Perl 6 - the future is here, just unevenly distributed

IRC log for #native, 2017-01-09

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

All times shown according to UTC.

Time Nick Message
07:55 plicease willthechill: it looks more like a problem with the makefile and the environment to me.
07:59 willthechill okay
08:01 willthechill so how can we move forward on trying to debug it?
08:06 plicease lemmesee if i can reproduce it
08:07 willthechill awesome
08:07 willthechill let me know what I can do to help on my end!  :-)
08:09 willthechill also, a different unrelated failure on one of BinGOs' test machines, I think  it may be caused by a firewall or other network issue, I have written him a private IRC message, still waiting on reply...     http://www.cpantesters.org/cpan/report/21e28342-d591-11e6-ab41-c893a58a4b8c
08:13 plicease yeah that indicates a failure to download.  Unfortunately there isn't much that can be done about that.
08:16 plicease the makefile assumes that USER is defined (which is not true on windows)
08:16 plicease also it assumes that the group is the same as USER which may not be true in unix.
08:16 plicease dumb thing is that it is unecessary
08:16 willthechill okay the USER & GROUP stuff is related to the failure I thought was a bug in Alien?
08:17 willthechill and we're talking about the astyle Makefile?
08:18 plicease correct
08:18 willthechill okay this is the one I'm looking at:  https://github.com/wbraswell/astyle-mirror/blob/master/astyle_2.06_linux/build/gcc/Makefile
08:19 willthechill ah okay yes I see it
08:19 willthechill https://github.com/wbraswell/astyle-mirror/blob/master/astyle_2.06_linux/build/gcc/Makefile#L41
08:19 willthechill INSTALL=install -o $(USER) -g $(USER)
08:20 willthechill you're saying that is the cause of the error?  or just one random bug you happened to find in the Makefile?
08:25 plicease I doubt the person who wrote the makefile intended to run it on windows.
08:25 plicease you can get around it by adding INSTALL=install to the install command
08:26 plicease that builds on my windows box, but the test fails.   In part because -x is not reliable on windows.
08:27 willthechill okay yes I have also been testing somewhat in Windows via AppVeyor, my tests currently fail as well due to -p compiler switch I think
08:27 willthechill however the failure I sent was running in Linux, not Windows?    http://www.cpantesters.org/cpan/report/002509dc-d591-11e6-a7d5-523f3119c05d
08:29 plicease yeah note: "install -o  -g " that looks like USER isn't defined for whatever reason.
08:30 willthechill ah good catch!  :-D
08:31 willthechill install: invalid user ‘-g’
08:31 willthechill yup that's an invalid user & group, alright, haha!  :-P
08:32 willthechill so what was the INSTALL=install thing about?
08:34 plicease https://github.com/plicease/Alien-astyle/commit/3f682ca28ca1f93c4ce4074682aff97b8a7e2370
08:34 plicease That works for me on Strawberry.
08:34 plicease it will probably also cover the linux where USER is not set as well.
08:35 plicease setting INSTALL=install just overides the makefiles default INSTALL which includes the -o $(USER) -g $(USER)
08:35 plicease which is unnecessary.
08:35 willthechill AH cool!
08:35 willthechill :-D
08:35 plicease it is nice that you can do it without patching the makefile :P
08:36 willthechill yeah that is SUPER nice
08:36 willthechill now as for the issue with the guy who can't download...
08:36 willthechill is it possible to use the 'local' keyword instead of 'http' in order to bundle the actual astyle source tarball with Alien::astyle on CPAN?
08:37 willthechill or does it always have to do an actual network download?
08:37 plicease it is an option.  it makes your tarball a lot bigger.
08:38 willthechill *sigh* yeah but failures...
08:38 plicease there are always going to be some failures .
08:38 plicease anyway I am not telling you not to, just explaining the tradeoffs.
08:39 willthechill yeah but failures in Alien::astyle will likely lead to RPerl failures, so I'm trying to achieve 100% coverage with Alien::astyle so as to check it off the list of items which are currently causing me larger RPerl headaches
08:40 plicease The Linux vendors also usually consider it a downside to bundle stuff like that.
08:41 plicease Also I think your test needs some work.  it doesn't seem to work on strawberry.
08:42 willthechill Linux vendors?
08:43 plicease Debian Fedora.
08:43 plicease any with any sense.
08:43 willthechill you mean like if I tried to create an Ubuntu package for Alien::astyle, then the people at Canonical would not like it if I had bundled the astyle source into Alien::astyle?
08:44 plicease I don't think they care if you create a package.  but if they are trying to create a package based on your software then it is a downside
08:45 willthechill hmm well the Linux vendors already have their own astyle packages, so presumably they would never want or need Alien::astyle?
08:45 plicease okay.  But you are making Alien::astyle a prereq for rperl.
08:45 willthechill yes correct
08:46 willthechill and we DO want an RPerl package for Linux vendors eventually
08:46 willthechill and presumably that RPerl package would have the vendor-specific astyle package as a Linux prerequisite, so Alien::astyle would find that `astyle` already exists and thus not need to build the source
08:47 plicease right.
08:47 willthechill which doesn't solve the issue of whether or not to bundle the source
08:47 willthechill :-P
08:47 plicease like I say it is up to you and how OCD you want to be with failures :(
08:47 plicease :)
08:48 plicease just trying to illustrate the tradeoffs that I've run into.
08:49 willthechill yes I definitely appreciate it
08:49 willthechill so is this failure from BinGOs for Alien::m4 also caused by his same machine network connection problem?   http://www.cpantesters.org/cpan/report/ed99a8e8-c77e-11e5-9aa9-0c4caef69d38
08:50 plicease that is some kind of network failure
08:52 willthechill okay then I'm guessing they are linked
08:53 willthechill next question...  what if I did something like create a subclass Alien::astyle::Bundled?
08:53 willthechill could that solve both the Linux vendor issue as well stopping the network failures on CPAN testers?
11:09 willthechill also, please make a pull request on your latest 'use msys on windows' commit...
11:10 willthechill *looks for the button labeled "pull request request"*  ;-)
15:47 awwaiid joined #native
15:48 awwaiid plicease: I just did some Alien promotion in freenode #perl6, planting the seeds a bit :)
16:03 jberger awwaiid++
16:04 jberger in a large part, though Alien::Base could use even more genericization (sp?) it does what most people need most of the time it seems
16:04 jberger I think I've at least proved the concept that preferring config-only (where possible) and plicease's work to allow minor extensions has been pretty successful
16:05 jberger from there it is only the dynamic loader, which I assume p6 has
16:16 awwaiid ya. nice
17:53 plicease awwaiid: saw that thanks!
17:53 plicease interested perl6ers are welcome here on #native too
17:55 plicease awwaiid: is it cool for me to spend a few minutes at the meeting tomorrow (dc,pm) on some pkgconf stuff that I have been working on?
17:55 plicease think abou 20mins should do it and we can do other stuff for the rest of the time.
18:47 awwaiid plicease: uhm, yes? OF COURSE. If you are worried that our schedule is full of other activities you are mistaken.
18:49 plicease it is my deepest darkest feer that I may interfear with something more important.
18:49 plicease seriosuly what I have to talk about can spur further discussion or we can go on to something else.
18:51 awwaiid cool
19:11 felliott` joined #native
19:19 willthechill joined #native
19:22 willthechill plicease: looking at github updates now...  :-)
19:22 plicease sure.  I just updated the PR to be more usable.  You may want to tweak for style.
19:24 willthechill sweet
19:27 willthechill plicease: the last commit turns back on `apt-get install astyle` in .travis.yml, which bypasses the Alien::astyle build...  I had purposefully turned off the Ubuntu install of astyle in order to actually test the build on Travis, why did you disable it?
19:28 plicease I also added ALIEN_INSTALL_TYPE to make sure both get tested.
19:28 plicease it you take a look at the results, you will see the share installs are passing, but the share installs are failing.
19:29 willthechill I assume that last line is a typo?  share installs can't pass AND fail?
19:29 plicease system installs are failing.
19:30 willthechill okay wow yes I see that, it's like Christmas again with the alternating red-and-green icons in Travis, haha!  :-P
19:30 plicease yup :)
19:31 willthechill okay so I assume the "system" installs are the ones using the Ubuntu `apt-get install astyle`, which was working before the current additions, so presumably something in the new code has broken the system builds?
19:31 plicease nope.  new code is fine.  the test is broken.
19:32 plicease for a system install Alien::astyle->bin_dir returns empty list.  as astyle is already in the PATH
19:33 plicease https://metacpan.org/pod/Alien::Base#bin_dir
19:34 willthechill OH I thought bin_dir() was supposed to return a valid path if `astyle` was executable via either system or share
19:35 willthechill so bin_dir() is not to be used in the same way as can_run()
19:35 willthechill my bad!
19:35 plicease no it is a list of directories that should be added to PATH
19:35 plicease np.  It isn't necessarily intuative
19:35 willthechill yeah I just totally assumed wrongly there
21:41 willthechill plicease: do you use appveyor?
22:48 willthechill https://metacpan.org/release/WBRASWELL/Alien-astyle-0.004000
22:50 willthechill good in linux & bsd so far!  :-)    http://fast-matrix.cpantesters.org/?dist=Alien-astyle%200.004000

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