Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 13 Jun 2013 10:17:37 +0200
From:      Michael Gmelin <freebsd@grem.de>
To:        Chris Rees <utisoft@gmail.com>
Cc:        Eitan Adler <lists@eitanadler.com>, crees@physics.org, freebsd-ports@freebsd.org
Subject:   Re: Are ports supposed to build and run on 10-CURRENT?
Message-ID:  <20130613101737.19f01b10@bsd64.grem.de>
In-Reply-To: <3d2e51e9-3aa2-43ba-91c4-13fbbbe9c94e.maildroid@localhost>
References:  <20130613031535.4087d7f9@bsd64.grem.de> <3d2e51e9-3aa2-43ba-91c4-13fbbbe9c94e.maildroid@localhost>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 13 Jun 2013 07:20:13 +0100
Chris Rees <utisoft@gmail.com> wrote:

> It's worth bearing in mind that head becomes a release every so
> often.  If you don't make sure your ports build there, it makes life
> harder for people running and testing it, and makes for a mad rush
> and scramble to fix broken ports before a major release.

Thanks for stating the obvious, but usually I'm crossing that bridge
once we have a release candidate available. As far as I remember
personally, the FreeBSD project didn't do a release by surprise within
the last 15 years :). Right now there's not even a planned release date
for 10 on the release engineering web site. Not even as TBD.

> 
> Since the port can often be fixed with a USE_GCC=any conditional on
> OSVERSION, I must question how much of a hardship it really is...

This was a general question and the hardship lies in the extra time
spent testing on CURRENT, including dependencies.

The fact that it builds using gcc doesn't mean that it runs correctly
on that version. In this case the unit tests provided upstream test
for many things and are able to find catastrophic failures, but part of
the QA I do as a maintainer is also testing more subtle scenarios
including staging real life software built using the port, which is
extremely time consuming and revealed mission critical problems in the
past. I can't guarantee that for 10, so why pretend it builds ok if it
hasn't been tested thoroughly?

So in this specific case I think going with Eitan's suggestion makes the
most sense:

> .if ${OSVERSION} > ${WHEREEVERITBROKE}
> BROKEN= Unit tests fail
> .endif

Which probably should be (since the unit tests fail for a good reason):

.if ${OSVERSION} >= 1000000)
BROKEN= Not ready for FreeBSD 10
.endif

Should I add a new patch to the PR or is this something the committer
could just put in there by herself?

Two more notes:
- As a user of the ports tree, I would expect committers to QA
  submissions on RELEASE version of the OS as well.
- If committers find a port that works on RELEASE but not on CURRENT,
  I would suggest to simply add a version specific BROKEN message to
  the Makefile by default, inform the submitter and not delay the
  submission any further.

Cheers,
Michael

> 
> Chris
> _______________________________________________
> freebsd-ports@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-ports
> To unsubscribe, send any mail to
> "freebsd-ports-unsubscribe@freebsd.org"



-- 
Michael Gmelin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20130613101737.19f01b10>