Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 3 Sep 1996 09:31:13 +0900 (JST)
From:      Michael Hancock <michaelh@cet.co.jp>
To:        Terry Lambert <terry@lambert.org>
Cc:        Kim Culhan <kimc@w8hd.org>, current@FreeBSD.org
Subject:   Re: Latest Current build failure
Message-ID:  <Pine.SV4.3.93.960903085814.28456B-100000@parkplace.cet.co.jp>
In-Reply-To: <199609022126.OAA03091@phaeton.artisoft.com>

next in thread | previous in thread | raw e-mail | index | archive | help
I haven't seen any real solutions to the buildable-current problem.

Just to make it easy I'll describe 4 scenerios that break current.  Note
the order doesn't necessarily imply frequency of occurences.

1) Programmer commits code that wasn't tested well enough.

2) A change is made to a global resource such as a header file or a
compile flag that breaks other programmer's code.

3) Operator error on the part of the (cv)sup user.

4) Corrupt CVS tree.

REMEDIES

1) Buildable-current meister (Louie, Terry, or whoever volunteers) sends
constructive criticism to programmer to shape up.  Others will probably do
the same, I will.

2) Assign global impact determination and resolution function to
Buildable-current meister.  CVS does not have the functionality to support
this automatically, so additional tools are required.

Perhaps the Buildable-current meister has UnixWare or Solaris and can
mount to a FreeBSD box containing the Buildable sources and run cscope to
determine the impact of the header file change.  A Perl script is another
solution.

The Buildable-current meister has a programmer database and uses a custom
script to automagically send mail to all programmers that the change will
impact using the output of cscope or equivalent tool.

The Buildable-current meister patiently collects all changes and does a
make world and if make world is successful the tree is made available for
public access.

3) Documentation.  freebsd-current@freebsd.org cvs-all@freebsd.org.

4) Someone volunteers to write a utility that checks the integrity of the
CVS repository.

Does anyone else have any better ideas?  I can think of other ways to do
2), but they all involve the assignment of a single Buildable-current
meister.

Regards,


Mike Hancock




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.SV4.3.93.960903085814.28456B-100000>