Re: howdee + 3.7 pkg testing

["harold felton" <hfelton_lists_(_at_)_hotmail_(_dot_)_com>]

> From: Theo de Raadt <deraadt_(_at_)_cvs_(_dot_)_openbsd_(_dot_)_org>
> To: Stuart Henderson <stu_(_at_)_spacehopper_(_dot_)_org>
> CC: harold felton <hfelton_lists_(_at_)_hotmail_(_dot_)_com>, misc_(_at_)_openbsd_(_dot_)_org
> Subject: Re: howdee + 3.7 pkg testing Date: Thu, 21 Apr 2005 09:59:25 -0600
>
> > Mirroring the packages (approaching 3000 ports, for multiple
> > architectures) must use a fair chunk of bandwidth, so it's no surprise
> > if it's not done for every single snapshot...
>
> It is roughly 15GB of data, for all the architectures.  It also takes
> a lot of time to build on some architectures.
>
> There is some attempt to syncronize base snapshots, X11 snapshots, and
> package snapshots.  But it is not foolproof.  Creating greater
> syncronization would slow the development and testing pace.  No
> changes are planned in the adhoc way we do it now.

thanx - i appreciate all the input.  i hadnt-yet gotten to doing a
-current source-build, so i think ill stick with snapshots for awhile.
they are easier to deal with as a _user_, while im still learning...

im sorry to bother-yall with this, but im still confused about the
pkg-vs-ports situation.  im reading the man-pages online [and now
noticed that partial-[PKGNM] was already documented] and i will
continue to do more research on my own...

my understanding:
1 - pkg_add $FTP_PKGLOC/PKG_NM will do the following:
 a - handle everything in its own staging-area [/var/tmp...], not
   in the ports-area (like /usr/ports would do i think).
 b - do the full-install of the pkg (download, unpack, yadda-yadda...)
 c - set up the un-install info in /var/db/pkg/PKG_NM
2 - ports do similar things, but are based on doing-it-by-sources
 and i believe do their staging in /usr/ports.
 a - if im reading things correctly, then an install will first 
create-its-pkg
   and then basically do everything that the pkg_add would do?
 b - seem to handle categories-of-ports better than pkgs directly.

questions:
1 - should i be doing individual pkg_adds like i mentioned originally?
 or should i be running my "tests" in the ports-area?
2 - i guess my original-issues were arrived-at in a semi-valid-way then?

> > issues: 1 - pkg_add samba-3.0.13 fails because it needs /etc/samba-dir to 
> > pre-exist.  i created it, then pkg installs ok. 2 - pkg_add 
> > postgresql-server-7.4.3p2 fails twice - first due to /var/postgresql 
> > missing, second due to /usr/local/bin/postgres failing md5-hash from 
> > -client (dependency) version.  i renamed-it.

my goal is still to help [in my own small way] testing the indiv. pkgs
that go from -current to -release.  for instance, on the 3.6-release-cd,
i noticed that there were three [?] msgs that went to stderr, rather than
stdout when doing the pkg_add *.tgz process.  [i was redirecting stdout,
but forgot to add stderr, when i did this - altho my memory is hazy now]

anyways, thanx a bunch for all the help/comments [and esp. the os-itself]
and ill try posting on ports@ when i get myself up-to-speed correctly.

have fun, h.
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/