Date: 17 May 1996 09:03:42 -0500 From: "Richard Wackerbarth" <rkw@dataplex.net> To: "Jordan K. Hubbard" <jkh@time.cdrom.com> Cc: "FreeBSD Hackers" <hackers@FreeBSD.ORG> Subject: Re(2): Standard Shipping Containers Message-ID: <n1379798653.81207@Richard Wackerbarth>
next in thread | raw e-mail | index | archive | help
> > [ Various arguments about CTM, CVS and so on elided ]
>
> Here's my perspective on all of this...
>
> There are 3 ways of getting the sources:
>
> 1. Extract the srcdist tarball.
> 2. Sup:
> 2a. Configure sup file.
> 2b. Run supfile first time.
> 3c. Add optional sup entry to crontab.
> 3. CTM
> 3a. Unpack initial deltas
> 3b. Subscribe to appropriate mailing list.
> 3c. Add optional alias to aliases.
Don't forget the "live file system"
> All three of these techniques constitute perfectly reasonable ways of
getting the sources, depending on your needs.
I agree.
What I am looking to do now is to combine (1) (2b) and (3a) in such a manner
that the user can mix and match.
>I'd really need someone else to do the coding.
> Considering the relative paucity of
> questions to ask ("What directory do you want to keep your sources
> in?"), it should be a fairly easy task.
I'm afraid it is a little more complicated. But not an impossible task.
Where do you want the updates collected?
Do you want the updates applied automatically?
Do you want to save the updates or delete them after they are applied?
In the case of CTM, we also need to run an ftp client daemon to get the
initial update and any that have gone out before the mailing list takes over.
If we are going to do this (and I think it is a good idea) it places an even
greater value on being able to do two things.
(1) The ability to run this setup program after a system is set up: Just like
the guy who wants to add a HD after the fact, a user is likely to first get
the system up either without the sources or with the sources from the tarball
or CD and later decide that he wants to add the ctm feature.
(2) The unified source product:
There is no need to download 28M again if only we can recognize what the user
already has. By making the source tree carry a tree isomorphic to a ctm
update, including the appropriate tag, we can automate the process of updating
from that point. This applies to both ctm and sup.
If you are willing to accept my idea (or a reasonable variation thereof) for
the unified source product, I'll write the config.
It will get a little "messy" in that I will have to add a user, add the alias,
send mail, etc.
How can we best handle the fact that these cannot be done in real time. They
will have to wait for the system to be up and running with everything
installed, etc. Do we have a mechanism to install an execute once script that
will be activated later in the sequence?
Also, what is your opinion on the "proper" (default) placement of the source
tree. I feel that there needs to be a standard location for each release
version. .../FreeBSD-2.1/src .../FreeBSD-2.2/src
.../FreeBSD-current/src .../FreeBSD-CVS/src so that they can co-exist.
/usr/src is then built by either (at the users option) copying or (preferably)
linking to the appropriate tree.
Or "mount union"?
I would tend to suggest /usr/FreeBSD/FreeBSD-2.1.1/src.
The sophisticated user can move this into /pub/FreeBSD, etc. with a link from
/usr/FreeBSD.
Your thoughts...
--
...computers in the future may have only 1,000 vacuum tubes and weigh
only 1/2 tons. -- Popular Mechanics, March 1949
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?n1379798653.81207>