Re: some permissions failures with Archive::Tar::Wrapper
On Wed, Jun 05, 2024 at 12:48:39AM -0300, Alceu Rodrigues de Freitas Junior via cpan-testers-discuss wrote: > Em 04/06/2024 17:07, David Cantrell escreveu: > >busybox tar doesn't mention -p in its docs so I don't think it's fair > >to say that it's broken. It just doesn't support that option. > I got something different here, maybe is related to the version? I was going by busybox's online doco: https://busybox.net/downloads/BusyBox.html I suppose the next stop is https://bugs.busybox.net/ :-) -- David Cantrell | Nth greatest programmer in the world
Re: some permissions failures with Archive::Tar::Wrapper
On 02/06/2024 03:17, Alceu Rodrigues de Freitas Junior via cpan-testers-discuss wrote: Hello there, Looks like the busybox tar from Alpine is indeed broken: localhost:~/.cpan/build/Archive-Tar-Wrapper-0.40-0$ id uid=1000(goku) gid=1000(goku) groups=10(wheel),18(audio),27(video),28(netdev),1000(goku) localhost:~/.cpan/build/Archive-Tar-Wrapper-0.40-0$ tar --version tar (busybox) 1.36.1 localhost:~/.cpan/build/Archive-Tar-Wrapper-0.40-0$ tar -x -v -p -f t/data/bar.tar bar/ bar/bar.dat bar/foo.dat localhost:~/.cpan/build/Archive-Tar-Wrapper-0.40-0$ ls -l bar/ total 8 -rw-r--r-- 1 goku goku 11 Jul 24 2005 bar.dat -rw-r--r-- 1 goku goku 11 Jul 24 2005 foo.dat localhost:~/.cpan/build/Archive-Tar-Wrapper-0.40-0$ tar -tvf t/data/bar.tar drwxrwxr-x mschilli/mschilli 0 2005-07-24 16:15:34 bar/ -rw-rw-r-- mschilli/mschilli 11 2005-07-24 16:15:27 bar/bar.dat -rw-rw-r-- mschilli/mschilli 11 2005-07-24 16:15:34 bar/foo.dat busybox tar doesn't mention -p in its docs so I don't think it's fair to say that it's broken. It just doesn't support that option. -- David Cantrell
Re: Verbose Output in Tests
On 27/09/2023 07:06, Bodo Hugo Barwich wrote: Hi everyone, As author and maintainer of a Perl Module I'm interested in using the great test matrix infrastructure of CPAN-Testers. Unfortunately I see that many test fail in the remote systems http://matrix.cpantesters.org/?dist=Process-SubProcess;os=linux;perl=5.32.1;reports=1 <http://matrix.cpantesters.org/?dist=Process-SubProcess;os=linux;perl=5.32.1;reports=1> while they never failed in any of my automated or manual tests. So, I actually can't reproduce it and also are unable to tell what the issue could be from the summary output Please, how can I see or activate the *verbose testing*? You have to ask the tester. There used to be a Thing on the CPAN-testers website that would tell you their email address but I have no idea if it's still there. However, in this case one of the failures is from me, so I'll test again and send you the detailed output. -- David Cantrell
Re: Getopt::Long::Descriptive fails to load with "undefined symbol: Perl_xs_version_bootcheck"
On 07/08/2023 06:13, Bodo Hugo Barwich wrote: Dear CPAN-Testers, Test Scenario "*Process-SubProcess-2.1.3 5.18.2 GNU/Linux*" fails when trying to load "/Getopt::Long::Descriptive/" as seen on: https://www.cpantesters.org/cpan/report/a62cd144-3498-11ee-98b0-b3c3213a625c <https://www.cpantesters.org/cpan/report/a62cd144-3498-11ee-98b0-b3c3213a625c> " Can't load '/home/cpan/pit/jail/cm2baDMkN6/lib/perl5/x86_64-linux-thread-multi-ld/auto/Params/Util/Util.so' for module Params::Util: /home/cpan/pit/jail/cm2baDMkN6/lib/perl5/x86_64-linux-thread-multi-ld/auto/Params/Util/Util.so: undefined symbol: Perl_xs_version_bootcheck at /home/cpan/pit/jail/cm2baDMkN6/lib/perl5/x86_64-linux-thread-multi-ld/XSLoader.pm line 96. FWIW I see dependency problems with it on 5.18.4 on my tester as well, but in my case it's because Module::CPANfile (a dependency of ExtUtils::MakeMaker::CPANfile) fails its tests. So I recommend not using that :-) When I get rid of that by putting your deps in the Makefile.PL, I see the same errors in the tests. They're because while dependencies have been installed in the tester's build of perl 5.18, you're running some of your tests against whatever random version of perl is at /usr/bin/perl, which is not compatible. XS modules need to be built for a particular version of perl. Specifically, the problem is that t/test_runner.t shells out to bin/run_subprocess.pl, which is executable and has /usr/bin/perl on the shebang line. Instead you should `use Config` and shell out to `$Config{perlpath} ../bin/run_subprocess.pl` to make sure that your tests use the right build of perl. NB that this isn't just a "testers are weird" problem. This will affect any user who doesn't use the system-provided build of perl, such as anyone I've worked for in the last decade and a half. Hope that helps! -- David Cantrell
Re: new releases of cpan-openbsd-smoker
On 19/05/2022 13:52, Alceu Rodrigues de Freitas Junior via cpan-testers-discuss wrote: On 18/05/2022 17:12, Slaven Rezic wrote: Actually I think there's no need to use Task::CPAN::Reporter. Just install plain CPAN::Reporter and make sure to use a http URL in your .cpanreporter/config.ini. This should do it, but it's not ideal since we cannot verify the identity of the reports destination. Oh how awful! Someone other than CPANtesters might find out whether some piece of software works or not! -- David Cantrell
Re: Should we start adding '-Utaint_support' to CPANtesters configurations
On 22/04/2022 07:30, Neil Bowers wrote: I’d hold fire on doing anything for a couple of days – there has been an extra-time objection to my taint change, which may result in it being reverted. I’ll update you here once it’s resolved either way. I thought that objection was just about how the user is asked about it when configuring interactively. But in any case, I have another potential minor objection - but whether it's a p5p thing or a CPAN::Reporter thing I don't know: $ ./blead Tie::Scalar::Decay ... CPAN::Reporter: Test result is 'pass', 'make test' no errors. CPAN::Reporter: preparing a CPAN Testers report for Tie-Scalar-Decay-1.1.1 CPAN::Reporter: sending test report with 'pass' via Metabase ... $ ./blead-no-taint Tie::Scalar::Decay ... CPAN::Reporter: Test result is 'pass', 'make test' no errors. CPAN::Reporter: preparing a CPAN Testers report for Tie-Scalar-Decay-1.1.1 CPAN::Reporter: this appears to be a duplicate report for the test phase: ... Note that CPAN::Reporter can't tell from what's in its list of previously sent reports that this is a different perl built with different options. Building with and without -Utaint_support makes no difference to `archname`, which is what CPAN::Reporter uses to tell the difference between eg threaded and non-threaded, or normal float vs longdouble vs quadmath builds. I'm not sure whether this difference should show up in archname or not, but if it shouldn't then CPAN::Reporter will need an update so it can dig it out of Config.pm. It appears to only exist in config_args. -- David Cantrell
Re: Should we start adding '-Utaint_support' to CPANtesters configurations
On 21/04/2022 22:00, James E Keenan wrote: As of perl-5.35.11, perl offers a configuration option, '-Utaint_support', which enables a user to build a perl executable without taint support. Neil Bowers contacted me with a pull request for a CPAN distribution that I co-maintain, IPC-System-Simple (https://github.com/pjf/ipc-system-simple/pull/39). The purpose of the p.r. was to enable certain test files to PASS if the user attempted to install this distro against a "no-taint perl." Yes. I will do so, as I have a module which relies on taint. > I myself don't do fully automated CPANtesting of the kind that > Andreas, Slaven, Bingos, Carlos and others so unflaggingly do. So I > don't know how much of a human and electronic work burden this might > be. I'm posting here to initiate a discussion. I already test against a build of blead which I update from git roughly weekly. It would be trivial to update the script that rebuilds it to spit out two slightly different builds, so I'll do that this weekend. -- David Cantrell
Re: test failed missing POD::Parser despite a dependency of required POD::Constants
On 29/08/2021 17:34, Slaven Rezic wrote: 17. 08. 2021. u 13:59, Jonas Smedegaard piše: Hi, I am trying to understand the real cause of the test failure here: http://www.cpantesters.org/cpan/report/9420f4a4-feca-11eb-bc66-57723b537541 Error message talks about missing POD::Parser, and yes that module is needed, but should be pulled in since it is a runtime-requirement of POD::Constants which the cpantest states is installed. What am I missing? Hi Jonas, could be something similar like https://rt.cpan.org/Ticket/Display.html?id=132139 Indeed. bin/licensecheck has `#!/usr/bin/env perl` so it's using whichever perl is first in the path, not whichever perl your code is being tested with. Best practice is to instead have `#!perl` as a placeholder shebang line. Then when you `make` (using EU::MM) the path to the correct perl will be filled in for you, in a copy of the script at blib/script/licensecheck, which is what will in turn be installed. Run the version in blib/script in your tests and you should be good. Alternatively you could explicitly invoke your script using the correct perl with something like: use Config; system($Config{perlpath}, 'bin/licensecheck', ...); -- David Cantrell
Re: How to create a tester?
On 08/08/2021 21:32, Felipe Gasper wrote: I’d like to try creating a CPAN testers box. Is there any documentation of how to do this? The wiki is back up - see http://cpanwiki.grango.org/wiki/GettingStarted
Re: How to create a tester?
On 08/08/2021 21:32, Felipe Gasper wrote: I’d like to try creating a CPAN testers box. Is there any documentation of how to do this? It's on the wiki, which, errm, appears to redirect to Barbie's personal site. I've prodded him about it. -- David Cantrell
Re: ExtUtils::CBuilder: Apple have started to apply -Werror by default, breaking XS
[CCed cpan-testers-discuss] On Sun, Nov 01, 2020 at 02:57:04PM +, Alberto Sim?es wrote: > Released a new version of EU::CB > Lets wait for CPAN Testers and see if we have any relevant feedback. > If everything looks fine, I will produce a proper patch including the Changes > file as well. CPAN testers will tell you whether EU::CB passes its own tests or not, but the error this fixes isn't amenable to CPAN-testers finding it and correctly reporting it fixed. The problem is that without this some XS modules that use EU::CB will fail to build. And then the testers will go "oh, it didn't build, I won't report anything at all". That's because most build failures are something like the testing machine not having some C library installed that the XS module depends on and so shouldn't generate a failure report. To verify that this patched release is correct people need to, manually, do this on Mac OS Catalina with the latest XCode: * with the previous version of EU::CB installed, find something XSy that uses it, and which fails to build because of -Werror=implicit-function-declaration * then install this new EU::CB and make sure that the build now succeeds. -- David Cantrell | Enforcer, South London Linguistic Massive
Re: CPAN Statistics API for mswin32 build failures
On 2020-02-14 18:55, Robert Ryley wrote: I'm looking to see if anyone had previously looked for patterns of errors across modules on a particular platform -- especially mswin32. Then you might want to look at Andreas's analysis site http://analysis.cpantesters.org/ -- David Cantrell
Re: cperl in the CPAN Testers Matrix
On Fri, Jun 07, 2019 at 08:45:27PM +0200, E. Choroba wrote: > I've just noticed one of my distribution failed in 5.28.1. When examining > the report, I found it failed under "strict hashpairs". I had no idea what > it was, so I Googled - and found it's a cperl thing. Reading the report > carefully I noticed it was indeed generated by cperl. See > http://www.cpantesters.org/cpan/report/0de93324-8933-11e9-9997-9db8de51d2a1 > > I have no problem with cperl smoking CPAN, but I'm not sure it's a good > idea to include its results among normal Perl versions. What do you think? FWIW I'd quite like to get notifications when my code fails on cperl even if those test results should probably be excluded from aggregates like the number of passes/fails. There's a lot to like about cperl. -- David Cantrell | Cake Smuggler Extraordinaire Your call is important to me. To see if it's important to you I'm going to make you wait on hold for five minutes. All calls are recorded for blackmail and amusement purposes.
Re: CPAN security improvements
On 06/04/2019 03:32, Alceu Rodrigues de Freitas Junior via cpan-testers-discuss wrote: Hello guys, Did you have the chance to read about this backdoor found in a popular Ruby gem? https://www.zdnet.com/article/backdoor-code-found-in-popular-bootstrap-sass-ruby-library/ I was wandering if there is anything we could do to avoid having the same thing happening. Three things stand out from the article. First, the rubygems access of the person who uploaded it was revoked *by the Bootstrap-Sass team*. Implying that he was previously authorised by them. He had co-maint, in PAUSE terms. Second, the package uploaded had a lower version number than what was current at the time, so most users wouldn't have been affected. I *ASS*ume that this means that it was a targeted attack on someone who was using the old version 3.2.0 specifically. Third, it was removed on the same day it was reported, which was only after an individual spotted that something strange had happened. Presumably it would have been spotted quicker if he'd uploaded $LATEST+1. There's nothing we can do about the first of those. The second and third, though, lead me to think that perhaps PAUSE should alert owners/co-maints whenever a package of theirs is uploaded but not indexed - not indexed either because the person uploading it doesn't have rights to that namespace or because it's older than the most recent version. That way the namespace owner can check what was uploaded, and then either discuss it with the PAUSE-admins or with the uploader. I don't think we should go so far as to do something like quarantining the offending dist until one of those owners/co-maints approves it though, mostly because there are legitimate cases where people upload "fixed" versions of other peoples' code where the "fix" is really just a difference of opinion. -- David Cantrell
Re: time to update the CPAN Testers Matrix?
On Wed, Mar 27, 2019 at 05:36:51PM -0700, Karen Etheridge wrote: > > we (well, p5p) should probably delete the entry in that list > > for Darwin (with a capital D) as it is potentially confusing > No, that's the value that's in use today in `uname -s` on OSX, as I > showed in my earlier reply. I know. And people call the platform some variation on "OS X", not "Darwin". -- David Cantrell | even more awesome than a panda-fur coat fdisk format reinstall, doo-dah, doo-dah; fdisk format reinstall, it's the Windows way
Re: time to update the CPAN Testers Matrix?
On 2019-03-27 17:31, Karen Etheridge wrote: Alceu, I don't understand what you're asking. `$^O` returns 'darwin' on MacOS systems. Are you suggesting that *that* be changed? He's saying that the CPAN-testers website should say "Mac OS X" instead of "Darwin", and ... It seems also that the official documentation should be updated regarding Darwin/MacOSX as well: https://perldoc.perl.org/perlport.html#Unix. > `perldoc perlport` is clear that $Config{archname} and $^O are derived > from uname. It would obviously be silly to change the value of $^O for OS X at this stage, but we (well, p5p) should probably delete the entry in that list for Darwin (with a capital D) as it is potentially confusing. Open Darwin barely existed and has been dead for over a decade. I doubt that there is a single person actually using any vaguely recent perl on it. -- David Cantrell | Enforcer, South London Linguistic Massive I think the most difficult moment that anyone could face is seeing their domestic servants, whether maid or drivers, run away -- Abdul Rahman Al-Sheikh, writing on 25 Jan 2004 at http://www.arabnews.com/node/243486
Re: Windows testing/debugging
On Wed, Feb 27, 2019 at 07:40:22AM +0100, Slaven Rezic wrote: > I find debugging with AppVeyor quite tedious --- it requires a fake git > commit for every thing to try and waiting until the AppVeyor system > picks up the change and runs it. The nice people at Appveyor pointed me at this: https://www.appveyor.com/docs/how-to/rdp-to-build-worker/ -- David Cantrell | Bourgeois reactionary pig Only some sort of ghastly dehumanised moron would want to get rid of Routemasters -- Ken Livingstone, four years before he got rid of 'em
Re: Windows testing/debugging
On 2019-02-26 17:32, Karen Etheridge wrote: Is there any documentation on that? `perldoc -f system` only describes the different handling of "system LIST" vs "system PROGRAM LIST" on windows, not any differences between different types of quote characters. I couldn't see anything in perlport either. I know that quoting is weird on Windows when the shell gets involved, but I thought system(LIST) always avoided that. -- David Cantrell | semi-evolved ape-thing For every vengeance, there is an equal and opposite revengeance. -- Cartoon Law XI
Re: Windows testing/debugging
On Mon, Feb 25, 2019 at 11:27:50PM -0800, Serguei Trouchelle wrote: > I looked into the problem, the issue lies in calling system() with > double quotes. Double quotes on Win32 behave different from Unix-like > systems and should generally be avoided. This diff fixes the problem: > > 25,26c25,26 > < [ "\\N{U+2603}", "\\N{U+1F4A9}"], > < [reverse "\\N{U+2603}", "\\N{U+1F4A9}"] > --- > > [ qq{\\N{U+2603}}, qq{\\N{U+1F4A9}}], > > [reverse qq{\\N{U+2603}}, qq{\\N{U+1F4A9}}] Thanks! So simple, but I would never have got that. -- David Cantrell | Reality Engineer, Ministry of Information If you have received this email in error, please add some nutmeg and egg whites, whisk, and place in a warm oven for 40 minutes.
Windows testing/debugging
I don't need to use Andreas's analysis tool to figure out what the common factor is in these test failures :-) http://cpantesters.org/distro/T/Test-Differences.html But I don't have access to any Windows machines, or any knowledge of how to use any of the tools on that platform or how to install stuff. Does anything exist now like the project a few years ago where Microsoft donated some cloudy VMs for use by perl people that already had a sensible toolchain and stuff installed? -- David Cantrell | http://www.cantrell.org.uk/david If you have received this email in error, please add some nutmeg and egg whites, whisk, and place in a warm oven for 40 minutes.
Heads up: Unix::Sudo incoming, it will ask for your password
I am soon going to unleash an abomination, Unix::Sudo, upon an unsuspecting innocent world. It runs a block of code as root: $uid = $<; # ordinary user $uid = sudo { $< }; # root $uid = $<; # back to the ordinary user The tests will ask for your password, and I would be very grateful if you would consider allowing it to run - but please read the source first, for both the module and the tests. https://github.com/DrHyde/perl-modules-Unix-Sudo Alternatively, consider this a warning that you might like to pre-emptively block it from running on your machines :-) -- David Cantrell
Re: First Time CPAN Testing on Windows
On Sat, Aug 25, 2018 at 01:10:15PM -0300, Alceu Rodrigues de Freitas Junior via cpan-testers-discuss wrote: > That being said, I wonder why we don't have something like a "make" > daemon, waiting to receive requests. The processe of doing smoke tests > involves creating a lot of processes that run ony for a short period of > time, at least that is what I can see checking OpenBSD with vmstat > during a smoke test (lots of context switch compared to other uses of CPU). > > I'm not saying that this is something easy to fix, far from it. But > maybe it would worth the effort, I guess. FWIW I wouldn't use a daemon like that, because I strive to make my testing environments as close as possible to those that ordinary users have. -- David Cantrell | Official London Perl Mongers Bad Influence In Victorian times, when every man wore a beard the size of a yew, Britain ruled the world. In the early 20th century, when the beard was trimmed to a moustache, we scraped through two world wars but lost an empire. Today, when Mach3 Turbo multi-blades are the norm, our national pride derives largely from beating the Swedes at Olympic cycling. Grow a beard. Your country needs you.
Re: Tests failing
On Wed, Jul 25, 2018 at 11:07:37AM +0200, Kai Schwarz wrote: > after having reviewed all PrePAN community concerns, I uploaded yesterday our > Module to PAUSE/CPAN. Now we got the 1st CPAN Testers Result Report where I > can see the tests failing. > http://www.cpantesters.org/cpan/report/38a33670-8f54-11e8-8fb1-ef5133556b3f > > Our module is available as WebService::Hexonet::Connector, source code can be > found at https://github.com/hexonet/perl-sdk. > > I am not sure why it fails, but maybe it is not possible to open urls on the > test servers? This will vary from one test machine to another, there's no standard build. > This could be then the reason why the response breaks expected test results. > Of course we need to review this to return at least an alternative error > response. Just my 2 cents.. > > Can someone help us here or at least provide some useful information? First of all, by default you're only notified of failures, not successes. Looking on the CPAN Testers website I can see 17 passes, 10 fails, and 126 'NA' results. The NAs are all on perl versions < 5.26, as specified in your Makefile.PL, so that's fine. The failures are all from Nigel Horne's test boxes, but I also see at least one pass from him as well: http://www.cpantesters.org/cpan/report/84e3a606-8f4e-11e8-9497-9e6cbe9bcd2c Andreas Koenig has a useful tool for trying to spot patterns in failures: http://analysis.cpantesters.org/?author=HEXONET=91.3_xxx=Submit there's nothing there at the moment but I assume it lags a bit behind. -- David Cantrell | Godless Liberal Elitist Anyone who cannot cope with mathematics is not fully human. At best he is a tolerable subhuman who has learned to wear shoes, bathe and not make messes in the house. -- Robert A Heinlein
Re: your reports for Archive-Tar-Wrapper
On Sat, Jul 14, 2018 at 02:52:26PM -0500, Doug Bell wrote: > > On Jul 11, 2018, at 7:24 AM, David Cantrell wrote: > >> On Wed, Jul 11, 2018 at 08:00:59AM +, Alceu R. de Freitas Jr. via > >> cpan-testers-discuss wrote: > >> Maybe I should force a higher version of EU::MM in the > >> Archive::Tar::Wrapper Makefile.PL? > > Yes, I think that makes sense. At least version 6.59. > Is this an issue that we should also detect in CPAN::Reporter, much like it > already detects certain kinds of failed dependency resolutions? Yes, but I don't think it's a high priority, as it only appears to affect some Ye Olde versions of EU::MM. -- David Cantrell | Godless Liberal Elitist
Re: testing beta distributions
On Wed, Jul 11, 2018 at 10:56:53AM +, Alceu R. de Freitas Jr. via cpan-testers-discuss wrote: > I tried to search everywhere but couldn't find if there is an automated way > to test "beta" distributions over CPAN. > Checking the documentation from Perl itself: > https://perldoc.perl.org/perlmodstyle.html#Version-numbering > "If you want to release a 'beta' or 'alpha' version of a module butdon't want > CPAN.pm to list it as most recent use an '_' after theregular version number > followed by at least 2 digits, eg. 1.20_01." > Indeed I cannot even search a beta module with the CPAN client (at least I > don't know how to do it, if it is possible), only list an author directory > for files available. > Is there any option to configure the CPAN::Reporter::Smoker to do that? The reason you can't search for them is because they're not indexed by PAUSE. You can, however, test and install them using the CPAN client, you just need to tell it exactly what file to download: $ cpan ARFREITAS/Some-Distribution-1.2_34.tar.gz It appears from the doco that CPAN::Reporter::Smoker will test trial versions automatically *provided that a non-trial version of the dist exists*. -- David Cantrell | Minister for Arbitrary Justice Guns aren't the problem. People who deserve to die are the problem.
Re: your reports for Archive-Tar-Wrapper
On 2018-07-09 22:05, Slaven Rezic wrote: David Cantrell writes: On Thu, Jul 05, 2018 at 03:55:30PM +, Alceu R. de Freitas Jr. wrote: Despite my efforts, the new release also showed the same error in one of your boxes: http://www.cpantesters.org/cpan/report/30beda64-7f90-11e8-bec4-48217347484a Could you please try to help me in discovering why Log::Log4perl is not being installed in this case? CCing cpan-testers-discuss because I'm stumped here. I think it's EUMM. If I add RUN perlbrew exec --with perl-5.14.4 cpan MSTROUT/ExtUtils-MakeMaker-6.59.tar.gz into the Dockerfile, then it works also with perl 5.14.4. Yep, confirmed. With EU::MM upgraded the dist's dependencies are correctly detected and all tests pass. -- David Cantrell | Enforcer, South London Linguistic Massive On the bright side, if sendmail is tied up routing spam and pointless uknot posts, it's not waving its arse around saying "root me!" -- Peter Corlett, in uknot
Re: your reports for Archive-Tar-Wrapper
On Thu, Jul 05, 2018 at 03:55:30PM +, Alceu R. de Freitas Jr. wrote: > Despite my efforts, the new release also showed the same error in one of your > boxes: > http://www.cpantesters.org/cpan/report/30beda64-7f90-11e8-bec4-48217347484a > Could you please try to help me in discovering why Log::Log4perl is not being > installed in this case? CCing cpan-testers-discuss because I'm stumped here. Here <https://pastebin.com/CFW7Aqs5> is the raw output from running perl-5.14.4/bin/cpan Archive::Tar::Wrapper As you can see it fetches and untars the dist, runs Makefile.PL, which correctly warns about the missing deps and does its usual dance. Here <https://pastebin.com/XRiWqYmd> is the generated MYMETA.yml. But then CPAN.pm doesn't spot that there are dependencies missing and carries on to the 'make' and 'make test' phases, which then blows up. I can't see anything especially weird in Makefile.PL that would cause this. Anyone else got any idea? This only happens in 5.14.4. I have test passes for 5.16 onwards, and also for 5.8.9. In 5.12, Makefile.PL dies thus: Warning: prerequisite IPC::Run 0 not found. Warning: prerequisite Log::Log4perl 0 not found. Warning: prerequisite Test::Simple 1.302073 not found. We have 1.302059. only nested arrays of non-refs are supported at /home/david/cpantesting/perl-5.12.5/lib/5.12.5/ExtUtils/MakeMaker.pm line 664 (/home/david/cpantesting/perl-5.12.5/bin/perl Makefile.PL exited with 65280) So I presume that there's something Odd in the metadata that only affects some combinations of versions of CPAN.pm and EU::MM, but I can't spot what it is. -- David Cantrell | top google result for "topless karaoke murders" Aluminum makes a nice hat. All paranoids will tell you that. But what most do not know Is reflections will show On the CIA's evil landsat.
Re: Archive::Tar::Wrapper makes sense with MS Windows?
On 2018-06-09 23:18, Alceu R. de Freitas Jr. via cpan-testers-discuss wrote: 2018/06/05 11:08:18 C:\WINDOWS\system32\tar.EXE jx -f C:\STRAWB~1\cpan\build\Archive-Tar-Wrapper-0.25-0\t\data\foo.tar.bz2 failed: tar.EXE: Error opening archive: Can't initialize filter; unable to run program "bzip2 -d" Does it make any sense at all to try to make the distro to work with MS Windows? I'm not using Windows for a while already, so I'm not sure if the tar.exe can work together with bzip2 or not. I would assume that GNU tar on Windows works just fine with bzip2 if it's installed. After all, it certainly works with gzip. -- David Cantrell | Minister for Arbitrary Justice Disappointment: n: No results found for "priapic dwarf custard wrestling".
Re: [off-topic] status of Archite::Tar::Wrapper
On Sat, May 26, 2018 at 05:51:39PM -0300, Alceu Rodrigues de Freitas Junior via cpan-testers-discuss wrote: > This is a bit off-topic, but do you guys now the status of > Archive::Tar::Wrapper project? > > I understand it is quite used by us (since it's faster to unpack > tarballs than the alternatives) but it just doesn't work on OpenBSD. > > I sent a Pull Request of a fix for it > (https://github.com/mschilli/archive-tar-wrapper-perl/pull/11) but so > far didn't get any feedback from Mr. Schilli. Github doesn't automatically send notifications when PRs are created, so unless the author is regularly checking for them he won't know that your PR exists. Have you tried emailing him? -- David Cantrell | Nth greatest programmer in the world comparative and superlative explained: worse, worser, worsest, worsted, wasted
Re: testing scripts inside your distro
On Wed, Feb 28, 2018 at 05:10:26PM +, Alceu R. de Freitas Jr. via cpan-testers-discuss wrote: > Hello guys, > I'm in need to test some scripts inside one of my distributions and I'm bit > tired of writing boilerplate code for doing it. > > Maybe you could suggest something available on CPAN for that? My approach is to have the script be mostly a wrapper around more easily-testable modules - the script just wrangles arguments and shows results. So the bulk of the logic can be tested in the usual fashion by testing the underlying modules. Then to test that it wrangles arguments and spits out the right results I use Capture::Tiny's "capture" sub thus: use Capture::Tiny qw(capture); use Config; my($stdout, $stderr, @result) = capture { system( $Config{perlpath}, (map { "-I$_" } (@INC)), qw( blib/script/... ... ) )}; and just make sure $stderr, $stdout, and @result (which contains the exit code reported by system()) are as expected. -- David Cantrell | Reality Engineer, Ministry of Information Repent through spending
Re: Marpa::R2: After 11 days only 2 results
On Tue, Jan 30, 2018 at 11:49:45AM +1100, Ron Savage wrote: > On this page http://matrix.cpantesters.org/?dist=Marpa-R2 there are only > 2 results 11 days after uploading. > > Any ideas? I can only assume that something in the report collection system burped around about the time you uploaded it, because I can see a load of passes (and more importantly no fails!) in my logs. I'll add it into my queueue again so hopefully somethign should come through from that. -- David Cantrell | Bourgeois reactionary pig Us Germans take our humour very seriously -- German cultural attache talking to the Today Programme, about the German supposed lack of a sense of humour, 29 Aug 2001
Re: is QNX relevant to Perl?
On 2018-01-11 17:27, Alceu R. de Freitas Jr. via cpan-testers-discuss wrote: Is QNX relevant for Perl, considering is basically only applicable to BlackBerry... or I'm wrong? QNX used to be fairly popular for real-time and embedded industrial systems, and those sorts of things have loong lifetimes. So even if Blackberry is the only *new* user of QNX there will be plenty of older systems out there. OTOH, they're also the sorts of users who really ought to have their own thorough acceptance tests for software changes, so I doubt they're relying on CPAN-testers to make sure their industrial robots don't fall over embarrassingly. -- David Cantrell | Minister for Arbitrary Justice If I could read only one thing it would be the future, in the entrails of the bastard denying me access to anything else.
Re: Fwd: *** SECURITY information for gateway.bandsman.co.uk ***
On 04/12/2017 18:52, Slaven Rezic wrote: Yes, that's from the test suite of my new module. Unfortunately there does not seem to be a possibility to check whether sudo (password-less) is possible without actually trying to run it and generate this security message. The sudo-using tests make sense in some environments (e.g. on travis-ci with sudo=true), but if testers find it too annoying I can move them to xt/ and run them only on systems where it actually works. If the test is sudo-specific then I don't have anything useful to say, but if it's just that some tests require root then take a look at File::Find::Rule::Permissions for a work-around I put in my tests. Anyone foolish enough to run all the tests as root will run them all, as will travis and my jenkins box, anything else will just skip those tests. -- David Cantrell
Re: *** SECURITY information for gateway.bandsman.co.uk ***
On 04/12/2017 14:21, A. Sinan Unur wrote: On Mon, Dec 4, 2017 at 9:08 AM, Nigel Horne <n...@bandsman.co.uk> wrote: gateway.bandsman.co.uk : Dec 4 08:17:02 : njh : a password is required ; TTY=pts/0 ; PWD=/home/njh/.cpan/build/Doit-0.022-21 ; USER=root ; COMMAND=/bin/true It's used in this code: https://github.com/eserte/Doit/blob/e45a040cbb838bb93b56c229192035eea1d4e95f/lib/Doit.pm#L1958 inside Doit::Sudo::do_connect. Given that it is described as a "scripting engine" it may be a legitimate use, but I'd rather not be running tests that sudo. It's not the only thing that wants to sudo. Every so often something asks me for a password, either to sudo or to ssh to localhost or something along those lines. I just bounce on the enter key or leave it to time out, and don't consider it to be a big problem. -- David Cantrell
CPAN testers website timeouts
Yesterday evening and this morning I've been getting messages similar to this when I try to look at the test failures that I get in my mail: Error 503 first byte timeout first byte timeout Guru Mediation: Details: cache-lhr6336-LHR 1502969464 1081180210 Varnish cache server or Error 503 backend read error backend read error Guru Mediation: Details: cache-lhr6336-LHR 1502969525 1081782213 Varnish cache server I generally get what I wanted after a few retries though. -- David Cantrell | Bourgeois reactionary pig Hail Caesar! Those about to vi ^[ you!
Teeny-tiny usability tweak for website
It would be handy if the columns in the table here: http://www.cpantesters.org/distro/N/Number-Phone.html#Number-Phone-3.4001_02 were sortable. I use this JS library for something similar on my own webshite. It's dead simple and Just Works with minimal effort: https://www.kryogenix.org/code/browser/sorttable/ -- David Cantrell | top google result for "internet beard fetish club" Planckton: n, the smallest possible living thing
Re: Removing/hiding data from CPAN Testers
On Thu, Jul 06, 2017 at 11:40:07AM +1000, Ron Savage wrote: > How much change would result, and how difficult would it be, if you > deleted data pertaining to version of Perl which are no longer supported? Wearing my author hat I very much want to see test results on older versions of perl. Even if p5p doesn't support something I might well do. Wearing my user hat, there are bazillions of systems out there using old versions of perl, for which test results on those old versions are very useful. For example, until March this year the stuff I work in $dayjob was on perl 5.14. While we're now on 5.22, I believe that will go out of p5p-support next year. Whether we upgrade then or not depends on whether we can get rid of a dependency on a dependency on a dependency on Coro by then. Finally wearing my tester hat, if I didn't think that test results from perl 5.8 were useful I wouldn't bother generating them! -- David Cantrell | Reality Engineer, Ministry of Information Repent through spending
Re: Test system configurations
On Tue, Oct 11, 2016 at 11:14:01AM -0400, Dan Collins wrote: > But other modules may work fine with these older versions of the library. > And, without testers reporting these failures, you wouldn't know that > Crypt::PKCS10 is failing on those platforms! > I suspect that you want to probe for the OpenSSL version at the Makefile.PL > stage, and if there is an insufficient version, fail there. Perhaps `openssl version` is what you need. It looks like the format is consistently simple across time: OpenSSL 1.0.1t 3 May 2016 OpenSSL 0.9.7e 25 Oct 2004 Extract the \d+\.\d+\.\d+ bit and compare it to the minimum acceptable in Makefile.PL, and exit(0) before writing Makefile if it's too ancient. -- David Cantrell | Bourgeois reactionary pig Safety tip: never strap firearms to a hamster
Re: Data::Dumper issue: V 2.124 'v' 2.125
On Wed, Sep 28, 2016 at 03:10:58AM +0200, Eirik Berg Hanssen wrote: > Data::Dumper and a whole slew of other modules are specified as > "configure" prereqs. The 'configure' step is when you run Makefile.PL. And because there's no META.yml, the only way to find out what the configure_requires modules are is to run Makefile.PL. By the time that has put the resulting data into Makefile it's too late to actually *use* that data. Change the CONFIGURE_REQUIRES to BUILD_REQUIRES and CPAN clients should DTRT. -- David Cantrell | Minister for Arbitrary Justice "The whole aim of practical politics is to keep the populace alarmed (and hence clamorous to be led to safety) by menacing it with an endless series of hobgoblins, all of them imaginary" -- H. L. Mencken
Re: Strawberry Win32::GetOSVersion seems to be returning invalid values or am I missing something?
On Tue, Apr 12, 2016 at 02:10:15PM -0700, L Walsh wrote: > Well, Instead of limiting myself to "MSWin32" in $^O, I think > I'll check for 'Win' somewhere in the $^O var. Seems like it might > be more forgiving (I hope?) Be very careful with $^O! Checking for 'Win' will miss 'cygwin', and checking case-insensitively will hit 'darwin' - that is, OS X. This is why I wrote Devel::CheckOS. -- David Cantrell | London Perl Mongers Deputy Chief Heretic I hate baby seals. They get asked to all the best clubs.
Online test reports are being HTML-ized
This test report, if you look at the source, contains a load of double-encoded XML noise: http://www.cpantesters.org/cpan/report/00414642-b19f-3f77-b713-d32bba55d77f It's wrapped in a tag to make it use a fixed-width font, but encoded entities like inside are interpreted by browsers so it renders thus: t/vo..junk ' instead. -- David Cantrell | Cake Smuggler Extraordinaire Hail Caesar! Those about to vi ^[ you!
Re: Beware testing Perl6/ distributions ??? CPAN::Reporter::Smoker users should upgrade
On Tue, Mar 01, 2016 at 01:22:35PM -0500, David Golden wrote: > We're now seeing P6 distributions uploaded to CPAN. The convention is that > they are going into an authors "Perl6/" subdirectory. > > Any smokers that smoke based on tailing recent uploads need to be fixed or > configured to skip distributions in that subdirectory to avoid spurious > reports. > > CPAN::Reporter::Smoker has been fixed. Anyone using it should upgrade. In CPAN distroprefs, I believe this is sufficient to ignore anything 6-ish: ~/.cpandistprefs$ cat PSIXDISTS.yml --- comment: "Ignore PSIXDISTS" match: distribution: "PSIXDISTS" disabled: 1 -- David Cantrell | semi-evolved ape-thing Arbeit macht Alkoholiker
Re: Is a mail server required to participate in cpantesters?
On Thu, Oct 08, 2015 at 08:16:39PM -0400, James E Keenan wrote: > I've read recently complaints that because most people submitting > reports are using cpanm-reporter, the reports being received have been > limited in variety. I also know that while Linux and some BSDs are well > represented in the reports, Darwin is surprisingly poorly represented. > (See, for example, http://matrix.cpantesters.org/?dist=File-Path, where > as of this writing there are 0 reports from Darwin.) If only I could find instructions *that actually work* for getting OS X running inside Virtualbox then I would send reports from there. > But, of course, the overwhelming majority of Mac users don't run mail > servers on their laptops or desktops You never needed to run a mail server. You just needed to be able to send mail. Being able to run 'mail' on the command line was sufficient IIRC, and OS X has always had that AFAIK. -- David Cantrell | London Perl Mongers Deputy Chief Heretic Erudite is when you make a classical allusion to a feather. Kinky is when you use the whole chicken.
Re: How to enable tests only when user ask for them?
On Fri, Oct 02, 2015 at 10:18:10AM +0300, Gabor Szabo wrote: > my $test_pw = prompt("Shall I test with password? [y|N]", 'n'); > > This will default to 'n' if this is an unattended test or if > PERL_MM_USE_DEFAULT was set to true. > > What I have not figured out yet is: > How to pass this information to the tests? > How to set an environment variable in Makefile.PL that will be seen during > the "make test" phase? You might want to look at the hoops I jump through for Devel::CheckOS. Depending on some stuff at Makefile.PL-time it writes some test files. You could do that, or drop a file full of test settings somewhere and have your test files read that. -- David Cantrell | Minister for Arbitrary Justice Blessed are the pessimists, for they test their backups
Re: Many FAILs due to unsatisfied prerequisites
On Thu, Jul 09, 2015 at 08:11:40AM -0700, Karen Etheridge wrote: autodie-2.29 is released with the INSTALLDIRS fix (thanks Paul!) -- David, can you please upgrade? Done. Let's hope that that fixes everything. -- David Cantrell | Enforcer, South London Linguistic Massive I think the most difficult moment that anyone could face is seeing their domestic servants, whether maid or drivers, run away -- Abdul Rahman Al-Sheikh, writing on 25 Jan 2004 at http://archive.arabnews.com/?article=38558
Re: Many FAILs due to unsatisfied prerequisites
On Wed, Jul 01, 2015 at 10:51:34AM -0700, Karen Etheridge wrote: Absent evidence to the contrary, I'm going to put this down to something else breaking the toolchain, and you're the one who noticed because your code is toolchain-ish and so interacted with it. Does this mean that your smoker is fixed now? No, it means that I see no evidence that it needs fixing. I've been seeing this issue a *lot*, across a lot of different distributions. Here are some other failures with the same issue: http://www.cpantesters.org/cpan/report/9ec56e7e-c332-11e4-af4e-59047db652c0 http://www.cpantesters.org/cpan/report/50166e4a-c332-11e4-a3e4-48ce7cb652c0 http://www.cpantesters.org/cpan/report/ca4a0030-c3fc-11e4-8ffc-88127db652c0 ... Thanks -- David Cantrell | Godless Liberal Elitist engineer: n. one who, regardless of how much effort he puts in to a job, will never satisfy either the suits or the scientists
Re: Many FAILs due to unsatisfied prerequisites
On Mon, Jun 29, 2015 at 07:44:14PM -0700, Karen Etheridge wrote: I've been continuing to receive reports (several a day, on average) like this: http://www.cpantesters.org/cpan/report/5421bfb0-1e88-11e5-b327-3e6ce14af301 ...where the test failures were clearly caused by unsatisfied prerequisites. It is my understanding that this should not result in a FAIL report, but rather NA -- any reasonable attempt to run tests and achieve a PASS should attempt to satisfy prerequisites first. Here's the full output from running this against a fresh install of 5.8.9. It only differs from stock 5.8.9 in the addition of Expect, CPAN::Reporter, and their pre-reqs: perl-5.8.9/bin/cpan Dist::Zilla::Plugin::StaticInstall 21|tee \ Dist::Zilla::Plugin::StaticInstall.cpan.output http://www.cantrell.org.uk/david/private/dist-zilla-plugin-staticinstall.cpan.output Note the CPAN::Reporter: test results were not valid, Prerequisite missing line, which is missing from the original report. I don't know whether that line is only ever spat to STDOUT/ERR or whether things were subtlely different in this run. If the latter, I can only assume that something else that I tested shortly before your module was a bit buggy and despite passing its tests (and hence getting installed) it broke the toolchain. Sorry about that. Absent evidence to the contrary, I'm going to put this down to something else breaking the toolchain, and you're the one who noticed because your code is toolchain-ish and so interacted with it. -- David Cantrell | top google result for internet beard fetish club On the bright side, if sendmail is tied up routing spam and pointless uknot posts, it's not waving its arse around saying root me! -- Peter Corlett, in uknot
Re: Please add version v5.20.2 to Barnyard CPAN archive
J G Miller wrote: For Barnyard CPAN v5.20.2, some modules appear to be missing - DynaLoader, Config, IO::Handle, Tie::Scalar All four of those are in core and have been since perl 5.003 or earlier. You shouldn't need to fetch them from my mirror. prerequisite module[DynaLoader] not known prerequisite module[Config] not known prerequisite module not known prerequisite module[Tie::Scalar] not known I've seen this myself occasionally on my CPAN testing machines, most commonly something whining about Config not being present. I'm not sure where it comes from, and it's not just confined to my mirrors. I've CCed the CPAN-testers mauling list, maybe someone there will have an Insight. -- David Cantrell Aluminum makes a nice hat. All paranoids will tell you that. But what most do not know Is reflections will show On the CIA's evil landsat.
Re: filtered list of reports for author
On Sun, Mar 15, 2015 at 11:30:13PM +0200, Alexandr Ciornii wrote: I'm planning to develop a tool that will show which modules of an author have problems on latest development and non-development major releases of perl (currently - 5.21.x and 5.20.x). For this I need to fetch list of cpantesters reports for this author for these versions. List of all reports is very large, for ex. for my modules it is 25M ( http://www.cpantesters.org/author/CHORNY.json ). Is it possible to make this list restricted to only perl = 5.x.y? Even better would be to have grouped data like OS/perl/result. AFAIK no such filter exists. A good starting point would be this script: https://github.com/DrHyde/cpXXXan/blob/master/refill-cpanstatsdb.pl It'll suck down any new reports each time you run it, and stuff them into a database so you can slice and dice the data however you like. That script is originally based on one by Andreas: http://repo.or.cz/w/andk-cpan-tools.git/blob/HEAD:/bin/refill-cpanstatsdb.pl And you'll need a database table something like this: https://github.com/DrHyde/CPANdeps/blob/master/tables.sql#L28 -- David Cantrell | Bourgeois reactionary pig The word urgent is the moral of the story The boy who cried wolf. As a general rule I don't believe it until a manager comes to me almost in tears. I like to catch them in a cup and drink them later. -- Matt Holiab, in the Monastery
Re: Description of platforms
On 16/02/2015 16:06, David Golden wrote: Sadly, perlport doesn't have a complete list of values for $^O. A hopefully mostly complete list can be seen here: https://github.com/dagolden/Perl-OSType/blob/master/lib/Perl/OSType.pm#L17 What OS has $^O eq 'gnu'? Hurd? -- David Cantrell | Godless Liberal Elitist There are many different types of sausages. The best are from the north of England. The wurst are from Germany. -- seen in alt.2eggs...
Re: Do I need something special to make the @INC of a child process in a test the same as the parent?
On Sat, Sep 13, 2014 at 03:28:30PM -0400, brian d foy wrote: In the t/compile.t, I call out to perl to do a syntax check, so I figure in the subprocess it doesn't know the @INC of the parent process. (And, I've just read the wiki stuff about $^X, so I'll fix that too). Should I set PERL5LIB with the parent @INC or something similar? You probably want something like this: https://github.com/DrHyde/perl-modules-Devel-CheckOS/blob/master/t/50-script.t#L23-L26 -- David Cantrell | top google result for internet beard fetish club engineer: n. one who, regardless of how much effort he puts in to a job, will never satisfy either the suits or the scientists
Re: Needed platforms for CPAN testers
On Tue, Aug 26, 2014 at 01:49:14PM +1000, Dean Hamstead wrote: Hence I wonder if there is something a matrix that shows deficiencies in testing coverage, which will give people like myself a quick list of platforms we could fire up in a VM (or old hardware we could beg/borrow/steal/ebay) The problem is that the interesting platforms that aren't getting tested are those for which you can't just spin up another VM. In particular I'm thinking of Solaris on Sparc, and Irix. Irix is an especially interesting platform, both for CPAN authors and p5p, because it doesn't so much have a compiler as a complier. Back when I was testing on it I found quite a few dodgy assumptions being made about data types. And one of these days I'll get round to buying another Irix machine. One day. -- David Cantrell | A machine for turning tea into grumpiness There is no one true indentation style, But if there were KR would be Its Prophets. Peace be upon Their Holy Beards.
Re: CPAN indexing of bignum/bigint/bigrat
On Thu, Aug 14, 2014 at 05:05:50PM -0400, Phillip Moore wrote: Sometime in the last year, ownership of the bignum distro changed hands from FLORA to PJACKLAM. On CPAN, FLORA's last release (bignum-0.32) is still present along with PKACJLAM's new one (bignum-0.37). However: $ zgrep bignum mirror/modules/02packages.details.txt.gz bigint 0.32 F/FL/FLORA/bignum-0.32.tar.gz bignum 0.32 F/FL/FLORA/bignum-0.32.tar.gz bigrat 0.32 F/FL/FLORA/bignum-0.32.tar.gz Math::BigFloat::Trace 0.36 P/PJ/PJACKLAM/bignum-0.37.tar.gz Math::BigInt::Trace0.36 P/PJ/PJACKLAM/bignum-0.37.tar.gz All 3 of the big* modules are indexed to the older FLORA release, not the newer PJACKLAM release. It appears that FLORA didn't hand over ownership of big{int,num,rat} but only of Math::Big{Float,Int}::Trace. -- David Cantrell | Bourgeois reactionary pig Safety tip: never strap firearms to a hamster
Re: How do you handle a perl process that never seems to exit?
On Mon, Aug 04, 2014 at 05:18:58PM -0400, Dave Horner wrote: Hanging perl process during test of Farabi and others. running cpan from cmd.exe - cpan shell -- CPAN exploration and modules installation (v2.05) cpan install Farabi ... t/00-compile.t ... ok t/01-basic.t . - The process never returns. Anyways, when smoking and running scripts with perl on windows sometimes I come to my unattended automated machine locked and blocking. No fun. I kill the most childish process and it continues; failed. I find that it happens seldom enough that I can just bounce on C-c. People who prefer a bit more automation would presumably have some kind of automatic time-out. Here's the bare bones of how to do that in the shell. I'm sure you can adapt it to however you're smoking: #!/bin/bash function long_running_process() { sleep 5 echo I didn\'t get killed } long_running_process CHILDPID=$! sleep 2 kill $! $ time ./foo.sh real0m2.008s ... Note the when long_running_process is invoked, so it's forked off. The parent then times out and kills the child before it finishes. -- David Cantrell | top google result for topless karaoke murders comparative and superlative explained: Huhn worse, worser, worsest, worsted, wasted
Re: How do you handle a perl process that never seems to exit?
On Thu, Aug 07, 2014 at 08:29:41AM -0400, Dave Horner wrote: it seems to happen more on windows than other platforms... Unfortunately I know exactly that much about Windows so can't help with that - although I'm sure you could write a watchdog thingy in perl if you don't have a decent shell available. I haven't gone so far as to schedule it to run in the background looking for problematic on my smokers, but i may in the future. I suppose you'd be better off forking twice. Fork once from whatever script controls your smoker *and wait for the child to terminate*. Then in the child, fork off a grandchild to watch the child and kill it if necessary. Of course, the child should, as the last thing it does before exiting normally, kill the grandchild. I suppose an automatic timeout that is set high enough takes care of many issues; but a test could be long running. An hour or so should do the trick. I can't think of anything off the top of my head that takes that long to test, even if you include the time it takes to fetch and test all its dependencies. And hardly anything needs the timeout, so waiting that long shouldn't be a problem. -- David Cantrell | Pope | First Church of the Symmetrical Internet Fashion label: n: a liferaft for personalities which lack intrinsic buoyancy
Re: How do you handle a perl process that never seems to exit?
On Thu, Aug 07, 2014 at 09:12:05AM -0400, Dave Horner wrote: The powershell zombie killer is not a great generic solution though for automated killing. Since looking at the hanging test on AZAWAWI/Farabi-0.47.tar.gz we'd need to look for: C:\strawberry\perl\bin\perl.exe t/01-basic.t which is generic enough that a background watch and kill script would find t/01-basic.t is often used. In Unix-land I'd go by PID, and it would be the 'make test' process that I'd kill, not the individual test script. If there was a way to also filter by the module we are testing; that would make it a bit betterbut I don't see the Farabi string anywhere in the command line arguments of any of the parent perl processes ... Can you also search by the name of the working directory? That would normally contain the distribution name and version - Farabi-0.47 - so could be used to disambiguate. -- David Cantrell | Minister for Arbitrary Justice Sobol's Law of Telecom Utilities: Telcos are malicious; cablecos are simply clueless.
Re: CPAN Testers
On Wed, Jun 11, 2014 at 09:52:24PM +0100, Alberto Simoes wrote: One of Nigel's machines reported on my module Math::GSL: http://www.cpantesters.org/cpan/report/97446734-f1a2-11e3-af0f-6440bde15059 It seems that the problem is that CC is somehow configured as ' cc'. Yes, with that space in the beginning. I am not sure how that space got there. But it is messing around with ExtUtils::CBuilder. I am not sure if: - ExtUtils::CBuilder should be tweaked to strip spaces in the command name... - or if this should never happen, and that Perl needs to be fixed before continuing sending these reports... Note that having spaces in $Config{cc} is valid. Back when I was the maintainer of Devel::CheckLib it came up, the example being 'ccache cc'. You need to split on spaces and build the compiler invocation from the resulting list, as well as $Config{ccflags} or whatever it's called. Possibly re-ordering the flags if you want to add your own. Thank god I don't maintain that code any more :-) I suppose that, theoretically, you might even need to parse something hateful with quoted whitespace in it. Yuck. However, leading/trailing whitespace like in this test report is probably an error. Given that Config.pm is created at perl build time, it might be worth taking up with p5p. -- David Cantrell | Bourgeois reactionary pig 23.5 degrees of axial tilt is the reason for the season
Re: unsatisfied configure_requires in perl 5.12.3?
On Tue, Apr 22, 2014 at 09:46:26AM -0700, Karen Etheridge wrote: On Tue, Apr 22, 2014 at 12:51:02PM +0100, David Cantrell wrote: A 'FAIL' result doesn't mean this code is broken. It doesn't even mean this code is broken on this version of perl and this OS. It means it didn't pass its tests on this particular setup. That's why in cpXXXan I don't pay any attention to failures at all. Are you saying you only look for the presence of a PASS, not the absence of a FAIL, when considering whether to index a distribution for a particular cpXXXan? Yup. Consider, for example, a report like this: http://www.cpantesters.org/cpan/report/03359184-b19f-3f77-b713-d32bba55d77f where the tests failed because they were run on a machine without enough memory available. It isn't possible to automatically and reliably categorise failure reports into valid and invalid^W^W^Wuseful and not useful for users, so I don't even try. But we can be pretty damned certain that *pass* reports are correct. -- David Cantrell | A machine for turning tea into grumpiness Compromise: n: lowering my standards so you can meet them
Re: unsatisfied configure_requires in perl 5.12.3?
On Tue, Apr 22, 2014 at 09:48:14AM -0700, Karen Etheridge wrote: On Tue, Apr 22, 2014 at 04:50:03PM +0300, Alexandr Ciornii wrote: I've investigated this problem (checking perllocal.pod helped) and caught Task-Git-LongList1 doing installation by itself, so even test of this module installs some modules. Ugh, those Task-Git-LongList* modules look like excellent candidates to put in distroprefs as never test. How useless! And Bundle-Git-LongList* as well. Nothing depends on them. -- David Cantrell | Hero of the Information Age Safety tip: never strap firearms to a hamster
Re: unsatisfied configure_requires in perl 5.12.3?
On Tue, Apr 22, 2014 at 07:48:35AM -0400, David Golden wrote: On Mon, Apr 21, 2014 at 11:25 PM, Daniel Staal dst...@usa.net wrote: In general, I'm against pushing that check into the testing framework - there are occasional situations where you can find yourself using truly ancient versions of perl and tools, and I would prefer to have requirements explicitly declared where I can check them than implicitly assumed. I do think that CPAN Testers should generally run a relatively modern toolchain, including CPAN clients. Testing whether something can pass tests on a bog-standard ancient Perl is not really useful data for maintainers because for a long, long time, the answer to I can't install is upgrade your toolchain. It is, however, useful for people wanting to use modules. -- David Cantrell | http://www.cantrell.org.uk/david What a lovely day! Now watch me spoil it for you.
Re: Developer version numbers on cpan testers?
On Mon, Apr 21, 2014 at 12:33:54PM -0400, NebCon, Inc wrote: My tests of LaTeXML are showing warnings like: Argument 2.07_02 isn't numeric in subroutine entry at /home/cpan/pit/bare/perl-5.8.9/lib/5.8.9/File/Temp.pm line 14. usually on OpenBSD systems. This doesn't appear to be possible with the version of File::Temp that was distributed with perl 5.8.9, or with the current version on the CPAN. What version do you have installed? Can you temporarily monkey-patch it to vomit out a stack trace so you can see where it's being called from and with what arguments? -- David Cantrell | http://www.cantrell.org.uk/david I remember when computers were frustrating because they did exactly what you told them to. That seems kinda quaint now. -- JD Baldwin, in the Monastery
Re: Developer version numbers on cpan testers?
On Tue, Apr 22, 2014 at 09:14:43AM -0400, David Golden wrote: On Tue, Apr 22, 2014 at 8:12 AM, David Cantrell da...@cantrell.org.uk wrote: This doesn't appear to be possible with the version of File::Temp that was distributed with perl 5.8.9, or with the current version on the CPAN. What version do you have installed? Can you temporarily monkey-patch it to vomit out a stack trace so you can see where it's being called from and with what arguments? He's complaining about a smoker, not his own machine: /home/cpan/pit/bare/perl-5.8.9/lib/5.8.9/File/Temp.pm I got confused because LaTeXML doesn't show up on metacpan. I see now that search.cpan *does* know about it. Is /home/cpan/pit/ one of your smokers? Nope. -- David Cantrell | Cake Smuggler Extraordinaire What is the difference between hearing aliens through the fillings in your teeth and hearing Jesus in your heart?
Re: deps.cpantesters.org broken?
On Tue, Jan 28, 2014 at 09:16:38PM +, Barbie wrote: If no-one else gets to it before then, I can perhaps help out with the new API at the QA Hackathon. I won't be there I'm afraid as it clashes with a Go tournament. I also may have a server which might be able to accommodate the deps site. Depending on how the cpxxxan sites are configured, I might also be able to help with those too. I've already got free hosting from my lovely employer. However, FYI cpxxxan is just a bunch of static files served up by Apache with a few scripts that get run from cron jobs to update them daily. -- David Cantrell | even more awesome than a panda-fur coat
Re: deps.cpantesters.org broken?
On Tue, Jan 28, 2014 at 10:53:04PM +0100, Andreas Koenig wrote: Me too:) And before that you may be interested in my cronjob refill-cpanstatsdb.pl that feeds the data from Barbie's API to a postgres database. It is in the repo http://repo.or.cz/r/andk-cpan-tools.git Very unpolished but does its job. Ah, excellent. That may be just what I need - instead of rewriting the sites to be vaguely modern/tested/sane, I should be able to just shim that layer in to the teetering tottering pile of shell and perl scripts that has accreted over time :-) -- David Cantrell | top google result for topless karaoke murders
Re: cpantesters system setting AUTHOR_TESTING=1
On Tue, Nov 05, 2013 at 08:11:19PM -0500, Nigel Horne wrote: Prerequisites is set to 'follow' in MyConfig.pm. I have now stopped setting it to follow when AUTHOR_TESTING is set. Has the noise stopped? You could unset AUTHOR_TESTING in the environment and use something like this in a distroprefs file to only turn it on when running the tests for your own code: match: distribution: ^NHORNE/ test: env: AUTHOR_TESTING: 1 See https://metacpan.org/pod/CPAN#Configuration-for-individual-distributions-Distroprefs -- David Cantrell | http://www.cantrell.org.uk/david Repent through spending
Re: Rocky vs CPAN Testers
On Tue, Jul 23, 2013 at 08:50:08PM +0100, Barbie wrote: My thanks to those who have replied to Rocky. I have written a rather lengthy reply (twice) now, but it has been blocked for mediation approval, which it seems Rocky has elected to not do. Interesting, my reply to him appeared on the site immediately, and I didn't know it was even possible for someone to turn on moderation on blogs.perl.org. Are you sure he's blocking you, or are you just suffering from blogs.perl.org being an unreliable piece of crap? -- David Cantrell | Cake Smuggler Extraordinaire While researching this email, I was forced to carry out some investigative work which unfortunately involved a bucket of puppies and a belt sander -- after JoeB, in the Monastery
Re: Failures in network operation
On Fri, May 24, 2013 at 10:42:32AM +0900, Shmuel Fomberg wrote: I released a module, WWW::Github::Files, that let you easily read files and directory from a github repository. I get a lot of test failures, where the first network call the test script does returns with 'Forbidden'. Is there anything I need to know about accessing the network in the tests? Make sure that whatever you're using to access the network knows about proxies. -- David Cantrell | semi-evolved ape-thing Perl: the only language that makes Welsh look acceptable
Re: getting strange reports regarding OS assertion
On Thu, Apr 18, 2013 at 09:14:46AM -0300, Alceu Rodrigues de Freitas Junior wrote: For that, I release a distribution that would to just like Devel::CheckOS ... This is the list of operational systems that would be supported (sub from Siebel::AssertOS): sub os_is { my $os = shift; given ($os) { when ('linux') { return 1 } when ('MSWin32') { return 1 } when ('aix') { return 1 } when ('solaris') { return 1 } when ('hpux'){ return 1 } default { return 0 } } } I would have written Siebel::AssertOS thus: package Devel::AssertOS::OSFeatures::SupportsSiebel; use Devel::CheckOS; sub matches { return qw(Linux AIX Solaris HPUX MSWin32); } sub os_is { Devel::CheckOS::os_is(matches()); } Devel::CheckOS::die_unsupported() unless(os_is()); sub expn { The operating system can run Siebel CRM } -- David Cantrell | even more awesome than a panda-fur coat Computer Science is about lofty design goals and careful algorithmic optimisation. Sysadminning is about cleaning up the resulting mess.
Re: Request for help understanding a FAIL
On Wed, Feb 27, 2013 at 08:53:21PM -0700, David Oswald wrote: The FAIL I'm having trouble understanding is this report for Bytes::Random::Secure: http://www.cpantesters.org/cpan/report/2437bd2e-6bf5-1014-a779-df2c4f8d0f8f It looks from that report like Crypt::Random::Seed is missing. If that's the case, B::R::S shouldn't even be getting tested, since its PREREQ_PM specifies Crypt::Random::Seed as a prereq. Judging from the very long @INC, the tester isn't *installing* modules after testing them, and is running into a long-standing problem with PERL5LIB being too long. -- David Cantrell | Hero of the Information Age People from my sort of background needed grammar schools to compete with children from privileged homes like ... Tony Benn -- Margaret Thatcher
Re: Distroprefs with multiple smokers
On Mon, Feb 25, 2013 at 04:23:03PM -0600, Matthew Musgrove wrote: I was thinking about setting up multiple smokers (at least two: 5.16.2 with and without thread support) on one box. Does anyone have a set up like this that also makes use of distroprefs? If so, how do you manage the distroprefs? I use the same distroprefs on all my smokers, distributed to them all by a small shell script and rsync. You can put conditions in them if you need something to behave differently depending on the build of perl. eg: --- comment: Ignore (ADAMK|TODDR)/IPC-Run-0* on Alpha/BSD because it hangs match: distribution: (ADAMK|TODDR)/IPC-Run-0 perlconfig: archname: alpha-netbsd -- David Cantrell | Reality Engineer, Ministry of Information You can't spell slaughter without laughter
Re: Perl 5.6.x smokers (Bytes::Random::Secure)
On Fri, Feb 01, 2013 at 11:29:34AM -0800, Dana Jacobsen wrote: By the way, I've had the same question, usually with regard to a particular O/S. On some of my modules, the test results just stop coming in, and I'm left wondering what I messed up, with no information about what might have happened. FWIW, I'm sorry I had to stop sending in test results for Irix. I hope to resume later this year. Mind you, I said that about this time last year. -- David Cantrell | Bourgeois reactionary pig Human Rights left unattended may be removed, destroyed, or damaged by the security services.
Re: Ignore lists for a specific PAUSE id + tester
On Fri, Feb 01, 2013 at 08:59:49PM -0800, Serguei Trouchelle wrote: At least two of Nigel's smoker setups generate errors because they use NFS, and everything that uses IPC::Run3 fails because File::Temp cannot guarantee successful unlinking on NFS. Probably, some modules that use File::Temp, may fail too. In my opinion, these results are useless for module authors. I sympathise. However, if Nigel isn't paying attention, it would seem that the best solution would be to file a bug report for File::Temp, that being the root of these problems. -- David Cantrell | even more awesome than a panda-fur coat Blessed are the pessimists, for they test their backups
Re: Ignore lists for a specific PAUSE id + tester
On Sat, Feb 02, 2013 at 01:14:44PM -0500, yary wrote: It would be nice to have something like $OSNAME for filesystem type discovery... Beef up Sys::Filesystem and put it in the core? Or at least encourage its use where helpful... That's tricky. What you're really interested in isn't the filesystem but the filesystem's capabilities. Consider, for example, NFS which ends up on a FAT filesystem has different capabilities from NFS which has an ext4 backend. And in the general case you can't tell what those capabilities are without just trying them. So, riffing off of Devel::CheckOS, you'd want an API something like this ... my $fs = Devel::CheckFS-new($filename); print $filename is on an FS that is case sensitive\n if($fs-supports('CaseSensitive')); and because you want this to work on Windows as well as Unix, you need to jump through all the painful hoops of making things like figuring out what filesystem a file is on portable. I'd love to see someone other than me write this code! -- David Cantrell | Pope | First Church of the Symmetrical Internet THIS IS THE LANGUAGE POLICE PUT DOWN YOUR THESAURUS STEP AWAY FROM THE CLICHE
Re: Ignore lists for a specific PAUSE id + tester
On Mon, Feb 04, 2013 at 09:16:40AM -0800, Marvin Humphrey wrote: On Mon, Feb 4, 2013 at 8:59 AM, David Cantrell da...@cantrell.org.uk wrote: And the mount options won't tell you whether your NFS volume is backed by something sensible like UFS or ext4, or whether the admin has done something crazy like export a FAT filesystem or re-export something from Samba. This is important if you want to know things like whether hard links are supported, or whether it is case-sensitive vs case-preserving vs case-smashing. Say that your CPAN module depends on one of those features, e.g. hard linking. When a CPAN tester runs your module's test suite on a file system where hard linking is not supported, what should happen? Once testing is underway, the only options are to either A) fail or B) skip, possibly obscuring true failures. Is hard linking not a legitimate use case in the judgment of CPAN testers because it is not sufficiently portable? Testers are happy to test modules that are specific to weird legacy proprietary OSes, so portability isn't a problem for us. We merely encourage people to declare their non-portability in advance. Is it the responsibility of module authors to fix their distributions to accommodate the lowest common denominator of file system features? No, I think it's authors' responsibility, if they don't want to fall foul of the tester running the wrong OS or whatever, to declare their pre-requisites, and, if they forget to declare some of them, to be prepared to put up with the occasional test failure report. For something that is so rarely a problem as working hard links, I'd be inclined to not care, no matter whether I'm wearing my testing hat or my author hat, until it becomes a problem for me. As it happens, I do have one project that depends on working hard links, and when it gets to the stage where that matters to my tests, I'll do something about it, and if I'm feeling generous I'll do it in such a way that others can add their own filesystem capability checks. -- David Cantrell | London Perl Mongers Deputy Chief Heretic Graecum est; non legitur
Re: Ignore lists for a specific PAUSE id + tester
On Mon, Feb 04, 2013 at 11:42:20AM -0800, Marvin Humphrey wrote: On Mon, Feb 4, 2013 at 9:49 AM, David Cantrell da...@cantrell.org.uk wrote: Testers are happy to test modules that are specific to weird legacy proprietary OSes, so portability isn't a problem for us. We merely encourage people to declare their non-portability in advance. Well, what about NFS, then? Ultimately it's no different from any other non-perl-module dependency. No, I think it's authors' responsibility, if they don't want to fall foul of the tester running the wrong OS or whatever, to declare their pre-requisites, and, if they forget to declare some of them, to be prepared to put up with the occasional test failure report. How does one declare to CPAN testers that a module does not support NFS? There is no tool yet for doing this easily, but I suggest you start by reading this: http://wiki.cpantesters.org/wiki/CPANAuthorNotes I recommend that you write some code involving Sys::Filesystem to figure out if your tests are trying to use an NFS volume and then either skip tests or bail out completely at Makefile.PL-time. -- David Cantrell | semi-evolved ape-thing Perl: the only language that makes Welsh look acceptable
Re: Ignore lists for a specific PAUSE id + tester
On Sun, Jan 27, 2013 at 09:36:04AM -0800, Karen Etheridge wrote: I have seen some mention of ignore lists - are these for specific combinations of authors + testers? How do I update this list? You don't. The only person who can update an individual tester's blacklists is that tester himself. I have been receiving a series of failure reports, for more than one distribution, where only the configure_requires dependencies have been installed but none of the runtime or testing requirements have been. These reports are nothing but spam, as obviously a distribution is going to fail if its prerequisites have not been properly installed first. If you can point us at a couple of example failure reports, maybe someone here can figure out what's going wrong. In my experience it is almost always the case that things like this are caused by an error in the distribution under test. -- David Cantrell | Bourgeois reactionary pig In My Egotistical Opinion, most people's ... programs should be indented six feet downward and covered with dirt. --Blair P. Houghton
Re: Locating Perl Modules that need extra attention (especially on Windows)
On Mon, Dec 17, 2012 at 06:02:25PM +0200, Gabor Szabo wrote: Recently there were a couple of blogs posts trying to identify the modules that would need the most attention. The solution were focusing on bug reports and recent activity, but I think the data collected by the CPAN Testers could be also used. So in this blog post http://blogs.perl.org/users/gabor_szabo/2012/12/how-to-find-cpan-modules-that-need-help-on-windows.html I asked: Is there a way to fetch all the distributions that have no successful test reports on Windows? Is there a way to create the ratio of (failed test reports)/(all test reports) on MSWin32 for each distribution, and then show the top 100 of that list? http://devel.cpantesters.org/cpanstats.db.bz2 -- David Cantrell | even more awesome than a panda-fur coat Only some sort of ghastly dehumanised moron would want to get rid of Routemasters -- Ken Livingstone, four years before he got rid of 'em
Re: smoking with 5.15, 5.13
On Wed, Jan 02, 2013 at 09:07:32AM -0800, MPR wrote: On Wed, Jan 2, 2013 at 8:06 AM, Andreas Koenig andreas.koenig.7os6v...@franz.ak.mind.de wrote: One might argue the ultimate goal of cpantesters is to find bugs. I'd say yes, but the bugs should be relevant. Bugs in very old and irrelevant combinations are *usually* noise that makes useful work harder for everybody involved. You say usually. Is there a case where it might be useful to test and send reports against an old version? Sure. Even if a module's author doesn't care, the module's users do. And cpXXXan cares deeply. -- David Cantrell | A machine for turning tea into grumpiness Aluminum makes a nice hat. All paranoids will tell you that. But what most do not know Is reflections will show On the CIA's evil landsat.
Re: smoking with 5.15, 5.13
On Fri, Jan 04, 2013 at 10:32:14AM -0500, Ricardo Signes wrote: * Nigel Horne n...@bandsman.co.uk [2013-01-04T09:43:13] On Fri, Jan 04, 2013 at 02:19:43PM +, David Cantrell wrote: On Wed, Jan 02, 2013 at 09:07:32AM -0800, MPR wrote: On Wed, Jan 2, 2013 at 8:06 AM, Andreas Koenig andreas.koenig.7os6v...@franz.ak.mind.de wrote: One might argue the ultimate goal of cpantesters is to find bugs. I'd say yes, but the bugs should be relevant. Bugs in very old and irrelevant combinations are *usually* noise that makes useful work harder for everybody involved. You say usually. Is there a case where it might be useful to test and send reports against an old version? Sure. Even if a module's author doesn't care, the module's users do. +1 ...but who is a user on a year old development version? No-one. And that's not what I was talking about. -- David Cantrell | Bourgeois reactionary pig It wouldn't hurt to think like a serial killer every so often. Purely for purposes of prevention, of course.
Re: CPAN::Reporter is disappearing
On Sat, Dec 15, 2012 at 07:28:19AM -0500, David Golden wrote: On Sat, Dec 15, 2012 at 1:37 AM, Todd Rinaldo to...@cpanel.net wrote: Out of curiosity what do you plan to skip? PrereqCheck tries to load every module to be sure it's installed correctly and working -- that all of *its* prereqs are installed and available. This is part of our attempt to ensure that broken dependency chains aren't reported as FAIL. It already skipped loading some older Module::Install. I've added Acme::Bleach to the skip list. The *version* of Acme::Bleach is still checked through static analysis, but it won't be loaded. I do wonder why Cpanel needs Acme::Bleach. BTW, you should probably also exclude Acme::Pony, Acme::Buffy, Acme::Buckaroo, Acme::EyeDrops, Acme::DoubleHelix, and probably a few others too. -- David Cantrell | semi-evolved ape-thing Longum iter est per praecepta, breve et efficax per exempla.
Re: CPAN::Reporter is disappearing
On Fri, Dec 14, 2012 at 08:04:55PM -0500, Kirk Kimmel wrote: On Thu, Dec 13, 2012 at 7:08 PM, Kirk Kimmel kimmel.k.program...@gmail.comwrote: So I tried 'cpan -t Task::Cpanel' and let it run. PrereqCheck.pm got bleached again so I am working on narrowing down the dependencies to see where exactly this happen. At least I can reproduce this bug now. I replicated the problem twice using 'cpan -t Task::Cpanel'. Each time PrereqCheck.pm got bleached by the time Mail-DomainKeys-1.0.tar.gz asked to enable tests but it was not that dist causing the trouble... http://grep.cpan.me is useful. And it shows up this: https://metacpan.org/source/CPANEL/Task-Cpanel-Internal-11.36.001/lib/Task/Cpanel/Internal.pm#L48 -- David Cantrell | Official London Perl Mongers Bad Influence Repent through spending
Re: Fwd: Help identifying a failure report issue
On Thu, Dec 06, 2012 at 06:36:28PM +0900, Shmuel Fomberg wrote: I'm not an expert, but I think that the .yml file is there for the automatic tools, not for installation. In the installation process only Makefile.PL is used. I'm not sure if that's still the case with a modern CPAN.pm, but it certainly is with the older one that you'll find installed just about everywhere. -- David Cantrell | Enforcer, South London Linguistic Massive If you can read this, thank a teacher. If you're reading it in English, thank Chaucer.
Re: no tests for latest versions of MooX::Shadow::Attributes
On Mon, Dec 03, 2012 at 11:24:23AM -0500, Diab Jerius wrote: I've put out one official distribution of MooX::Shadow::Attributes followed by two dev releases. The official release has been accumulating test results, but the dev releases have been out for a few days and haven't had any test results. The number of results for the old official release is still increasing after the release of the dev distributions. Could someone check if there something weird with the newer dev distros? Looks fine to me, and I've sent reports for 0.01_04 (all PASSes). There are many layers of Magic that reports go through before showing up on the cpantesters.org site and on search.cpan.org, and I see that cpantesters.org is under quite heavy load at the moment, so I presume that they'll show up some time but are in a queue right now. One other weird thing is that if I go to http://www.cpantesters.org/distro/M/MooX-Attributes-Shadow.html version 0.01_03 has a green bar, but no reports. You mean in the left column? Yeah, I see that too. Not sure why. -- David Cantrell | top google result for topless karaoke murders [OS X] appeals to me as a monk, a user, a compiler-of-apps, a sometime coder, and an easily amused primate with a penchant for those that are pretty, colorful, and make nice noises. -- Dan Birchall, in The Monastery
Re: Intermittent failures for Number-Phone
On 29/11/2012 00:04, Slaven Rezic wrote: David Cantrell da...@cantrell.org.uk writes: Can anyone figure out what it is that's Special about these failure reports? Tests fail if Number::Phone::JP is installed. Ah yes, that would explain it. Thanks! -- David Cantrell | Hero of the Information Age Your call is important to me. To see if it's important to you I'm going to make you wait on hold for five minutes. All calls are recorded for blackmail and amusement purposes.
Intermittent failures for Number-Phone
I've had a few FAIL reports for Number-Phone-2.2: http://www.cpantesters.org/distro/N/Number-Phone.html?grade=3perlmat=1patches=2oncpan=2distmat=2perlver=ALLosname=ALLversion=2.2 but also loads of passes, and I can't see what the pattern is. For the failures, it's the same test failing every time, but the failure is in automatically generated code, and the same auto-generated code (just with different data) passes its tests elsewhere. Can anyone figure out what it is that's Special about these failure reports? -- David Cantrell | Godless Liberal Elitist What a lovely day! Now watch me spoil it for you.
Re: cpan.cpantesters.org missing files
On Fri, Nov 16, 2012 at 07:48:34AM +, Barbie wrote: On Thu, Nov 15, 2012 at 04:13:07PM +, David Cantrell wrote: I just spotted that Data-Domain-1.02.tar.gz is missing from http://cpan.cpantesters.org/authors/id/D/DA/DAMI/ IIRC this is one of the fast mirrors - are things occasionally slipping through the cracks? I hope not :( I'll restart the rrr script and see if that picks it up with a full check. That seems to have fixed it. Thanks. -- David Cantrell | Official London Perl Mongers Bad Influence Nuke a disabled unborn gay baby whale for JESUS!
Re: failure installing dependencies
On Sun, Nov 11, 2012 at 11:07:36PM +0900, Shmuel Fomberg wrote: Let's take as a sample case Test::Class. It have a meaningless test fail, a patch waiting in the RT, and a lot of modules depending on it. How can I find the list of modules that directly depend on it? http://deps.cpantesters.org/depended-on-by.pl?module=Test%3A%3AClass -- David Cantrell | Cake Smuggler Extraordinaire Anyone who cannot cope with mathematics is not fully human. At best he is a tolerable subhuman who has learned to wear shoes, bathe and not make messes in the house. -- Robert A Heinlein
cpan.cpantesters.org missing files
I just spotted that Data-Domain-1.02.tar.gz is missing from http://cpan.cpantesters.org/authors/id/D/DA/DAMI/ It is listed in the 02packages.details.txt file, and was uploaded on the 26th of October 2012. IIRC this is one of the fast mirrors - are things occasionally slipping through the cracks? -- David Cantrell | Cake Smuggler Extraordinaire Suffer the little children to come unto me, as their buying habits are most easily influenced. -- Marketroid Jesus
Re: failure installing dependencies
On 9 Nov 2012, at 01:28, Shmuel Fomberg shmuelfomb...@gmail.com wrote: And finally, if a module author upload a new version, and this new version have test failure, I think that it is possible to notify the authors of modules that depend on it without spamming them, but it is really tricky. That's not particularly difficult. CPANdeps already has all the required info, as it has reverse deps too - that is, it can also report on which modules depend on any particular module as well as what dependencies a module has. Using that info to spot failures in your prerequisites should be a fairly simple patch to the scripts that import the cpan-testers data. Just need a web by interface for authors to sign up (probably per dist as well as per author) and some way of keeping track of which dist/version failures have been reported to who. Finally, a link from the reverse deps page to lists of reports. Patches welcome! -- David Cantrell This electrogram was despatched by wireless field telegraph. I would therefore ask that the recipient be so kind as to excuse any failures of courtesy or linguistic inelegance as an unfortunate side-effect of the technology.
Re: Scalar::Util 1.23 openhandle changes
On Sat, Oct 20, 2012 at 11:12:40AM +0200, Kjetil Kjernsmo wrote: I had a bunch of failures, see below, which is caused by the usage of Scalar::Util::openhandle in a new parser in RDF::Trine, typically, I get: Undefined subroutine Scalar::Util::openhandle called at constructor RDF::Trine::Parser::Turtle::Lexer::new (defined at /usr/home/cpan/pit/thr/conf/perl-5.10.1/.cpanplus/5.10.1/build/RDF- Trine-1.001/blib/lib/RDF/Trine/Parser/Turtle/Lexer.pm line 600) line 43. Now, I've discussed this with the author of RDF::Trine, and opened an issue on it: https://github.com/kasei/perlrdf/issues/75 So, as kasei says, openhandle has been there for a long time, and we'd like to understand why this happens now. Can anyone please shed some light on this? I don't really have much to add apart from what's already been discussed there on github. But if you upload a distribution to the CPAN with a _ in its version - eg Test-RDF-1.11_001 - then that will get tested but won't show up in the CPAN index. See if there's anything helpful here: http://analysis.cpantesters.org/solved?distv=Test-RDF-1.11 If I'm reading things right, then your code breaks with Scalar::Util 1.22 and 1.25. -- David Cantrell | Official London Perl Mongers Bad Influence fdisk format reinstall, doo-dah, doo-dah; fdisk format reinstall, it's the Windows way
Re: 4 months of reports missing
On Tue, Oct 16, 2012 at 02:17:50PM -0400, David Golden wrote: Latin-speakers, what's the plural of Mea Culpa? Nostrae culpae (our faults/sins) Meae culpae(my faults/sins) HTH. HArepentfulD. -- David Cantrell | Hero of the Information Age Wow, my first sigquoting! I feel so special now! -- Dan Sugalski
Re: Devel::Trepan smoker failure on v5.16.1 GNU/Linux
On Thu, Sep 06, 2012 at 07:41:22AM -0400, Rocky Bernstein wrote: [FAIL report because of overly-paranoid signature checking?] http://www.cpantesters.org/cpan/report/a2bf7306-f1d1-11e1-93e0-c591eff0cc48 When I run cpansign -v on your dist, I get an error, but it looks like it's because the key server is down, not that there's something wrong with the signature: david@pigsty:~/cpantesting/Devel-Trepan-0.35$ ../perl-5.16.0/bin/cpansign -v Executing gpg --verify --batch --no-tty --keyserver=hkp://pool.sks-keyservers.net:11371 --keyserver-options=auto-key-retrieve SIGNATURE gpg: Signature made Tue Aug 28 03:16:28 2012 BST using DSA key ID 8275EC21 gpg: requesting key 8275EC21 from hkp server pool.sks-keyservers.net gpgkeys: HTTP fetch error 7: couldn't connect: eof gpg: no valid OpenPGP data found. gpg: Total number processed: 0 gpg: Can't check signature: public key not found == BAD/TAMPERED signature detected! == So in this case I think cpansign is correct to complain, and that it's legitimate for the tester to say it didn't pass - although maybe NA would be a more accurate result than FAIL. If you don't want to rely on dodgy third-party infrastructure, get rid of the SIGNATURE file! CCed cpan-testers-discuss because maybe the CPANPLUS test reporting stuff needs fixing; CCed Florian Ragwitz because maybe cpansign needs to use a different server. -- David Cantrell | http://www.cantrell.org.uk/david Computer Science is about lofty design goals and careful algorithmic optimisation. Sysadminning is about cleaning up the resulting mess.
Re: I want to make CPAN::Reporter catch and report errors in the pre-PL phase
On Wed, Aug 08, 2012 at 09:14:51AM -0400, David Golden wrote: This is, of course, tricky. I wouldn't mind seeing an 'optional' CPANTS analysis that does static analysis of dependencies to try to detect cycles ... Apart from dynamic dependencies, it is a SMOP to extract this from CPANdeps. If anyone wants to help out with that, then the data already exists in db/META.yml/*{yml,json} and db/reverse/*.dd, so you just need to load the files and grovel over the resulting data structures. If you just want to do a one-off report, then I expect that it wouldn't hurt too badly if you just scraped the website with URLs like this: http://deps.cpantesters.org/depended-on-by.pl?module=XML::Feed -- David Cantrell | top google result for internet beard fetish club Blessed are the pessimists, for they test their backups
Re: smoke tester hanging during tests
On Wed, Jul 25, 2012 at 09:10:37PM -0300, Alceu Rodrigues de Freitas Junior wrote: Every time that I start the program, it became hanging while executing the following test: Running make test Delayed until after prerequisites Running test for module 'IPC::Pipeline' Running make for C/CP/CPANEL/IPC-Pipeline-0.6.tar.gz I'm not surprised. Anything IPC-ish is a bit flakey on Unix-a-likes too, so I wouldn't expect it to work properly on Windows. -- David Cantrell | semi-evolved ape-thing Arbeit macht Alkoholiker
Re: I would like to search other people's test reports
On Tue, Jul 24, 2012 at 10:38:58AM +0900, Ben Bullock wrote: Some of the error messages from I get from CPAN testers are pretty obscure, like Free to wrong pool or perl: fatal: relocation error. It would be a big help if there was a way to search the other test reports on the site to look for other modules which had the same problem and see how they solved it. google://perl+free+to+wrong+pool At the moment the site seems to be blocked from Google. A search query for site:www.cpantesters.org Free to wrong pool comes up a blank. Is it possible to have some kind of search function on the site? I investigated downloading the database and trying to search it myself but it is dauntingly large. If web search robots are disallowed from the site, is there any way to set up an alternative search? The database doesn't include the reports themselves anyway, just metadata for things like perl version, OS, pass/fail/etc -- David Cantrell | top google result for topless karaoke murders Immigration: making Britain great since AD43
Re: cpantesters CPAN mirror
On Wed, Jul 18, 2012 at 07:47:49PM +0100, Barbie wrote: On Wed, Jul 18, 2012 at 02:36:20PM +0100, David Cantrell wrote: rsyncing from cpantesters.org::CPAN has been timing out for the last few days. Are you using the rrr tool or rsync'ing directly? rsyncing dirctly. -- David Cantrell | http://www.cantrell.org.uk/david The whole aim of practical politics is to keep the populace alarmed (and hence clamorous to be led to safety) by menacing it with an endless series of hobgoblins, all of them imaginary -- H. L. Mencken
Re: Testing Tapper::MCP
On Mon, Jun 11, 2012 at 09:48:43PM +0200, Slaven Rezic wrote: Steffen Schwigon steffen.schwi...@amd.com writes: Is there something else I could do with such long running test suites? Make sure that there's some output, or spit out a warning at the beginning of the time-consuming test file, so that testers can see that it hasn't just hung. Chance of all tests passing for Tapper-MCP's dependencies is: 1.8% (According to http://deps.cpantesters.org/?module=Tapper%3A%3AMCPperl=any+versionos=any+OS ) If you specify 'any version' of perl then it assumes that you mean 5.005 for the purposes of figuring out what's in core, so many of those dependencies won't actually be dependencies on a more modern perl, and it will include test failures on very old perls too. That figure doesn't mean very much anyway: http://deps.cpantesters.org/static/overall-chance.html Proc::ProcessTable is the biggest culprit, and is likely to be quite platform-dependent. The next biggest is String::Diff, which you may be able to get rid of. Finally, the biggest cause of things not getting tested is if they require some external setup (a database, for example, or an environment variable) or there's a non-perl dependency such as on a library or an application. -- David Cantrell | A machine for turning tea into grumpiness Us Germans take our humour very seriously -- German cultural attache talking to the Today Programme, about the German supposed lack of a sense of humour, 29 Aug 2001
Re: Failed: PAUSE indexer report WPMOORE/NetApp-500.001.tar.gz
On Sun, Jun 03, 2012 at 11:55:45PM +0200, Aristotle Pagaltzis wrote: But in the meantime, the gist is: 2. Always use the same version in all files within a distribution. Not practical. I'm not going to manually update the version number in a bunch of files just so I can then automagically generate new versions of Number::Phone::StubCountry::*. -- David Cantrell | top google result for topless karaoke murders I hate baby seals. They get asked to all the best clubs.
Re: Failed: PAUSE indexer report WPMOORE/NetApp-500.001.tar.gz
On Wed, Jun 06, 2012 at 04:45:56PM +0200, Aristotle Pagaltzis wrote: * David Cantrell da...@cantrell.org.uk [2012-06-06 15:40]: On Sun, Jun 03, 2012 at 11:55:45PM +0200, Aristotle Pagaltzis wrote: But in the meantime, the gist is: 2. Always use the same version in all files within a distribution. Not practical. I'm not going to manually update the version number in a bunch of files just so I can then automagically generate new versions of Number::Phone::StubCountry::*. You???ll likely make users suffer at some point. The only way that could happen is if I update something, but forget to update its $VERSION, and then upload it to the CPAN. Your scheme doesn't prevent that from happening. -- David Cantrell | Cake Smuggler Extraordinaire I caught myself pulling grey hairs out of my beard. I'm definitely not going grey, but I am going vain.
bogus FAIL for CPAN-FindDependencies?
http://www.cpantesters.org/cpan/report/25244182-9d04-11e1-a770-889a5de0fff0 defined(@array) is deprecated at /home/src/perl/repoperls/installed-perls/perl/v5.15.9/165a/lib/site_perl/5.15.9/x86_64-linux/PPI/XS.pm line 45. My module doesn't use PPI or XS! And it passes on 5.16.0-RC1, so methinks there's something wrong with Andreas's tester here. -- David Cantrell | Minister for Arbitrary Justice PLEASE NOTE: This message was meant to offend everyone equally, regardless of race, creed, sexual orientation, politics, choice of beer, operating system, mode of transport, or their editor.