cpet/openbsd-ports
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
126,763 Commits 1 Branch 0 Tags
Commit Graph

4686 Commits

Author SHA1 Message Date
espie
719977d8ca just ran into a lock removed while scanning the dir 2018-02-18 16:44:17 +00:00
gonzalo
b52f15231c add suricata's user 2018-02-15 08:33:38 +00:00
sthen
53dfe11879 simplify the GH_* vs MASTER_SITES comment. 2018-02-09 17:08:33 +00:00
espie
9c419212b5 write an actual list of DPB_PROPERTIES, since I didn't know where to put 
lonesome. There's some duplication with the other description, but this also
yields a single reference location for all keywords

don't document memoryhog, I might scrape it entirely
 2018-01-29 18:20:12 +00:00
espie
6f2ae7c1be support for a lonesome property, so that a job can swallow all cores on 
a host. Needed by shit like rust on i386, which is very very close to memory
limits
 2018-01-29 15:45:29 +00:00
naddy
92bb18cbb2 add a hint how @version is derived; discussed with espie@ three months ago 2018-01-26 13:10:08 +00:00
landry
2d23374af5 Remove the update cookies in make clean=package. Fixes typo/thinko in r1.1371 
ok espie@
 2018-01-24 13:42:03 +00:00
phessler
706f4b6bc5 when we make plist, check to see if 'aarch64' is in the filename 
OK jca@ sthen@
 2018-01-19 17:53:00 +00:00
phessler
03aa877297 arm is now using clang as the system compiler, chase base 
OK jca@
 2018-01-19 17:50:53 +00:00
espie
3d979511ef really fix state machine 2018-01-19 11:59:47 +00:00
espie
b0a9b72909 setting PORTS_PRIVSEP=Yes without SUDO=doas makes no sense 2018-01-17 13:33:54 +00:00
espie
958cf17ce3 automaton bug: removing a lock will make the "frozen" go away but when 
it comes back, since we already fixed the "Awaiting lock", override
is still defined, though it shouldn't be.

Didn't notice before because you have to fudge locks on ports that actually
stay frozen to notice, thank you boost
 2018-01-16 21:51:44 +00:00
bentley
3e81b07c71 Replace distcache.freebsd.org/ports-distfiles/ with _MASTER_SITE_FREEBSD. 
Some ports used _MASTER_SITE_FREEBSD already but others hardcoded the URL.

Switch _MASTER_SITE_FREEBSD and MASTER_SITE_FREEBSD_LOCAL to https.
 2018-01-13 03:00:46 +00:00
sthen
322897489f update some apache mirror URLs, drop some dead ones 2018-01-12 22:06:28 +00:00
rpe
34b689b187 Reflect the change of the shebang line from /bin/sh to /bin/ksh for 
rc.d daemon scripts in the template too.
 2018-01-11 19:30:18 +00:00
claudio
e42a21ffd2 take uid/gid 799 for upcoming prometheus port 2018-01-10 16:20:58 +00:00
espie
ab0ab94863 surprisingly enough, split returns a minimal result (e.g., 0 element) 
when confronted with an empty line.
Reorg parsing to handle this properly.
While there, have the parsers display a name/line number for bogus lines
in stat files (most often produced by filesystem burps)
 2018-01-07 10:49:15 +00:00
espie
795b86dbad somehow, distfile with sz but no sha can reach there. 
so protect.

fixes an immediate crash reported by rpe@, even though I don't quite see
how this happens...

ports with moving distfiles are somewhat bad news... I need to rescan
more forcibly, probably.
 2018-01-04 12:11:26 +00:00
espie
a9baff7820 "normally" depend does everything that's needed, but ports with a 
:patch/:configure depend still want the "prepare" stage.
Otherwise, show-prepare-results may get confused and register too much.

So index it on BEXTRA, so that we don't run the extra stage when it's
not needed, but run it when required.

problem noticed by nigel@, thanks for the sleuthing.
 2018-01-03 23:19:51 +00:00
landry
64ff0ec2e3 Reserve uid/gid 797 for _influx, and while here 798 for upcoming 
_grafana port/user.
 2018-01-03 21:08:29 +00:00
espie
f6f27269fc move back "peeking at watch file" in the parent. 
this could be used by other watchers
 2018-01-03 14:23:06 +00:00
espie
8200732ee9 the 'Awaiting lock' check doesn't belong in generic watcher, so inherit from 
it for the special "build port" job, and track the lock just once, on the
first tasks that will run in ports (e.g., show-prepare-results and inbetween)
 2017-12-31 13:03:07 +00:00
espie
77333b9891 set the watcher before starting the build, so that it exists when we want 
to tweak it.
 2017-12-31 13:01:53 +00:00
espie
e7343419e8 Hum, forgot about pkg_add in the depends stage. 
revert temporarily
this check needs to move anyway, this is highly Job/Port specific
 2017-12-30 10:47:23 +00:00
espie
5854dc8693 Only check for Awaiting Lock on the *first* frozen moment. 
There's no need to reset the slate later, as we're already building.
 2017-12-30 10:40:58 +00:00
espie
933c63bfa3 document recent login.conf, pf.conf evolution. 
talk briefly about PORTS_PRIVSEP.
 2017-12-29 17:58:42 +00:00
espie
2fc804fd2b if we don't own the host lock, replace the "waiting for lock" message entirely 
teach minicurses about this, since it's higher priority than blue.
 2017-12-29 17:08:18 +00:00
espie
b8ee79a26c document new message "stuck on..." 2017-12-29 16:57:21 +00:00
espie
01996a616b so, that's a DPB::UserFile, use the correct method. 
tweak the message so that it has a better chance to fit on the file

teach minicurses about it, since that kind of problem will require manual
intervention.
 2017-12-29 16:52:34 +00:00
espie
de7ab17cb0 so when ports get stuck, try to obtain a bit more information as to why 
(awaiting lock from the ports tree, and host locks from other dpbs)
 2017-12-29 15:49:21 +00:00
espie
8a68a83320 fully document lock types 2017-12-29 15:00:34 +00:00
espie
e482c686a6 probably more than you want to know, all those messages are dpb diagnostics 
for gory things, some of which do not always work 100%
 2017-12-29 14:48:43 +00:00
sthen
d9186f795b Add xgettext and gettextize to the list of poisoned executables used when 
gettext-tools is not a listed dependency. This resulted in 2 new failures in
an i386 bulk, both fixed already. (naddy already bumped into and fixed a few
others),
 2017-12-26 23:38:37 +00:00
landry
578f36e973 take uid/gid 796 for upcoming minio server port 2017-12-23 20:23:03 +00:00
feinerer
d91f96af6b Reserve uid/gid 795 for _turnserver user/group 2017-12-23 17:43:21 +00:00
zhuk
c652e6b6f0 Check that rc.d scripts and README files are not forgotten. 
Inspired by bcallah@
 2017-12-22 22:59:53 +00:00
espie
691677b2cf add a DIAGNOSTICS section with a list of relevant messages, to be documented 2017-12-19 16:08:15 +00:00
zhuk
3fc9af2224 Do not warn about missing dependency on x11/dbus,-suid for x11/dbus,-suid. 2017-12-17 14:02:02 +00:00
zhuk
88807ce481 Make sure that neither @rcscript or share/doc/pkg-readmes/* items have 
@owner, @group or @mode set.
 2017-12-16 18:36:54 +00:00
zhuk
a9b1ae45c7 Quiet a few tweaks for portbump. It features: 
* Add second-priority potential places where to add REVISION;
  * Better WANTLIB tweaking: it should now roll and unroll
    ${MODFOO_WANTLIB} items in WANTLIB declarations correctly.

Some of those were rotting in my tree for months. Sorry for laziness. :-\

There is still tests/portbump/t5 failing, I'm working on it. Sloooowly...
 2017-12-16 18:10:49 +00:00
espie
d68fcbffe2 if FAKESUDO can be something, then we need a SETENV to handle environment 
problem noticed by jca@
 2017-12-07 20:14:25 +00:00
espie
7d31773d38 next stage of PORTS_PRIVSEP: run most steps of the build under 
BUILD_USER (aka _pbuild usually)

as far as I know, only _DEPFILE stuff is left out for now.

Tested on roughly 6000 ports without any ill effects

based on an initial remark by pirofti@
 2017-12-05 17:46:43 +00:00
espie
02c3481452 write PORTS_PRIVSEP = Yes by default. 
It's overrideable as an extra mk.conf line anyway, and it makes more
sense in the default case: chroot useable for running dpb as root.
 2017-12-03 14:55:36 +00:00
espie
b17992612c oops 2017-12-03 12:57:35 +00:00
espie
adf97138bc tweak pre-fake and post-install modules, so that they're run in the 
exact same way (e.g., with FAKE_SETUP done and with changed
identity already).

Adjust user modules accordingly.
 2017-12-03 11:10:10 +00:00
espie
20e4b3344d add a "status fullpkgpath" that displays what it currently knows about a given path, to be augmented eventually. 2017-11-30 14:54:00 +00:00
espie
63f09e1325 a bit of glue to prevent adding the same host twice 2017-11-30 11:17:53 +00:00
espie
2b052aa2ad oops, I forgot to give the core back after scanning. 2017-11-29 11:21:49 +00:00
jca
d99f43a337 Remove gettext remnants, no longer relevant 
ok sthen@
 2017-11-28 18:30:16 +00:00
espie
48800a9e33 more love after feedback from danj@ 2017-11-28 15:38:07 +00:00