[2123] in SIPB_Linux_Development
Re: RedHat-Athena 5.1 packages
daemon@ATHENA.MIT.EDU (Derek Atkins)
Tue Sep 1 14:58:16 1998
To: Salvatore Valente <svalente@MIT.EDU>
Cc: linux-dev@MIT.EDU
From: Derek Atkins <warlord@MIT.EDU>
Date: 01 Sep 1998 14:57:50 -0400
In-Reply-To: Salvatore Valente's message of Tue, 1 Sep 1998 14:16:31 -0400 (EDT)
Salvatore Valente <svalente@MIT.EDU> writes:
> I boot it into Linux mode to work on RedHat-Athena 5.1, and boot it
> back into NetBSD mode so others can use it.
So, how are others supposed to test it? Or don't you think it's
ready for others to test?
> You should have used the Linux-Athena build directory,
> /mit/linux/devel/athena. This way we can save the build tree easily
> to rebuild the sources later.
>
> I don't understand. We don't need a shared build tree to easily
> rebuild the sources. If someone does create a build tree in AFS,
> I have no interest in using it.
And why not? How often do you need to rebuild the whole thing?
Generally, once you build it, you only make small modifications and
just those mods get rebuilt. So, having a common, shared build tree
saves a lot of time in the long run.
Also, if we're not using a common tree, how can we all be sure we're
testing the same code/build? It doesn't help the process if N people
are working from N code-bases.
> Also, the configuration file modification should be made in the source
> tree...
>
> When the configuration file modifications seem to work, either Aaron
> or I will post them to sipb-source-review. If they are accepted, they
> will be committed to the source tree. Sometime after that, we will
> make "official" packages. (The packages I've created so far are for
> more-or-less alpha testing of the configuration files and the rest of
> the source tree for that matter.)
I agree that this is the correct process. It just feels like there
are single-points-of-failure here, or at least a procedural
bottleneck. I mean, I have plenty of time to work on it, but I don't
know where to pick up from where you left off, or how I can speed up
the testing.
> Alternatively, we can have a Linux-Athena source tree checked out of
> AFS.
>
> We already do, in the source-sipb locker. However, that's just for
> reference. If you want to make any changes, check out a copy for
> yourself.
I wasn't asking to make changes to the sources. I wanted to do a
complete build of "stable" sources so I have something to test. But I
was under the impression that the code in /mit/source-sipb wouldn't
build as-is under Linux. Can it?
> > 4. Once every package builds, I will do a clean reinstall of RedHat
> > 5.1 on snork...
> You'll lose the build tree you made, which would be bad. You'll
> also lose the srvd you created, which would also be bad.
>
> There's no problem with losing the initial build tree and the initial
> srvd. Once I've used a system running with just my initial 5.1
> packages for a couple of days and have an idea about what works and
> what doesn't, I'll make whatever changes seem appropriate, make sure
> all my source tree changes are approved and committed, and rebuild the
> srvd and the packages.
Can I recommend that the "final" build-tree/srvd be in AFS? That way
there is a single place that many can go to rebuild packages when
small changes are made.
> Once you do have working packages, I'd like to know...
>
> Thanks. I'll keep you updated.
Thanks. I'd like to help (since I seem to have a lot of time on my
hands during the day). What can I do to help you?
> Have a nice day,
> Sal.
-derek
--
Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
Member, MIT Student Information Processing Board (SIPB)
URL: http://web.mit.edu/warlord/ PP-ASEL-IA N1NWH
warlord@MIT.EDU PGP key available
