Re: Release early, release often

Andreas (andreas.rick@free.fr)
Sun, 18 Jul 1999 19:42:12 +0200

Petter Reinholdtsen wrote:
>
> I'm one of the strange programmers who beleave in the slogan "Release
> early, release often". I find the release rate of SANE too late and
> too slow.

I agree to that.

> To get the "current" version of SANE, I have to fetch the latest
> release and patches from all over the world to get the latest backends
> and frontends. I beleave this is a problem easily solved using
> version control software and by giving more then one person write
> access to the master source. With a public read-only source tree, us
> wanting to do development on SANE don't have to wait 3-6 months to see
> if our patches made into the distribution.
>
> To solve this problem I sat up a CVS tree at cvs.hungry.com with
> public read-only access from anoncvs.hungry.com, and gave the current
> maintainer David write access to the server. Before the backend
> maintainers get write access, David have to say OK.

> The CVS tree is hardly used at all. I beleave the reason for this is
> that David is unfamiliar with CVS, and also want to keep the master
> source under his control, and not on some random source code
> repository set up by a complete stranger. I hope I'm mistaken.

I'm not too familiar with CVS myself, but I'm willing to learn.

> If this situation don't change, I've been thinking about using the CVS
> server for SANE development myself, collecting all reasonable patches
> and commiting them into my own branch, releasing snapshots every week
> (or more often), like Alan Cox is releasing his experimental version
> of Linux. I do not want to do this, as it will generate a code fork,
> and unless most or all SANE developers colaborates, the CVS version
> would be only for my benifit.
>
> Am I wasting my time?

No! I think you may save some by this initiative...

>
> Among the things I would like to fix with SANE are Win32 support,
> better network support (access control and encrypted communication)
> and improving SnapScan and Coolscan backends. Until the development
> model changes, it seems a waste to start working on these.

I wouldn't mind injecting my work on the Coolscan Backend into CVS.
It needs some changes to sane.h and I was wondering what I have
to do to get them accepted and included.

Andreas

--
Source code, list archive, and docs: http://www.mostang.com/sane/
To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com