Date: Sat, 18 May 1996 06:06:02 +1000 (EST) From: Gary Roberts <gdr@ajax.che.curtin.edu.au> To: freebsd-hackers@freebsd.org Subject: Re: Standard Shipping Containers Message-ID: <199605172006.GAA11345@ajax.che.curtin.edu.au>
next in thread | raw e-mail | index | archive | help
>> The great CTM - SUP - CVS and "who's taking cheap shots at who" debate.
You've all read the arguments (if you are interested -- and I certainly am)
so I won't burden you with unnecessary context.
I would regard myself as one of Richard's `Joe Users'. I've followed this
thread and feel I should give feedback from my perspective. Someone said
"Let the users speak" so here goes...
I have a love of unix in general, BSD unix in particular, and I think
FreeBSD is just the greatest thing since sliced bread. I have nothing
but praise and admiration for _all_ the developers and contributors who
have made FreeBSD what it is today. I'd love to have the ability and
time to contribute.
Whilst I have no real ability or need to customise the sources, I have
always installed full sources right back from the 386BSD+patchkit days.
I can read makefiles and do basic sysadmin type tasks and I always
customise my kernel by removing excess baggage. Occasionally I rebuild
things and sometime even port things if it proves to be not too difficult.
FreeBSD is essential to my work these days and I want to have the
benefits of stability and reliability with as much of the new development
as I can possibly get. I have therefore chosen to go the -stable route
for the moment anyway. Over the years I have done a number of full
installs on a number of machines and I have really appreciated all the
work that has been done by Jordan and Co on sysinstall. A year ago I
used to think that installing SNAPS was the easiest way to stay relatvely
up-to-date. I used to do complete installs and then spend quite a lot
of time re-customising everything. Because of the work involved in this
process, I still have two machines running August 95 SNAPS, rather than
something a bit more current.
My personal work machine is a 486 notebook _still_ running 1.1.5.1.
Recently, on a spare identical notebook, I installed 2.1.0R on a 1.28Gig
drive and used ctm to bring it up to -stable. I found (after overcoming
my initial reluctance and inertia to experiment with the unknown) that
this was the simplest and most painless upgrade I have ever done. After
ftping the `biggie' and all the subsequent deltas and letting ctm do its
thing, I did a make world, rebuilt and installed the kernel and rebooted
into -stable with amazing ease. I now have the new deltas flowing in
automatically by e-mail and whenever I feel like it, I can run ctm again
and repeat the process. My thanks go to phk for ctm and also to rkw for
keeping up the supply of deltas (any anyone else involved as well).
Even though I am _normally_ very well net connected, ctm suits me very
well for a number of reasons:-
1. I'm now acquainted with the procedure and it is quite painless :->.
2. After the initial `biggie', the deltas are environmentally friendly
(by conserving a precious resource - net bandwidth).
3. By extending the default 3 day expiry time for sendmail bouncing stuff
on my mail server machine, all the new deltas will still eventually
arrive on my personal machine even if my work takes me `on the road'
for up to a month to places where I don't have internet access.
The same is equally true for mailing list stuff like -hackers :->.
(It's quite spectacular when I reconnect and watch a huge batch of
stuff hit my poor little notebook :->.).
Having said all this and having read the various comments and counter
comments to rkw's proposals, I find myself wanting to support him 150%
if what he is proposing will make his life easier. I believe that there
is a relatively large group of `slightly better than novice' users who
are capable enough to do `make world' and rebuild their kernel and who
would really appreciate being able to painlessly maintain a steady flow
of fixes, enhancements, new functionality or whatever through the ctm
mechanism. At least having discovered it, _I'm_ very appreciative of it.
Since starting this series of comments, my machine has received quite a
flood of further e-mails on this thread, including this last snippet from
Jordan:-
> Or you could also make the point that for getting the *CVS* tree, for
> which read-only access is the norm, sup or CTM are fairly
> interchangeable and it's back down to choosing by required latency
> again. As disk space gets cheaper, I think I'm going to be advocating
> local copies of our CVS repository as the holy grail of src tree
> management. :-)
A number of questions:-
Is the CVS tree available as ctm deltas or is it only -current and -stable?
Am I correct in thinking that with an up-to-date CVS tree (maintained by
sup or ctm) you can create any of the supported source trees in a fully
consistent and functional state at a time of your choosing?
Also would the only real penalty be extra disk space to keep it all?
Finally am I correct in thinking that at the time of a release (either
SNAP or full release) you could just checkout your own release from your
locally maintained CVS tree and build it all yourself in a reasonably
trivial manner even for a moderately unskilled or semi-skilled user?
If `yes' then where was that phone number for my HD supplier :->.
Or if I'm way off the planet please let me down gently :-<.
Cheers,
--
Gary Roberts (gdr@wcs.uq.edu.au) (Ph +617 3844 0400 Fax +617 3844 0444)
4th Floor, South Bank House, 234 Grey St, South Bank QLD 4101 Australia.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199605172006.GAA11345>