How to do source management?

[Jan-Benedict Glaw]

On Wed, 2004-07-28 22:32:52 +0100, Kenn Humborg <kenn at linux.ie>
wrote in message <20040728213252.GA24556 at excalibur.research.wombat.ie>:
> On Wed, Jul 28, 2004 at 09:56:02PM +0200, Jan-Benedict Glaw wrote:
> > X-VAX-is-cool: Yes!
> > Hi!
> > 
> > As you all already know from my previous email to you, I'm about to
> > import toolchain's sources. Before that, I'd like to discuss the "how".
> > Please choose your weapons:
> > 	[ ] CVS
> > 	[ ] Arch
> > 	[ ] Subversion
> > 	[ ] Other: _____________
> 
> I've only used CVS.  But I'm willing to learn.

I've used CVS and Arch; Arch is alot "cooler" in some circumstances, but
quite new software and not many people actually use Arch these days
(mostly because CVS works "well enough" in most circumstances:)

So I think CVS is the way to go.

> > distance:
> > 
> > 	[ ] Just patch sources with whatever is mentioned in upstream
> > 	    CVS status emails and check it in onto HEAD.
> > 	[ ] Import patches into a separate upstream branch.
> > 	[ ] Import snapshots into a separate upstream branch
> 
> Tough call.  I'm familiar with the 3rd option from importing Linus kernels
> into the Linux/VAX CVS at SourceForge.  When it goes smoothly, it's great.
> However, sometimes issues with SF's CVS server make it tricky for such
> large source trees.

I think I'll push in patches directly on HEAD done semi-automatically
(ie. by script, but called by hand at least for the first few weeks to
get a feeling about how often I have to expect conflicts eg. in
./configure files... )

> > and ammunition:
> > 
> > 	[ ] Import binutils, gcc and glibc into three modules reflecting
> > 	    the three upstream modules. Con: IMHO hard to build, at
> > 	    least I need to figure out how to do that at all:)
> > 	[ ] Import binutils+gcc as a combined tree; most probably that
> > 	    would make some sense if we used upstream snapshots then.
> > 	    Pro: easy to build; Con: glibc is "extra" then.
> > 	[ ] Other: _________
> 
> I'd suggest creating a local CVS/SVN/whatever repository and play around
> with the various options yourself using a selection of recent FSF 
> tarballs.

I think I'll try the combined-tree approach. Though, it possibly makes
submitting changes a bit harder (since our tree doesn't look like those
FSF trees), but it'll give advantages during build time.

Does anybody really *know* which parts I actually need from which tree?
Eg. I think I won't really need tcl, libreadline, ...

> BTW, anyone have any objection to making the Reply-To: point to the list?

Should be done now, plus a hopefully correct link to an email archive:)
Still need to work on lug-owl.de's search engine...

MfG, JBG

-- 
Jan-Benedict Glaw       jbglaw at lug-owl.de    . +49-172-7608481             _ O _
"Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur | Gegen Krieg  _ _ O
 fuer einen Freien Staat voll Freier Bürger" | im Internet! |   im Irak!   O O O
ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lug-owl.de/pipermail/vax-toolchain/attachments/20040802/7578f85e/attachment.pgp