawk:  How do I do that?
14:53 mohawk https://github.com/PDLPorters/pdl/branches/all
14:53 mohawk see your branch, there's "New pull request"
14:54 opkick [pdl] devel-chm opened pull request #74: Make Filter::Util::Call the default engine for PDL::NiceSlice (master...niceslicerevert) http://git.io/pjHE
14:54 mohawk chm, nice work
14:55 chm very cool!
14:55 mohawk and in the fullness of time, travis will report here with its results
14:55 mohawk welcome to github
14:57 mohawk chm++
14:58 mohawk as promised, i'll kick off a win32 test run now
15:00 mohawk chm, in the meantime, others following the repo will see the notification
15:00 mohawk and can weigh in on the PR ticket with any thoughts
15:00 mohawk including line by line comments
15:00 mohawk i'll make one now to demo
15:01 chm mohawk:  I've taken the PDL::NiceSlice tickets off of the 2.008 list (>6) and the imag2d and twiddle look good.
15:02 opkick [pdl] mohawk2 comment on pull request #74 27764a1: It's a pity this is necessary for now, but evidently it is. We should probably add a test exercising the case in #25 / https://sourceforge.net/p/pdl/bugs/238/ http://git.io/pj7w
15:02 mohawk great
15:03 mohawk as you can see, this irc channel acts as a console for everything happening on the repo, in real time
15:03 chm mohawk:  Are there any other fixes that are done but just not pushed to master?
15:03 mohawk there's my corebuild-improve i'd like to get in there
15:03 mohawk but realistically, that's not critical
15:05 mohawk since it's dev-orientated, leaving that to 2.008_01 is fine
15:05 chm I wouldn't object since it is a developer-visible issue mainly and has a work around if needed (edit perldl.conf by hand)\
15:07 mohawk short version, if we're leaving Changes till post release, i think we're solid
15:07 mohawk please call a feature freeze now :-)
15:08 chm Err, the PDL::NiceSlice still needs to be updated in master
15:08 mohawk (with the niceslice exception, actually)
15:08 mohawk yes, well said
15:08 chm Also, did the better error message SF#332 get into master yet?
15:09 mohawk i can't bear to use the SF interface, i'll check the gh issues list
15:09 mohawk https://github.com/PDLPorters/pdl/issues/33
15:09 mohawk see my comment at bottom
15:09 mohawk gives link to commit on master
15:12 chm how does the commit link indicate it is on master?
15:13 mohawk it doesn't, but i found it by looking in master's commits list :-)
15:13 mohawk Mithaldu, what's the git way to find if a commit is on a spec'ed branch?
15:13 Mithaldu wat
15:14 Mithaldu i'm not sure i understood the question correctly
15:14 Mithaldu that said
15:14 Mithaldu i do git log with tortoisegit, look for the commit in question and follow the lines
15:14 mohawk "is commit xyz on branch abc"
15:15 mohawk http://stackoverflow.com/questions/1419623/how​-to-list-branches-that-contain-a-given-commit
15:15 Mithaldu hey look, magic
15:16 mohawk git branch --contains
15:16 mohawk git branch --contains 4b77820
15:18 mohawk (yes, it's on master)
15:20 chm What are the steps to merge the PDL::NiceSlice change into master?
15:20 mohawk chm, once it's passed its various tests
15:20 mohawk these are the steps:
15:21 mohawk git fetch --all
15:21 mohawk # get master up to date as above
15:21 mohawk git checkout master
15:21 mohawk git merge --ff-only niceslicerevert
15:22 mohawk whoa, i got a test fail on win32
15:22 opkick [pdl] coveralls comment on issue #74:
15:23 mohawk on t/ops.t
15:23 Mithaldu mohawk: you forgot a step
15:23 mohawk chm, look at the branches page again
15:23 mohawk final step:
15:23 Mithaldu 1: note down the commit id of niceslicerevert before doing anything
15:23 mohawk git push
15:23 mohawk Mithaldu, why?
15:23 Mithaldu 2. git diff master SAID_ID_HERE
15:24 Mithaldu 3. make sure that there were no unexpected changes from git's merge logic
15:24 mohawk Mithaldu, that is more easily done using github web ui
15:24 mohawk Mithaldu, did you notice --ff-only on my merge?
15:24 Mithaldu so?
15:24 Mithaldu that only prevents it from doing things in case of conflict
15:24 Mithaldu it *will* happily fuck up your shit if the merge logic does something you didn't expect
15:24 mohawk it's not really a merge
15:24 mohawk it's a fastforward
15:25 mohawk no conflicts are possible
15:25 Mithaldu no rebase involved?
15:25 mohawk why don't you have a read of the "merge" man page and get back to me
15:25 mohawk chm, status - the red X is because coverage dropped, which is fine
15:25 mohawk let me look at why i'm getting fail on win32
15:25 chm mohawk: it looks like the build passed but the coverage "failed" which is a bit bogus
15:26 Mithaldu yeah, ok, i'm done
15:26 Mithaldu bye
15:26 Mithaldu chm: feel free to email if you need anything about opengl or whatever, or just /msg me on here
15:26 chm Mithaldu:  Thanks.
15:27 Mithaldu i don't care to try and help a project where the main actor responds to earnest attempts to help with abuse
15:27 Mithaldu left #pdl
15:27 mohawk oh dear
15:28 mohawk chm, that's the criteria it uses
15:28 mohawk let's not worry :-)
15:31 chm It's merged.  Lunch now.  o/
15:31 mohawk chm, that is not good
15:32 mohawk i just told you it's getting test fails and you merged anyway?
15:34 opkick [pdl] wchristian merged niceslicerevert into master: http://git.io/hevC
16:00 travis-ci PDLPorters/pdl#116 (master - 27764a1 : Chris Marshall): The build passed.
16:00 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/31266c24d4ef...27764a129141
16:00 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/54458462
16:11 mohawk luckily enough, the commit merged above isn't what's causing the win32 test breakage
16:28 mohawk situation: this snippet of t/ops.t is breaking on my strawberry perl:
16:28 mohawk $INT_MAX=2147483647;
16:29 mohawk cmp_ok indx($INT_MAX*4)%2, '==', 0, 'big indx modulus';
16:29 mohawk cmp_ok double($INT_MAX*4)%2, '==', 0, 'big double modulus';
16:29 mohawk these two tests fail
16:34 opkick [pdl] mohawk2 created intmaxfix (+1 new commit): http://git.io/heWK
16:34 opkick pdl/intmaxfix 0fc1c2b Ed J: Change t/ops.t indx and double modulus tests to cmp_ok
16:35 opkick [pdl] zmughal comment on issue #33: Oh yes! It's done! http://git.io/helO
16:35 opkick [pdl] zmughal closed issue #33: SF#332 "isn't numeric in null operation" warning could be more helpful http://git.io/xgC5
16:38 mohawk sivoais, welcome back
16:38 mohawk we have a minor drama now - win64 fail
16:38 sivoais :-D I just read through the log!
16:38 sivoais ooh, what's happened?
16:38 mohawk see log
16:39 sivoais well, we could bisect, but that might take a while...
16:39 * sivoais goes to look at git log
16:41 sivoais I just had a wonderful idea for my thesis.. have to write it down :-)
16:43 mohawk that takes priority
16:44 sivoais done! :-) ok, back to the log
16:44 mohawk i sent an email re the win64 issue to pdl-d, we'll see what transpires
16:45 mohawk chm, once a branch is merged, the way to tidy it up is like so:
16:46 mohawk git checkout niceslicerevert
16:47 mohawk git push WHATEVERYOUCALLSF --delete niceslicerevert
16:48 opkick [pdl] mohawk2 deleted niceslicerevert at 27764a1: http://git.io/heRm
16:51 opkick [pdl] mohawk2 opened pull request #75: t/ops.t fail on 64-bit Strawberry Perl (master...intmaxfix) http://git.io/he0I
16:58 sivoais mohawk: how does that fix it on MSWin?
16:59 opkick [pdl] mohawk2 comment on issue #75: To add to this, the `indx` number got is 0x200000000, while the `double` one is 0x2fffffffc. http://git.io/heuI
17:00 sivoais I'm going to pull out my new Win environment and give this a go
17:11 opkick [pdl] mohawk2 comment on issue #75: The relevant-looking bit of Strawberry Perl's -V is:... http://git.io/he2p
17:13 sivoais mohawk++
17:13 sivoais that would mean we can't just multiply by 4
17:13 mohawk indeed
17:18 opkick [pdl] mohawk2 force-pushed intmaxfix from 0fc1c2b to 5cbd281: http://git.io/herW
17:18 opkick pdl/intmaxfix 5cbd281 Ed J: Change t/ops.t indx and double modulus tests to see got vs expected
17:21 opkick [pdl] coveralls comment on issue #75:
17:39 drrho joined #pdl
17:40 opkick [pdl] zmughal opened issue #76: Continue PDL_Anyval work http://git.io/he1X
17:44 opkick [pdl] zmughal opened issue #77: Add information about MAX and MIN values of each PDL::Type http://git.io/heM5
17:44 opkick [pdl] coveralls comment on issue #75:
17:45 opkick [pdl] zmughal comment on issue #67: Also, make sure that the format of the Changes file is documented in `DEVELOPMENT` and part of the release process. http://git.io/heDC
17:50 sivoais mohawk: if you want to fix intmaxfix, then you can look at howbig() and use that to determine the multiplier to use
17:51 mohawk sivoais, ah HA
17:51 mohawk where is that doc'ed?
17:51 sivoais PDL::Core
17:51 mohawk in fact, if you can give me the code snippet i can go from there?
17:51 sivoais yes
17:51 mohawk like this? $size = howbig($piddle->get_datatype);
17:52 sivoais yeah
17:52 sivoais that gives you the bytes
17:52 mohawk is this right? howbig(byte)?
17:53 sivoais yes
17:53 mohawk perl -MPDL::LiteF -MPDL::Core=:Internal -e 'die howbig(long)'
17:53 mohawk 4 at -e line 1.
17:53 mohawk magic
17:53 sivoais howbig(indx) / howbig(long)
17:54 mohawk even more concise: perl -MPDL::LiteF -e 'die PDL::Core::howbig(long)'
17:56 mohawk problem is that howbig on both indx and double report same on both platforms
17:57 mohawk so that won't actually help
17:58 mohawk question is, what is indx($INTMAX*4) % 2 actually SUPPOSED to be?
17:58 mohawk why is this a meaningful test?
18:01 sivoais mohawk: I believe it is trying to make sure that the values can only fit in those types without overflowing
18:02 sivoais mohawk: though, it might be better to do indx($INTMAX)*4
18:02 sivoais not $INTMAX*4
18:02 mohawk what's the * 4 intended to achieve?
18:03 mohawk isn't the fact that's it's bigger than the max integer already a problem?
18:07 sivoais I'm looking at the commit where it was introduced
18:08 sivoais 7ebd2d5
18:10 sivoais mohawk: the intention is to test large values... ones that are larger than those that fit in long()
18:10 mohawk and that's by using eg a double?
18:11 sivoais it has to test each type separately because they go down different code paths
18:11 mohawk ok
18:12 sivoais that's why I made a category of bugs specific to the type-system :-) It's a huge part of what makes PDL powerful, but also complex
18:13 mohawk so how does it actually store these things?
18:13 mohawk that's kind of important to figuring out the limits of how one can meaningfully test it
18:15 sivoais mohawk: man PDL::Internals
18:15 sivoais the code is in pdl.h
18:15 sivoais struct pdl
18:16 mohawk can you guide me to which bit of the 600 line doc is relevant? :-)
18:17 sivoais mohawk: where it says "So what does the structure look like"
18:18 sivoais mohawk: what type of thing are you looking for?
18:19 mohawk top-level - i want t/ops.t to check the right thing on all platforms
18:19 sivoais right
18:20 mohawk so if that tests asks a question that the individual platform can't handle, and we know that, the test is wrong
18:21 sivoais well, all platforms should handle those types... but there are some assumptions that don't hold
18:21 sivoais (my guess)
18:22 sivoais I'm going to take a look at it and try to reproduce on my Windows box after the changes fix
18:23 mohawk did you see chm's email?
18:23 mohawk do you have PDL 2.007 installed?
18:24 sivoais just saw it. I do not
18:25 sivoais add the PR to the 2.009 milestone
18:25 mohawk i'm just contemplating reinstalling it
18:25 sivoais err, I have added
18:25 mohawk i'll do it now
18:25 mohawk oh, wait
18:25 mohawk does my activeperl have it
18:27 mohawk it doesn't, but i'll put it on there
18:28 mohawk because i'm pretty sure 2.007 wouldn't work on my "perl in space" rig anyway :-)
18:28 mohawk (ap is installed in a non-space place)
18:38 mohawk this is quite something
18:38 mohawk on 2.007, Basic/Core/Makefile ends up with this at the bottom:
18:38 mohawk pdlsections.c: pdlsections.g Types.pm
18:38 mohawk $(PERL) -e "require './getdev.pl'" pdlsections.g > pdlsections.c
18:39 mohawk what you may not be able to tell is that is not a tab, it's spaces
18:39 mohawk dmake doesn't like it
18:40 mohawk that was unexpected
18:40 mohawk anyway, fixing that manually allows it to continue building on AP
18:53 sivoais I don't think it really matters that this follows YAML. Inline and Inline::C don't either
18:54 sivoais because of multiline scalars
18:54 sivoais and the headings aren't maps (they don't have colons after them)
18:54 mohawk sounds fine
18:54 mohawk we'll be dev-releasing anyway
18:55 mohawk so 2.007 blows up on longlong modulus, as in segv
18:55 sivoais eep
18:55 mohawk inorite
18:56 mohawk (the error has been there in the "if dmake" branch of B/C/Makefile.PL since before 2011, when chm fixed some tab->space errors but not that one)
18:57 mohawk actually, i'm wrong
18:58 mohawk it fails the longlong modulus (as does linux 2.007_04)
18:58 mohawk but it blows up after that, just finding where
19:01 mohawk this crashes on AP 5.18.1: C:\test\PDL-2.007_13>perl -MPDL::LiteF -e "all(byte($a) % byte($b)==byte($c)), 'byte modulus'"
19:01 mohawk (on 2.007)
19:02 mohawk so does strawperl 5.20.0 on 2.007_09
19:03 mohawk but it DOESN'T crash on same strawperl, 2.007_13
19:03 mohawk so that's good
19:16 mohawk sivoais, the ops.t thing is still failing
19:16 mohawk but this snippet passes:
19:17 mohawk perl -Mblib -MPDL::Version -MPDL::LiteF -MTest::More=tests,1 -e "cmp_ok indx($INT_MAX*4)%2, '==', 0, 'big indx %';diag qq{'$PDL::Version::VERSION'\n}"
19:17 opkick [pdl] wchristian comment on issue #75: Observation on windows:... http://git.io/hvmX
19:19 sivoais mohawk: is $INT_MAX defined?
19:19 mohawk that would be why
19:20 mohawk that's quite reassuring :-)
19:20 mohawk now it correctly fails
19:20 mohawk so now i can try on 2.007
19:20 mohawk perl -Mblib -MPDL::Version -MPDL::LiteF -MTest::More=tests,1 -e "cmp_ok indx(2147483647*4)%2, '==', 0, 'big indx %';diag qq{'$PDL::Version::VERSION'\n}"
19:21 mohawk yes, same failure on 2.007
19:21 mohawk phew
19:24 opkick [pdl] mohawk2 comment on issue #75: @wchristian, that's valuable.... http://git.io/hv38
19:43 opkick [pdl] mohawk2 pushed 1 new commit to intmaxfix: http://git.io/hvc0
19:43 opkick pdl/intmaxfix fb9880b Ed J: TODO the "big indx/double modulus" tests since fail <= 2.007
19:43 mohawk sivoais, your review please :-)
19:43 mohawk (also waiting on my win32 test + travis, no rush)
20:08 opkick [pdl] coveralls comment on issue #75:
20:10 mohawk so all tests have passed
20:11 mohawk sivoais, you about?
20:13 sivoais mohawk: yeah
20:13 sivoais I'm looking at how to turn these tags into dates for the changelog
20:14 mohawk does the changelog need dates?
20:14 mohawk oh, hang on
20:14 mohawk i'll look at that
20:14 mohawk can you spend 20 secs ok-ing PR 75?
20:14 mohawk https://github.com/PDLPorters/pdl/pull/75/files
20:16 mohawk sivoais: $ git log -1 --pretty="format:%ai" v2.007
20:16 mohawk 2013-10-12 12:56:52 -0400
20:19 opkick [pdl] zmughal comment on issue #75: Looks good. We should merge per @devel-chm's rec to make this a TODO at <http://thread.gmane.org/gmane.comp.​lang.perl.pdl.devel/6373/focus=6374>.... http://git.io/hv06
20:19 sivoais mohawk: oh, thanks, I was using grep and sed :-P
20:20 mohawk i used google
20:22 sivoais haha
20:25 mohawk (and then tweaked it a bit)
20:26 mohawk any scriptlets you make for modifying, i assume you'll capture them on the issue?
20:26 mohawk sivoais, please comment (or make an issue) with any fix ideas!
20:26 mohawk you're teasing us like fermat ;-)
20:27 mohawk while i'm here, any objection to merging corebuild* as well?
20:27 mohawk https://github.com/PDLPorters/pdl/pull/73
20:28 sivoais mohawk: I'd like to test my idea first to make sure that I'm hitting the same code paths
20:28 sivoais I did this in vim!
20:28 sivoais :.!grep -o '[[:digit:]_.]\+' | xargs -I{} echo v{} | xargs -I{} git log -1 --pretty="format:{} \%ai" {}
20:28 mohawk sivoais, fair enough
20:28 mohawk even ideas might inspire or remind someone though
20:29 sivoais that turns each of the "Release Notes for PDL 2.WXY_Z" into "v2.WXY_Z [[date]]"
20:29 mohawk ha, nice
20:29 sivoais not much beyond trying howbig()
20:30 mohawk my earlier attempt at howbig showed no difference
20:30 mohawk anyhoo
20:30 mohawk if you like corebuild, i can merge it with int* and there'll only need one build for both change sets
20:31 sivoais ok, there was also trying to create indx($INTMAX) first then multiplying it
20:31 mohawk noted
20:31 mohawk (the build being when i push to master, in case that's not obvious)
20:32 sivoais I'm looking at the PR
20:32 mohawk awesome
20:34 sivoais eep, that's a little ugly... going down into Basic/Gen, then back up
20:36 mohawk yes
20:36 mohawk but Gen is pure perl
20:37 sivoais mohawk: is this something that can be improved with a newer EU::MM?
20:38 sivoais or is that in EUMM's DEV now
20:38 mohawk which "this" do you mean?
20:38 sivoais the parallel builds in subdirs
20:38 sivoais that you mention in the comment
20:39 mohawk well it's not parallel as such
20:39 mohawk it's serial across subdirs
20:39 mohawk then parallel within each
20:39 mohawk EUMM will, when i make it do so, be able to be parallel across subdirs
20:41 sivoais cool! Well, then I say merge
20:42 mohawk shweet
20:42 opkick [pdl] zmughal comment on issue #73: This looks good to me. A little ugly, but as long as we get this working more nicely with future EU:MM releases, go for it! http://git.io/hv2c
20:44 opkick [pdl] mohawk2 pushed 3 new commits to master: http://git.io/hv2b
20:44 opkick pdl/master 8f8cda2 Ed J: "core" target also needs Math and MatrixOps
20:44 opkick pdl/master 906156b Ed J: Change Core.o dep to just make blib/arch, avoid always recompiling
20:44 opkick pdl/master fc32b6b Ed J: Work around parallel-make problem for "core" target
20:44 opkick [pdl] mohawk2 closed pull request #75: t/ops.t fail on 64-bit Strawberry Perl (master...intmaxfix) http://git.io/he0I
20:45 opkick [pdl] mohawk2 force-pushed corebuild-improve from d0799c8 to fc32b6b: http://git.io/hv2h
20:47 opkick [pdl] mohawk2 deleted intmaxfix at fb9880b: http://git.io/hvaM
20:48 opkick [pdl] mohawk2 deleted corebuild-improve at fc32b6b: http://git.io/hvaS
20:49 mohawk bit of a forest-fire :-)
20:50 mohawk sivoais++
20:58 opkick [pdl] zmughal pushed 4 new commits to changesfix: http://git.io/hvw5
20:58 opkick pdl/changesfix 4ea83c3 Zakariyya Mughal: rename Release_Notes to Changes...
20:58 opkick pdl/changesfix 4ef264d Zakariyya Mughal: remove install notes and warning about PDL C-API changes
20:58 opkick pdl/changesfix bf2922f Zakariyya Mughal: remove PDL v2.007_13 Changes note (not released yet)
21:00 opkick [pdl] zmughal opened pull request #78: Move to standard Changes format (master...changesfix) http://git.io/hvrt
21:12 travis-ci PDLPorters/pdl#120 (master - fc32b6b : Ed J): The build passed.
21:12 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/27764a129141...fc32b6ba1c3c
21:12 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/54486559
21:23 opkick [pdl] mohawk2 comment on issue #78: Fantastic work! http://git.io/hvPw
21:24 mohawk sivoais++
21:26 sivoais I think this should be enough for now. If MetaCPAN can't parse between the vW.XYZ lines, I'll write a patch
21:27 mohawk that's why we have dev releases
21:27 mohawk does json_pp like it?
21:27 sivoais nah, the changelog format in general doesn't map to YAML anyway
21:28 mohawk i just looked at this: https://github.com/CPAN-API/metacpan-web
21:28 mohawk guess who most recent commit is?
21:28 mohawk https://github.com/CPAN-API/metacpan-web/com​mit/27f69c5a4765f3edbd74b57c94d45871df210979
21:28 sivoais yeah, there's a MetaCPAN::Web::Model::API::Changes::Parser
21:28 sivoais heh, Mithaldu!
21:29 sivoais Mithaldu++
21:29 mohawk he amn't here
21:29 mohawk he's EVERYWHERE
21:29 sivoais :-)
21:29 mohawk https://github.com/CPAN-API/metacpan-developer
21:30 sivoais ok, so MetaCPAN deps on <https://metacpan.org/pod/CPAN::Changes>
21:32 sivoais hmm, I'm seeing some invalid releases being parsed with it
21:32 sivoais I'll fix that
21:32 mohawk sivoais++
21:34 mohawk sivoais, i'm going to do a mass rebasing/pushing starting with test-cleanup
21:34 mohawk ok with you?
21:35 sivoais mohawk: go for it!
21:35 sivoais I think there might be a merge conflict on t/ops.t
21:35 sivoais but that shouldn't be hard to fix
21:35 mohawk i can handle it
21:36 sivoais :-)
21:36 mohawk pity mit's not here to come up with a cheesy meme
21:37 mohawk no conflict with t/ops.t itself because i already took that out
21:38 opkick [pdl] mohawk2 force-pushed test-cleanup from f458fa1 to 2e25053: http://git.io/xXCA
21:38 opkick pdl/test-cleanup 84f3fc9 Zakariyya Mughal: convert aaa_load.t to Test::More
21:38 opkick pdl/test-cleanup 9c3f76c Zakariyya Mughal: consolidate test helper function of form `ok( $no, $result )`...
21:38 opkick pdl/test-cleanup f153cee Zakariyya Mughal: done_testing not needed
21:40 mohawk just push -f'ed changesfix, be aware
21:40 opkick [pdl] mohawk2 force-pushed changesfix from db196ee to 274affd: http://git.io/pxve
21:40 opkick pdl/changesfix 274affd Ed J: Changes file maintained by person not dumped from git log
21:40 mohawk (and i cancelled the build on it)
21:41 opkick [pdl] mohawk2 force-pushed doc/vsearch-example from a96743c to f222c25: http://git.io/p7hw
21:41 opkick pdl/doc/vsearch-example 0aa8dd9 Zakariyya Mughal: add vsearch example...
21:41 opkick pdl/doc/vsearch-example f222c25 Zakariyya Mughal: add links to other vsearch_* functions
21:48 opkick [pdl] mohawk2 force-pushed hainest-assgn_badflag from 2f47a52 to 6b24622: http://git.io/xgxg
21:48 opkick pdl/hainest-assgn_badflag 6b24622 Tim Haines: Add propagation of badflag with .= (Ops::assgn) for sf.net bug 3543056.
21:50 opkick [pdl] mohawk2 force-pushed alienhdf4 from 0000530 to 8da63c2: http://git.io/x2TT
21:50 opkick pdl/alienhdf4 8da63c2 Ed J: Use Alien::HDF4 in PDL::IO::HDF build
21:51 opkick [pdl] mohawk2 force-pushed longlong-double-fix from 59bb36a to 4e5b006: http://git.io/p7j1
21:51 opkick pdl/longlong-double-fix efb1c6c Chris Marshall: Implement PDL_Generic for PDL_Anyval type
21:51 opkick pdl/longlong-double-fix 5b60a29 Chris Marshall: Add 8byte array as first member of PDL_Union...
21:51 opkick pdl/longlong-double-fix 1cb9afb Chris Marshall: Add ANYVAL_IS_EQ() macro to replace PDL_Anyval == PDL_Anyval constructs
21:51 opkick [pdl] zmughal pushed 5 new commits to changesfix: http://git.io/hvy1
21:51 opkick pdl/changesfix bd27260 Zakariyya Mughal: rename Release_Notes to Changes...
21:51 opkick pdl/changesfix 4fd820b Zakariyya Mughal: remove install notes and warning about PDL C-API changes
21:51 opkick pdl/changesfix 6c93c4f Zakariyya Mughal: remove PDL v2.007_13 Changes note (not released yet)
21:54 opkick [pdl] zmughal force-pushed changesfix from 64a30b5 to f4e7512: http://git.io/pxve
21:54 opkick pdl/changesfix f4e7512 Zakariyya Mughal: indent some changes on older releases so that CPAN::Changes can parse the file
21:54 mohawk sivoais, do you mind me cancelling the changesfix build?
21:54 mohawk (build(s) in fact)
21:55 sivoais doing so now
21:55 mohawk nice
21:58 sivoais I'll add some info the DEVELOPMENT too
21:58 mohawk dang
21:59 mohawk (if there are tools you're using to check these things, are they going on the PR?)
22:00 sivoais well, there could be, but the thing is ... who runs tests after modifying Changes ;-)
22:00 sivoais we're not using Dist::Zilla... :-P
22:00 mohawk might be useful to the future :-)
22:00 mohawk if you got stuck on your thesis and i needed to do the same checks you did...
22:01 mohawk be nice to cut+paste :-)
22:01 mohawk (including your neato vim line)
22:01 sivoais heh, I can add the info to DEVELOPMENT
22:01 mohawk could even make it a test
22:01 mohawk https://coveralls.io/builds/2118529
22:02 mohawk we're covering 8011 lines of 16346
22:02 sivoais mkdir xt # ?
22:02 sivoais that should be an author test
22:02 mohawk that might be the way forward
22:02 mohawk yes, absolutely
22:03 mohawk of the 8335 we're missing, 2267 are in lib/PDL/Transform/Proj4.pm and 1212 are in lib/PDL/Graphics/PGPLOT/Window.pm
22:03 sivoais actually, a test would be even better than just text in DEVELOPMENT
22:04 mohawk yes it would
22:04 mohawk and travis could run it
22:04 mohawk and MANIFEST.SKIP would have ^xt\/
22:04 mohawk and MANIFEST.SKIP would have ^xt/, rather
22:07 opkick [pdl] coveralls comment on issue #53:
22:15 mohawk i have a yen to make PDL::PP put #line statements in
22:22 * mohawk first wants to increase test coverage so gets stuck into t/proj*
22:27 opkick [pdl] zmughal pushed 2 new commits to changesfix: http://git.io/hvdj
22:27 opkick pdl/changesfix 72b00ce Zakariyya Mughal: make dates in Changes conform to the W3CDTF standard
22:27 opkick pdl/changesfix 2a6ba1e Zakariyya Mughal: add author test to check Changes log...
22:28 sivoais this might fail because CPAN::Changes isn't in .travis.yml
22:28 mohawk was about to say
22:28 mohawk but i don't suppose it'll be run anyway
22:29 sivoais that too
22:29 sivoais I'll add a commit adding both
22:29 mohawk grin
22:29 sivoais actually, travis-perl helpers run author tests if possible by default
22:30 mohawk oh, great
22:30 mohawk i guess that's what $(test-files) achieves, then
22:30 sivoais yes
22:31 opkick [pdl] zmughal pushed 1 new commit to changesfix: http://git.io/hvFK
22:31 opkick pdl/changesfix f4198bd Zakariyya Mughal: Travis-CI: install CPAN::Changes so that author test runs
22:31 sivoais ETOOMANYBUILDS
22:31 mohawk ha ha!
22:31 mohawk ETOOPRODUCTIVE
22:32 mohawk (plus the distro is much too vast as is)
22:32 mohawk 2.009 = changesfix
22:32 mohawk 2.010 = split/bundle distro?
22:33 sivoais sounds reasonable
22:34 sivoais <https://github.com/PDLPorters/pdl/milestones>
22:35 opkick [pdl] coveralls comment on issue #51:
22:49 mohawk installing proj4...
22:56 mohawk and apparently PDL's M.PL can't get it working
22:56 * mohawk sobs
23:04 mohawk ...because despite WITH_PROJ=>undef in perldl.conf, it's set to 0 by the time it gets to Lib/Transform/Proj4/Makefile.PL
23:05 mohawk seems Lib/GIS/Proj/Makefile.PL is also getting involved
23:05 mohawk yes, there are two separate, not-very-related Proj4 related modules
23:26 mohawk and the lesson learned here is, when we install libraries, don't forget to run ldconfig

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