cpet/openbsd-ports
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
100,656 Commits 1 Branch 0 Tags
Commit Graph

587 Commits

Author SHA1 Message Date
espie
70de7e8c72 fix creation of path log files 2015-05-01 20:49:46 +00:00
espie
d40042f30e lock_user is not needed, defaults to log_user for all practical cases 2015-05-01 19:42:54 +00:00
espie
2c469073c2 give size logs to log_user. 2015-05-01 18:27:43 +00:00
espie
795a501ff4 remove debug scaffolding. Found the bug (elsewhere, not surprisingly) 2015-05-01 15:41:52 +00:00
espie
f8d8faa9ef slight cleanup 2015-05-01 09:03:20 +00:00
espie
ad230f14e0 tweak the tweak 2015-04-30 21:43:28 +00:00
espie
5296a6691e a few more "tweak the user" for fetch handling 2015-04-29 13:28:36 +00:00
espie
388883bea1 display more junk info for now, trying to track down things 2015-04-29 10:59:03 +00:00
espie
017830ee1c blabla that doesn't matter 2015-04-29 08:18:42 +00:00
espie
1f51a7929e in a bizarre twist, since we store umask an an octal string, we need 
to convert it back to use the built-in properly.
 2015-04-27 17:01:37 +00:00
espie
c148e02ce4 more or less "the final" dpb pieces for fetching as a separate user. 2015-04-27 13:32:57 +00:00
espie
0163547b4f reexpand will work much better with the correct anchor 2015-04-27 08:14:20 +00:00
espie
e94eae598f use UserProxy and make_path method. 
Have the default host handle chroot correctly.

(still a bit of fetch to fix)
 2015-04-26 18:00:19 +00:00
espie
107eee3773 Add a way to make_path with uig/gid. 
Add a way to run local jobs as another user.
 2015-04-26 17:36:20 +00:00
espie
6968205389 do minimal discovery before reading config files, then do full discovery. 
allow things like proper chroot in the config file only.
 2015-04-25 14:24:03 +00:00
jasper
d42f9ea727 provide a dummy start_build() for DPB::SubEngine::NoBuild; this prevents a 
crash when removing a lockfile in a fetch-only (nobuild) run.

ok espie@
 2015-04-25 12:43:32 +00:00
espie
e06efd44da make most of the creatio files code able to change identity so that logging, 
locking   don't happen as root.

Still need to do the fetchfiles/package linking part.
 2015-04-25 11:40:58 +00:00
espie
f823ed0adf don't forget to copy all the users for later 2015-04-25 11:40:06 +00:00
espie
1184eb9bae more users, for later 2015-04-25 11:32:42 +00:00
espie
446fd12db9 don't bypass the logger method 2015-04-25 11:25:08 +00:00
espie
3286531cff zap duplicate redirection, was harmless 2015-04-25 11:24:24 +00:00
espie
4444edc95e a bit of cleanup: building the host properties is totally separate from 
building the host (especially ensure overrides/finalize is run together).
Do default handling in a saner way, and use that to get a build_user
(and soon a log_user, a fetch_user).

Add run_as scaffolding to allow logging properly when dpb is run as root
(to be used soon)
 2015-04-25 11:23:20 +00:00
espie
d14016b0b5 figure out whether we're running as root on localhost, and then do most 
things internally (chroot, setuid, etc)
 2015-04-25 10:07:19 +00:00
nicm
81baec9d23 Use objdump(1) rather than file(1) to detect binary and shared object 
files.

ok espie
 2015-04-22 14:03:43 +00:00
espie
8c4240d8d0 make things cleaner and less magical 2015-04-21 13:00:00 +00:00
espie
a91c124bcc setting the stage for further changes 2015-04-21 09:53:13 +00:00
espie
6444679ec9 start reorg the way we handle users. 2015-04-21 09:23:57 +00:00
espie
954aba1596 commonalize check for option values... 2015-04-21 08:19:52 +00:00
espie
ffd33fb22c typo, make sure full diagnostic message goes out 2015-04-16 15:49:29 +00:00
espie
8b55e21202 use make -C 2015-03-14 20:45:11 +00:00
espie
0c44a25b22 regorg code, no actual change 2014-12-25 15:14:14 +00:00
espie
480dc42a5d wrap die so that it drops into the debugger if we run under perl -d. 
change the way we do rescans again, to avoid race conditions:
just store away *all* pkgpaths related to the rescan (so that multi-packages
are covered) and then do the rescan. Hold those like we do for normal locks
but in a different list, and release them when the scan is over.

remove the atrocious shennanigans wrt fullpkgnames, it shouldn't be needed
again, hopefully.
 2014-12-07 15:18:50 +00:00
espie
95850b1a81 finally tracked down a long-standing bug (hard to reproduce): when 
building a port with EXTRA depends (meaning it depends on the dependency
being extractible), the normal rules don't really apply, it's not enough
for the package to exist, the listing must have solved its depends names.
 2014-12-06 18:34:05 +00:00
espie
e5f2d3ddbf before dying, set DB::single... if we're running under the debugger, we get 
a chance to catch up.
 2014-12-06 17:39:21 +00:00
espie
6997eedf84 post fake-as-root, we don't need to clean twice. If it fails the first time, 
there *is* a problem.

with new show-size, we don't need root there either.
 2014-11-22 12:44:37 +00:00
espie
fad523ceb8 I had a few "doh" moments where I pointed dpb at an empty config file, 
and waited for it NOT to start up. So prevent that from happening, which
is easy enough.
 2014-10-11 09:03:18 +00:00
espie
8b3963e860 refactor the engine code, so that new_path can put some stuff directly on 
the queue. This shaves significant cpu during initial scanning...
 2014-09-23 18:19:21 +00:00
espie
e1f5ffa474 fix .a printed names and such. 2014-08-09 21:34:12 +00:00
ajacoutot
b09337f520 No more BSD.local.dist. 
reported by richard at aaazen dot com
 2014-07-19 07:04:42 +00:00
espie
cfb049f98e store host info in individual logs 2014-07-18 07:02:20 +00:00
espie
3bea39f566 prevent the local core from running interactive jobs too. 
tricky: have to test for a terminal, as sometimes, we pipe stuff to them
 2014-07-14 17:54:11 +00:00
espie
5ce51d54dd add glue necessary for check-lib-depends to scan out-of-order packages 
problem noticed by Giovanni Bechis
 2014-07-09 11:26:11 +00:00
afresh1
edaddea4f0 Work around perlbug 77706, only hit infrequently on vax. 
ok espie@
 2014-06-14 01:23:48 +00:00
espie
2cf36ad340 make HISTORY_ONLY work again... I managed to break it twice :) 2014-04-28 12:51:41 +00:00
afresh1
03fb3296fa Lexical $_ fixes in package tools 
okay espie@
 2014-03-24 15:18:17 +00:00
espie
f82962b283 prepare for perl 5.18 
- don't bother declaring $_ for while(<>) loops. assume current default
will still work.
- rename variable for readability elsewhere.
 2014-03-17 10:48:40 +00:00
espie
1af9cd4e04 don't bother re-adding already shunted depends. 2014-03-17 10:47:45 +00:00
espie
cfb2f4c5fd log when we actually run engine, since this is hard to match with perf log. 2014-03-17 10:47:12 +00:00
espie
87ed35ed60 obvious typo. spotted by jasper@ 2014-03-16 17:22:05 +00:00
espie
9c85c8f67b do the depends computation before the fork and after the lock, so we're 
aware of everything going on on the same host. In particular, group together
other ports waiting for lock on the same host, and run one single pkg_add.
since they are now in order, this helps quite a bit.
 2014-03-16 08:52:53 +00:00