cpet/openbsd-ports
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
132,379 Commits 1 Branch 0 Tags 544 MiB
4109a05f37
Commit Graph

5072 Commits

Author SHA1 Message Date
espie
80ff67ad9a do not pass TRUEPREFIX through, as it's subpackage independent 
adjust post update-plist changes
 2018-10-18 09:38:32 +00:00
espie
2bee44f298 improve FETCH_PACKAGES 
- add _FETCH_RECURSE_HELPER that records whether we're actually building
or trying to grab a dependent package.
- if grabbing dependent packages while FETCH_PACKAGES is active, *do not*
depend on build dependencies, so that we only fetch the minimal set of packages
needed for what we're actually doing.
 2018-10-16 14:16:04 +00:00
espie
eafa5ba175 move dependence on build cookies on a separate line for further work 
no functional change
 2018-10-16 14:10:31 +00:00
espie
ed0ff30246 change the semantics of FETCH_PACKAGES variable. 
Default is "no" : no fetch
Set to options you wish to pass to pkg_add in addition to normal options
(e.g., FETCH_PACKAGES=-Dsnap recently)

To use it without passing options to pkg_add, just 'make FETCH_PACKAGES='
 2018-10-16 13:59:14 +00:00
sthen
be94a60297 just call apxs and apachectl instead of apxs2/apachectl2, there is no more 
conflict with Apache httpd 1.x so the suffix is no longer needed.
 2018-10-16 13:11:30 +00:00
espie
4236b0f8cc make darn sure find does see all it's supposed to 
problem noticed by solene@
 2018-10-09 09:47:28 +00:00
espie
40cd6f544e die is a bit terse. Explain why, helps debugging user issues. 2018-10-08 15:20:28 +00:00
espie
1e6f822edd add a "fix-permissions" utility target for PORTS_PRIVSEP = Yes 
to be refined later, maybe.

okay sthen@
 2018-10-08 13:28:05 +00:00
espie
3df1949da5 say something is MASTER_SITESn is empty 2018-10-08 12:03:16 +00:00
espie
9dcbd84a9a make sure we get a value, as noticed by naddy@ 2018-10-04 09:02:26 +00:00
sthen
d802ba16e5 add alpha to arch-defines.mk, this was missed when alpha support was added. 
from pascal@
 2018-09-26 13:23:53 +00:00
espie
e871125460 simplify setup, both _usermode and _pkgmode are no longer required at all 2018-09-22 07:25:04 +00:00
espie
d9c37d55e3 remove some older doas calls that no longer make sense now that FAKE_AS_ROOT 
is dead
 2018-09-22 07:13:05 +00:00
robert
c6d51f57c7 add the USE_LLD knob which makes it possible for a port to use LLD for 
linking instead of bfd

ok espie@
 2018-09-20 10:27:27 +00:00
espie
447d63002f as noticed by krw@, move the finalization code for jobs/squiggles/parallel 
into the job finalizer, because ncpu does not necessarily get run.
 2018-09-17 13:57:05 +00:00
espie
65e2763682 zap debug message 2018-09-14 11:36:07 +00:00
espie
2c8b572819 scrape stupid debug code 2018-09-13 15:53:19 +00:00
naddy
ef1fe4c5fe lld does not have a default linker emulation. In order to create 
an object file with a .data section from a binary blob (ld -r
-b binary foo foo.blob), the target emulation must be specified
explicitly.  Provide a variable LLD_EMUL that maps the architecture
to the respective emulation name.  We include the -m switch for
direct use in an ld command line.

ok sthen@ espie@
 2018-09-12 19:29:08 +00:00
espie
4fc18975ba for_list integrates the next, so don't do it twice 2018-09-12 15:34:31 +00:00
espie
dc9f6babcf ouch, don't try things twice if it doesn't work the first time. 2018-09-12 13:05:32 +00:00
espie
5630250b91 do a transitive closure, this should fix incomplete licensing issues 2018-09-12 10:48:13 +00:00
espie
416cc26d12 do things slightly more directly, to be able to pass SUBDIRs manually (prepare 
for transitive closure)
 2018-09-10 15:15:55 +00:00
espie
6be83f422b get rid of globals 2018-09-10 15:07:33 +00:00
espie
8e814a5b02 unify the conflicts and dir checking code some more so that I can go further 2018-09-10 14:54:41 +00:00
espie
214ebb33fd don't redo the same package twice 2018-09-10 14:34:17 +00:00
bentley
f756e3c091 Reserve a user for vnu. 2018-09-10 06:25:52 +00:00
espie
498590d80c also check for licences 
the ports tree scanner *will* need transitive closure support, as stuff
like py3-cairo is only reachable thru deps right now!
 2018-09-08 15:33:04 +00:00
espie
43a304357e more 2018-09-08 12:53:37 +00:00
espie
43328d0459 keep going after an error (like databases/ports-readmes) 2018-09-08 12:13:37 +00:00
espie
b48bff8ae5 fix: this requires dependencies actually 
and it's just a part of pkg_check-problems, ever since I unified tools,
so there are no other consumers to check so far.
 2018-09-08 11:15:03 +00:00
espie
e21c8cafe3 also moved to bin 2018-09-07 09:05:56 +00:00
espie
c6192746c3 all that stuff actually moved to bin/ years ago 2018-09-07 09:04:34 +00:00
espie
d7dc3f67ac thinko. fix 2018-09-06 15:01:39 +00:00
espie
0cd41bc4d8 add a "parallel2" property that defaults to the parallel value 
chrome is taking so long to build, I want to experiment with giving it
more cores...
 2018-09-06 11:31:30 +00:00
espie
e9afd6c5c8 remove the extra doas for running _internal-clean 
you shouldn't have to drop from root to _pbuild/_pfetch
 2018-09-05 14:10:15 +00:00
sthen
3eb429508f use hw.ncpuonline rather than hw.ncpu for default #/cores in DPB and the 
number of parallel "conflict finder" runs of pkglocatedb in update-plist.

ok espie@
 2018-09-05 10:31:55 +00:00
espie
ad0ba6e372 prevent infinite recursion 2018-09-05 09:16:19 +00:00
espie
67096a6d1a the ReverseSubst fragment is used to put *back* PKGSTEM where it belongs, 
it still needs to be basically ignored.

noticed by rsadowski@
 2018-09-04 16:33:56 +00:00
ajacoutot
9d0a6f0ab0 FULLPKGNAME -> PKGSTEM 
ok espie@
 2018-09-04 13:40:26 +00:00
espie
ee2d3810b4 introduce PKGSTEM 
intended to fix a problem reported by tb@: pkg-readmes docs used to have
the FULLPKGNAME encoded, which makes for gratuitous differences from version
to version.

- use PKGSTEM inside the README for less variation
- ditch the FULLPKGNAME as the install location, PKGSTEM is enough.

PKGSTEM ?=  is overrideable, because stuff like php branches need to install
several readmes.

Next commit will fix all affected ports: all readmes change location, so lots
of plist tweaks and bumps
 2018-09-04 12:41:51 +00:00
remi
55d0616792 register user for the upcoming port of netshot. 
suggested by danj@
 2018-08-16 21:43:08 +00:00
espie
ba234ba60b vim swapfiles are not a problem 2018-08-10 09:50:54 +00:00
espie
a461d6ed8a give LOCALBASE != /usr/local 
a fighting chance wrt pkgconfig and perl

(this does not affect the default /usr/local case at all)
 2018-08-09 19:53:41 +00:00
espie
7065c9344a tweak imake-cf: use LOCALBASE more consistently (some of these should be 
PREFIX, to look up)

Also USE_GROFF to not require groff for generated rules.

not really used, according to Ingo, but won't hurt...
 2018-08-09 17:43:00 +00:00
espie
c4a1c5c02b pull USE_GROFF up so I can test it in modules 2018-08-09 17:41:43 +00:00
sthen
bebfb52e62 provide and use RUST_ARCHS, from semarie. sounds good to espie 2018-08-08 09:19:40 +00:00
espie
3403aed295 pass _FAKE_TREE_LIST thru to update-plist (tricky glob/quoting) 
this allows removing some special-cases from mtree, as we get the
real directory structure pre-built in fake.
 2018-08-06 09:36:32 +00:00
fcambus
d0673a0a70 Reserve GID for _libpostal. 2018-08-06 09:05:54 +00:00
espie
548326f6c5 zap some dirs we don't want 2018-08-04 19:44:31 +00:00
espie
e7a7af3e4e pass AUTOCONF_VERSION and AUTOMAKE_VERSION unconditionally into MAKE_ENV 
and also define AUTOCONF_ENV always, with PATH set as well

should simplify any autoreconf/bootstrap definition
 2018-08-04 09:01:08 +00:00