2016-11-22 00:41:50-!- Irssi: #gnupg-devel: Total of 7 nicks [2 ops, 0 halfops, 0 voices, 5 normal]
2016-11-22 00:41:50< gitbot> [2016-11-21 11:54:45] [git] GnuPG - branch master updated by Werner Koch:
2016-11-22 00:41:50< gitbot> [2016-11-21 11:54:45] 362afb4 build: Add repo-only maintainer script append-signature.sh.
2016-11-22 00:41:50<@K_F> [2016-11-21 14:40:33] timezones are fun, but if gniibe gets back when I'm not here, tried out 9a707a223a3d45ccf245dee7989ca144e4e6bb49 and tests in question still fails
2016-11-22 00:41:50<@K_F> [2016-11-21 14:43:57] also --disable-tools has build failure before tests stage, so needs to be fixed outside of skipping tests
2016-11-22 00:41:50<@K_F> [2016-11-21 21:24:11] (although I should note that it is likely not a regression introduced in 2.1.16, I was cleaning up the ebuild to actually disable tools rather than build it unconditionally and only install them if enabled for the version...)
2016-11-22 00:41:50< neal> [2016-11-21 21:26:39] justus: I love TESTFLAGS=--parallel
2016-11-22 00:41:50< neal> [2016-11-21 21:27:34] Given that the output is correctly "interleaved", why can this not be enabled by default?
2016-11-22 00:41:50< gniibe> [2016-11-21 22:48:27] K_F: Thanks for testing.  I will look into other things.
2016-11-22 00:41:50<@K_F> [2016-11-21 22:49:44] gniibe: on another note, although maybe better for IRC, I'm finally trying to use GnuK and NeuG, started with the latter, but it shows up as a mass storage device instead of as ttyACM0, any tips for the udev .rules to correct it?
2016-11-22 00:41:50<@K_F> [2016-11-21 22:50:27] cdc_acm module is loaded..
2016-11-22 00:41:50< gniibe> [2016-11-21 22:52:51] It is intentionally started as a MSD, to convey GPL text.  It also has html and README.
2016-11-22 00:41:50< gniibe> [2016-11-21 22:53:34] Howto stop this mode is also in html.
2016-11-22 00:41:50<@K_F> [2016-11-21 22:54:39] ah, touch ... DROPHERE/foo :) 
2016-11-22 00:41:50< gniibe> [2016-11-21 22:55:27] I wakeup by earthquake today.
2016-11-22 00:41:50<@K_F> [2016-11-21 22:56:04] gniibe: not too bad I hope
2016-11-22 00:41:50< gniibe> [2016-11-21 22:56:56] Tsunami! Radio warns.  No, problem here.
2016-11-22 00:41:50< gniibe> [2016-11-21 22:59:10] No ploblem here.  Sorry, wrong punc.
2016-11-22 00:41:50< gitbot> [2016-11-21 23:10:25] [git] GnuPG - branch master updated by Neal H. Walfield:
2016-11-22 00:41:50< gitbot> [2016-11-21 23:10:25] 037f9de g10: Cache the effective policy.  Recompute it when required.
2016-11-22 00:41:50< gitbot> [2016-11-21 23:10:25] 182efc5 g10: Correctly parameterize ngettext.
2016-11-22 00:41:50< gitbot> [2016-11-21 23:10:25] 7142b29 g10: Don't use the same variable for multiple SQL compiled statements
2016-11-22 00:41:50< gitbot> [2016-11-21 23:10:25] 91a0483 g10: Add a convenience function for checking if a key is a primary key
2016-11-22 00:41:50<@K_F> [2016-11-21 23:27:06] gniibe: hmm, there I got news alert on it, 7.3, yaikes
2016-11-22 00:41:50< gniibe> [2016-11-21 23:49:24] K_F: My place is far from sea (150km), so no problem for Tsunami.  Magnitude 7.3 is not that rare, here in Japan.  On the other hand, we have hot springs.
2016-11-22 00:41:50<@K_F> [2016-11-21 23:51:11] :)
2016-11-22 00:41:50-!- root changed the topic of #gnupg-devel to: No one has the intention of erecting a surveillance state.
2016-11-22 00:41:50< gnupg-devel> werner has set the subject to: No one has the intention of erecting a surveillance state.
2016-11-22 00:41:50-!- Irssi: Join to #gnupg-devel was synced in 0 secs
2016-11-22 03:30:26< gniibe> K_F: I can reproduce same failure when I change GPG definition in tests/migrations/common.scm, so that gpg-agent will be invoked with wrong arg.  I change "|--debug-quick-random" to "|--no-such-option".
2016-11-22 03:30:49< gniibe> K_F: Could you please apply http://paste.debian.net/897511 to see what's going on?
2016-11-22 03:41:33< gniibe> This is better: http://paste.debian.net/897512
2016-11-22 03:43:51< gniibe> Along with this: http://paste.debian.net/897513
2016-11-22 09:46:12<@K_F> 2016-11-22 09:41:52 gpg-agent[22780] socket name '/var/tmp/portage/app-crypt/gnupg-2.1.16/temp/gpgscm-20161122T084152-from-classic-GTecRG/S.gpg-agent.browser' is too long
2016-11-22 09:46:59<@K_F> build and gpg-agent log at https://download.sumptuouscapital.com/gnupg/test-failure-3/
2016-11-22 09:56:22<@K_F> yup, moving build to shorter path and tests works
2016-11-22 09:58:30<@K_F> "gpgscm-20161122T084152-from-classic-GTecRG" uses 43 of a max of 108 chars :|
2016-11-22 09:59:28<@K_F> s/43/42/
2016-11-22 10:04:05<@K_F> gniibe: thanks for helping to debug this, at least now its starting to make sense :)
2016-11-22 10:05:37<@K_F> presumably this can be fixed around line 270 of tests/gpscm/tests.scm
2016-11-22 10:11:46< gitbot> [git] Pinentry - branch master updated by Werner Koch:
2016-11-22 10:11:46< gitbot> d015020 Create SWDB entry during make distcheck
2016-11-22 10:11:46< gitbot> 2fac0e8 Post release updates
2016-11-22 10:11:46< gitbot> 5eb5580 Release 1.0.0
2016-11-22 10:11:46< gitbot> d126036 emacs: Add curses handler stub.
2016-11-22 10:11:46< gitbot> 626819f Update config.{sub,guess} .
2016-11-22 10:18:24<@K_F> https://download.sumptuouscapital.com/gnupg/test-failure-3/0001-gpgscm-Use-shorter-socket-path-lengts-to-improve-tes.patch
2016-11-22 10:19:41< gitbot> [git] gnupg-doc - branch master updated by Werner Koch:
2016-11-22 10:19:41< gitbot> 18c75d8 swdb: Release Pinentry 1.0.0
2016-11-22 10:20:40<@K_F> will have to look into the build failure with --disable-tools later on, real work to do :)
2016-11-22 10:55:23< gitbot> [git] gnupg-doc - branch master updated by Werner Koch:
2016-11-22 10:55:23< gitbot> b121063 web: Announce pinentry 1.0.0
2016-11-22 11:27:08< gitbot> [git] gnupg-doc - branch master updated by Werner Koch:
2016-11-22 11:27:08< gitbot> 1c99eac swdb: release nPth 1.3
2016-11-22 11:28:45< gitbot> [git] Pinentry - annotated tag pinentry-1.0.0 created by tag pinentry-1.0.0
2016-11-22 11:28:45< gitbot> Tagger: Werner Koch <wk@gnupg.org>
2016-11-22 11:28:45< gitbot>  
2016-11-22 11:28:45< gitbot> After 14 years we reached version 1.0.0
2016-11-22 11:28:45< gitbot> -----BEGIN PGP SIGNATURE-----
2016-11-22 11:28:45< gitbot>  
2016-11-22 11:28:45< gitbot> iQEzBAABCAAdFiEE2GkhI8QGXepeDzq1JJs50k8l47YFAlgz+yUACgkQJJs50k8l
2016-11-22 11:28:45< gitbot> 47b43wf+LZn8J9+r1OJJCG1qQLeM0VvfOAxwBTaJxpCyDrC5iXkglGYlXJLrY3FO
2016-11-22 11:28:45< gitbot> mHUuZt+rVu6Vz6vZSR7soxyExE3mh5CJndNg9bs9mn5MMZj1tkRY+OCdIbI0FZBP
2016-11-22 11:28:45< gitbot> g3aLvLMx1fP/y56hdO+9zeJHUK6dpDtDqRqmOundlfKv5H1s53IZThAxMrfPZj0W
2016-11-22 11:28:45< gitbot> Y5OgJx4DMLtSgrfV6djNLNmEuPkm8a6jRcGhMoXqMVAp9M+uvm3fXbyqB2Z1vT8e
2016-11-22 11:28:45< gitbot> qEapMdzKM5EbaOHsd7enghUFqHxoJXm8fyrnRwCTT1/W0g9GGRvonsFerPSRAQfM
2016-11-22 11:28:45< gitbot> oO3XzWn0M83uwtpWVpfeUjqpyZVSCQ==
2016-11-22 11:28:45< gitbot> =Gzjw
2016-11-22 11:28:45< gitbot> -----END PGP SIGNATURE-----
2016-11-22 11:28:45< gitbot>  
2016-11-22 11:28:45< gitbot> Werner Koch:
2016-11-22 11:31:25< justus> K_F: that is why the test suite creates temporary files in /tmp
2016-11-22 11:31:32< justus> why isn't that working for gentoo?
2016-11-22 11:31:44< neal> it looks like tmp is overriden
2016-11-22 11:32:30< justus> i will not commit that kind of patch
2016-11-22 11:33:09< justus> i will not sacrifice debuggability over some stupid unix api limitation from the eighties
2016-11-22 11:33:21< neal> but it is limitation today :-)
2016-11-22 11:33:25< neal> +a
2016-11-22 11:33:28< justus> no it is not
2016-11-22 11:33:37< justus> as i have demonstrated
2016-11-22 11:33:45< justus> but i'm tired of that
2016-11-22 11:33:52< neal> I'm not sure what you mean
2016-11-22 11:34:31< justus> the limitation limits the relative path to the socket, not the absolute one
2016-11-22 11:34:50< justus> so unless basename(path) exceeds 100something characters, there is no problem
2016-11-22 11:34:57< neal> ah
2016-11-22 11:35:22< neal> so what is the problem that k_f is running into?
2016-11-22 11:35:51< justus> the absolute path being larger than 100something chars, and the absolute path being passed to connect
2016-11-22 12:11:04< gitbot> [git] GnuPG - branch master updated by Justus Winter:
2016-11-22 12:11:04< gitbot> a3b258d gpgscm: Refactor.
2016-11-22 12:11:04< gitbot> d8df804 gpgscm: Fix property lists.
2016-11-22 12:11:04< gitbot> 7b4e2ea gpgscm: Fix installation of error handler.
2016-11-22 12:11:04< gitbot> 66834eb gpgscm: Use a static pool of cells for small integers.
2016-11-22 12:11:04< gitbot> 893a3f7 tests: Delay querying the avaliable algorithms.
2016-11-22 12:11:04< gitbot> 6ce14a8 g10: Fix memory leak.
2016-11-22 12:11:48<@K_F> neal: justus TMP path is set to dir within build directory, that is sandboxed 
2016-11-22 12:12:55<@K_F> so TMP env var in this case is /var/tmp/portage/app-crypt/gnupg-2.1.16/temp
2016-11-22 12:15:33< neal> justus: When creating the socket, why not do: chdir(TMP); setenv(TMP="."); create socket; restoreenv(TMP); chdir(-)
2016-11-22 12:16:03< justus> indeed, why not?
2016-11-22 12:16:33< justus> i suggested to do so
2016-11-22 12:16:56< neal> I mean in the code
2016-11-22 12:17:05< neal> like in gpg-agent
2016-11-22 12:17:12< justus> K_F: if we cannot use /tmp then your build environment does not conform to posix
2016-11-22 12:17:36< justus> i'm pretty sure we require a posix-conform environment for building
2016-11-22 12:18:19< justus> 10.1 says
2016-11-22 12:18:19< justus> The following directory shall exist on conforming systems and shall be used as described:
2016-11-22 12:18:19< justus>  
2016-11-22 12:18:19< justus> /tmp
2016-11-22 12:18:19< justus>     A directory made available for applications that need a place to create temporary files. Applications shall be allowed to create files in this directory, but shall not assume that such files are preserved between invocations of the application. 
2016-11-22 12:18:43< neal> You seem pretty upset about this (or something else) for some reason that I don't know / understand.
2016-11-22 12:19:17< neal> I'm not advocating for k_f's patch, but if we can come up with a reasonable technical solution, I think that would be good
2016-11-22 12:20:45< neal> if the solution is for gpg-agent / dirmngr, then this problem will be solved not only for those who are building gpg, but also those who want to use some non-standard GNUPGHOME directory.
2016-11-22 12:59:10< gitbot> [git] GnuPG - branch master updated by NIIBE Yutaka:
2016-11-22 12:59:10< gitbot> 5c2db9d scd: Fix receive buffer size.
2016-11-22 15:26:45< gitbot> [git] GnuPG - branch master updated by Neal H. Walfield:
2016-11-22 15:26:45< gitbot> 44c17bc g10: If the set of UTKs changes, invalidate any changed policies.
2016-11-22 15:48:08< neal> Is the mode parameter to es_fopen the same as to fopen?
2016-11-22 15:50:54< neal> Also, I'm unclear about the portability problem of open
2016-11-22 15:54:58< neal> is es_fopen(xxx, "a") guaranteed to changed the ctime?
2016-11-22 17:12:43< gitbot> [git] GnuPG - branch master updated by Neal H. Walfield:
2016-11-22 17:12:43< gitbot> bfeafe2 g10: Use es_fopen instead of open.
2016-11-23 21:49:02< neal> [2016-11-22 12:20:45] if the solution is for gpg-agent / dirmngr, then this problem will be solved not only for those who are building gpg, but also those who want to use some non-standard GNUPGHOME directory.
2016-11-23 21:49:02< gitbot> [2016-11-22 12:59:10] [git] GnuPG - branch master updated by NIIBE Yutaka:
2016-11-23 21:49:02< gitbot> [2016-11-22 12:59:10] 5c2db9d scd: Fix receive buffer size.
2016-11-23 21:49:02< gitbot> [2016-11-22 15:26:43] [git] GnuPG - branch master updated by Neal H. Walfield:
2016-11-23 21:49:02< gitbot> [2016-11-22 15:26:43] 44c17bc g10: If the set of UTKs changes, invalidate any changed policies.
2016-11-23 21:49:02< neal> [2016-11-22 15:48:08] Is the mode parameter to es_fopen the same as to fopen?
2016-11-23 21:49:02< neal> [2016-11-22 15:50:54] Also, I'm unclear about the portability problem of open
2016-11-23 21:49:02< neal> [2016-11-22 15:54:58] is es_fopen(xxx, "a") guaranteed to changed the ctime?
2016-11-23 21:49:02< gitbot> [2016-11-22 17:12:42] [git] GnuPG - branch master updated by Neal H. Walfield:
2016-11-23 21:49:02< gitbot> [2016-11-22 17:12:42] bfeafe2 g10: Use es_fopen instead of open.