Bug#983239: libbio-db-ncbihelper-perl: (autopkg)test failures when network is available
Hi Aaron, Hi Gregor, Aaron M. Ucko, on 2021-02-21 17:57:37 -0500: > Étienne Mollier writes: > > > I have been looking in the issue below in the package > > libbio-db-ncbihelper-perl. If I understood correctly, the main > > point of the package is to rely on resources made available on > > the Internet. > > I'm not sure I've been personally involved with this package, but that's > my understanding as well. > > > a perhaps magic index to refer to human genome, but maybe it is > > a "well known index".) > > Per [1], taxonomic ID *numbers* are stable in general, but the > associated *names* occasionally change to reflect improved > understandings of the underlying science. AFAICT from [2] (as linked > from [3]), this change is correct and legitimate; moreover, it looks > like 'Actinobacteria' should now appear in $n->common_names, if anyone > wants to verify that. > > [1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7408187/ > [2] > https://www.microbiologyresearch.org/content/journal/ijsem/10.1099/ijsem.0.003920 > [3] > https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mode=Info=1760=3=f=1=1 Thank you very much! I humbly have to admit I know nothing to taxonomy classification, but it's nice to know that people try to keep improving one of the references. This was enough to brighten my day. :) Will see how upstream changes the entry, but given our thoughts, I would guess updating the expected $n->scientific_name with the new value would be the way to go. Out of curiosity, from the $n->common_names I obtained the character string: authorityActinomycetia (Stackebrandt et al. 1997) Salam et al. 2020 gregor herrmann writes: > I see your point, and in the end it's a matter of taste, I guess; or > a choice between two bad options: > - skipping tests and missing bugs > - enabling tests and having to deal with failures because of internet > problems, server problems, changes in returned data etc. > > Having seen too much of the latter, I prefer the former but as I > said, that's not the only option :) In doubt, since the package is team maintained and developer time is precious, I believe I will defer to your experience. I replace the "needs-internet" by "superficial", and set NO_NETWORK_TESTING=1 in debian/tests/pkg-perl/smoke-env. Hopefully if further offline testing became possible, that would allow us to remove the "superficial" flag in the future. > As a side note to the autopkg-tests: The "heavy" tests don't exists, > that was just an idea when we started with the framework, so you can > remove the last paragraph. Thanks for the side note, I removed the entry. Thanks again to you both for your time, Kind Regards, -- Étienne Mollier Fingerprint: 8f91 b227 c7d6 f2b1 948c 8236 793c f67e 8f0d 11da Sent from /dev/pts/2, please excuse my verbosity. signature.asc Description: PGP signature
Bug#983239: libbio-db-ncbihelper-perl: (autopkg)test failures when network is available
Étienne Mollier writes: > I have been looking in the issue below in the package > libbio-db-ncbihelper-perl. If I understood correctly, the main > point of the package is to rely on resources made available on > the Internet. I'm not sure I've been personally involved with this package, but that's my understanding as well. > a perhaps magic index to refer to human genome, but maybe it is > a "well known index".) Per [1], taxonomic ID *numbers* are stable in general, but the associated *names* occasionally change to reflect improved understandings of the underlying science. AFAICT from [2] (as linked from [3]), this change is correct and legitimate; moreover, it looks like 'Actinobacteria' should now appear in $n->common_names, if anyone wants to verify that. [1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7408187/ [2] https://www.microbiologyresearch.org/content/journal/ijsem/10.1099/ijsem.0.003920 [3] https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mode=Info=1760=3=f=1=1 > In any case, I though upstream would be interested, so I took > the liberty to open an issue in their VCS. Thanks! gregor herrmann writes: > I still think that NO_NETWORK_TESTING=1 should be set in debian/rules > to make sure there's no internet access attempted during the build, > as that is a policy violation. Right, we're just discussing what to do about the autopkgtest. -- Aaron M. Ucko, KB1CJC (amu at alum.mit.edu, ucko at debian.org) http://www.mit.edu/~amu/ | http://stuff.mit.edu/cgi/finger/?a...@monk.mit.edu
Bug#983239: libbio-db-ncbihelper-perl: (autopkg)test failures when network is available
On Sun, 21 Feb 2021 22:23:20 +0100, Étienne Mollier wrote: > As the purpose of the package is to fetch resources on the web, > I did something better described by the following commit[1]. > > [1] > https://salsa.debian.org/med-team/libbio-db-ncbihelper-perl/-/commit/bde35697e319417ae17367d51aaa868c43e1715d > > It looked to me like the most sensible thing to do, since the > entire set of tests part of smoke are otherwise skipped. I see your point, and in the end it's a matter of taste, I guess; or a choice between two bad options: - skipping tests and missing bugs - enabling tests and having to deal with failures because of internet problems, server problems, changes in returned data etc. Having seen too much of the latter, I prefer the former but as I said, that's not the only option :) I still think that NO_NETWORK_TESTING=1 should be set in debian/rules to make sure there's no internet access attempted during the build, as that is a policy violation. (Sorry if that's already the case for _this_ package, I might confuse the 3 libbio.*-perl packages I looked at today and which all have a similar situation.) As a side note to the autopkg-tests: The "heavy" tests don't exists, that was just an idea when we started with the framework, so you can remove the last paragraph. Cheers, gregor -- .''`. https://info.comodo.priv.at -- Debian Developer https://www.debian.org : :' : OpenPGP fingerprint D1E1 316E 93A7 60A8 104D 85FA BB3A 6801 8649 AA06 `. `' Member VIBE!AT & SPI Inc. -- Supporter Free Software Foundation Europe `- NP: Peter, Paul and Mary: Leaving On A Jet Plane signature.asc Description: Digital Signature
Bug#983239: libbio-db-ncbihelper-perl: (autopkg)test failures when network is available
Control: tag -1 upstream Control: forwarded -1 https://github.com/bioperl/Bio-DB-NCBIHelper/issues/6 Hi Aaron, I have been looking in the issue below in the package libbio-db-ncbihelper-perl. If I understood correctly, the main point of the package is to rely on resources made available on the Internet. On Sun, 21 Feb 2021 14:05:39 +0100 gregor herrmann wrote: > not ok 44 > > # Failed test at t/Taxonomy.t line 102. > # got: 'Actinomycetia' > # expected: 'Actinobacteria' For reference, I see along those lines: 101 ok $n = $db->get_Taxonomy_Node('1760'); 102 is $n->scientific_name, 'Actinobacteria'; I would think the test 44 in t/Taxonomy.t is fooled following an update in the resource the package fetches on the Internet (https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi if I trust the package description). Hence I would be tempted to think the kind of bacteria is not necessarily that important, but that a character string could be fetched. But I'm not at all a specialist of the field so prefer referring to your expert view. (and I am also concerned that the previous tests rely on a perhaps magic index to refer to human genome, but maybe it is a "well known index".) In any case, I though upstream would be interested, so I took the liberty to open an issue in their VCS[0]. [0] https://github.com/bioperl/Bio-DB-NCBIHelper/issues/6 Gregor, Many Thanks for having spotted this in the first place! As the purpose of the package is to fetch resources on the web, I did something better described by the following commit[1]. [1] https://salsa.debian.org/med-team/libbio-db-ncbihelper-perl/-/commit/bde35697e319417ae17367d51aaa868c43e1715d It looked to me like the most sensible thing to do, since the entire set of tests part of smoke are otherwise skipped. I am not sure if there were more simple ways to approach that. I am concerned that all test are skipped and considered valid. Here is what I see if I smoke-env NO_NETWORK_TESTING=1: autopkgtest [18:43:55]: test autodep8-perl-build-deps: /usr/share/pkg-perl-autopkgtest/runner build-deps autopkgtest [18:43:55]: test autodep8-perl-build-deps: [--- t/EntrezGene.t . skipped: NO_NETWORK_TESTING t/GenBank.t skipped: NO_NETWORK_TESTING t/GenPept.t skipped: NO_NETWORK_TESTING t/Query-Genbank.t .. skipped: NO_NETWORK_TESTING t/RefSeq.t . skipped: NO_NETWORK_TESTING t/Taxonomy.t ... skipped: NO_NETWORK_TESTING Files=6, Tests=0, 0 wallclock secs ( 0.02 usr 0.00 sys + 0.07 cusr 0.00 csys = 0.09 CPU) Result: NOTESTS autopkgtest [18:43:55]: test autodep8-perl-build-deps: ---] autopkgtest [18:43:55]: test autodep8-perl-build-deps: - - - - - - - - - - results - - - - - - - - - - autodep8-perl-build-deps PASS I'm sorry, I'm not sure that was that much of a Low Hanging Fruit, but that might do for this month's session hopefully. :) Thanks for your thoughts, -- Étienne Mollier Fingerprint: 8f91 b227 c7d6 f2b1 948c 8236 793c f67e 8f0d 11da Sent from /dev/pts/8, please excuse my verbosity. signature.asc Description: PGP signature
Bug#983239: libbio-db-ncbihelper-perl: (autopkg)test failures when network is available
Source: libbio-db-ncbihelper-perl Version: 1.7.6-2 Severity: serious Justification: autopkgtest is relevant, network access is a policy violation -BEGIN PGP SIGNED MESSAGE- Hash: SHA512 As seen on ci.d.n: https://ci.debian.net/data/autopkgtest/testing/amd64/libb/libbio-db-ncbihelper-perl/10589853/log.gz we have a test failure: not ok 44 # Failed test at t/Taxonomy.t line 102. # got: 'Actinomycetia' # expected: 'Actinobacteria' This happens because internet access is available, and internet using tests are disabled during not disabeld. I think they are supposed to be disabled during build, but this doesn't seem to work either: t/EntrezGene.t . skipped: no host: www.google.com t/GenBank.t skipped: no host: www.google.com t/GenPept.t skipped: no host: www.google.com t/Query-Genbank.t .. skipped: no host: www.google.com t/RefSeq.t . skipped: no host: www.google.com t/Taxonomy.t ... skipped: no host: www.google.com means that libtest-requiresinternet-perl doesn't find internet access (expected in my chroot), whereas it would output something different if NO_NETWORK_TESTING was respected. (Also: what is --network supposed to do?) With override_dh_auto_test: NO_NETWORK_TESTING=1 dh_auto_test we get: t/EntrezGene.t . skipped: NO_NETWORK_TESTING t/GenBank.t skipped: NO_NETWORK_TESTING t/GenPept.t skipped: NO_NETWORK_TESTING t/Query-Genbank.t .. skipped: NO_NETWORK_TESTING t/RefSeq.t . skipped: NO_NETWORK_TESTING t/Taxonomy.t ... skipped: NO_NETWORK_TESTING Summary: - - the tests try to go out to the internet (policy violation during build) - - if this succeeds the build fails (which doesn't happen in restricted chroots) - - autopkgtests fail - - setting NO_NETWORK_TESTING=1 in d/rules help - - doing the same in debian/tests/pkg-perl/smoke-env fixes the autopkgtest failure Cheers, gregor - -- System Information: Debian Release: bullseye/sid APT prefers unstable APT policy: (990, 'unstable'), (500, 'unstable-debug'), (500, 'oldoldstable'), (500, 'experimental'), (500, 'testing'), (500, 'stable'), (500, 'oldstable') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 5.10.0-3-amd64 (SMP w/4 CPU threads) Kernel taint flags: TAINT_OOT_MODULE, TAINT_UNSIGNED_MODULE Locale: LANG=C, LC_CTYPE=de_AT.utf8 (charmap=UTF-8), LANGUAGE not set Shell: /bin/sh linked to /bin/bash Init: sysvinit (via /sbin/init) LSM: AppArmor: enabled -BEGIN PGP SIGNATURE- iQKTBAEBCgB9FiEE0eExbpOnYKgQTYX6uzpoAYZJqgYFAmAyWqJfFIAALgAo aXNzdWVyLWZwckBub3RhdGlvbnMub3BlbnBncC5maWZ0aGhvcnNlbWFuLm5ldEQx RTEzMTZFOTNBNzYwQTgxMDREODVGQUJCM0E2ODAxODY0OUFBMDYACgkQuzpoAYZJ qgZpqg//QFq38bzNVQzMtyxMbWHQ7rHWERS2IcdGdsu0PIkluPTX86dn/XOvGZgG XlTSJVlhoaZLU9/+O6D+IGUTMGrMCMskNW7sDfB3Q5eTISCCf25ua93iXQaD8iz9 HFScCQx/oc9lAhXyxZwa9qvgMOjNkyFGvz0chjJhqhwsBp9o7mnFFwMd9a675xgv CgXSecGvbkO2ri951Qj42rHblEC1GI1dh12aDZKPuopFkmF43TT5Encem5Ekbw5N ShP2LJsw9+kJ0MEnzClHSYZSetrkqDise/bgkC0icxARzFz7FZVsskwJQq/fx4jm JUjTtiZTDPZlFym3bw2eN2x46187UpdEYF7x1Ez6RxxwUMYOxdZTWf+EzGkXgnQA 5Nd8PC907tzHYsz3U17ABIRZn1pploxwn4S0Rt58m/Z1d4UIhywpdJPw8Ioq3o+6 KvWqdO2mWQxOX0UucvURSNjlMqyX3qifqPvj8yCQRBuPEu2oL3ai3vlN6rzuYVq4 qgWoWofLe+l6OUFt95ZSzRfUw7E9n+XJGjrfcL5hkvdmRa73LAWDkLcgyyqcB7th nwshKOA7h1/IvRSdQJbBenddEL6UtsE6auhkw0300ifF3PHhVxwz1H7zDTKF1JLx z9Bmlqh9cjNx+G4wyXK5LZA83NCgsF9vSzOBwb1KLtc/lx9OgUI= =abvI -END PGP SIGNATURE-