[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Criteria for experimental uploads

On 21 Apr 1997, Guy Maor wrote:

> There is way too much stuff in experimental, and no reason for it.
> Some people erroneously believe that a new package must first go into
> experimental and then be moved into unstable.
> Starting immediately, an upload to experimental must meet the
> following criteria:
> 1. The package already exists in unstable.
> 2. The package is such a major change from previous versions that it
> will break the setups of all installed users.  Hardly anybody will
> want to upgrade without investing an enormous amount of time.

The reasoning behind dpkg-mountable going into experimental (I was
planning to move it out in the next version anyway) was that, if it went
wrong, it could do some very obscure things to peoples' systems
(although, from later experimentation with dpkg, it seems that most of
my fears were unfounded), and I didn't want to release what was basically
very-alpha software on the unsuspecting public. Had I been packaging
software which was already written, and which had been tested by the
community at large, it would have gone straight into unstable.

Perhaps this was wrong, and I certainly haven't seen these criteria
before. However, what are people to do it they wish to upload software
which has a good chance of breaking peoples' systems (if linuxconf ever
comes into the distribution with all its daemon functions, this is a
prime example) but which is not yet in the main distribution, presumably
for this very reason?

I see two alternatives. Either create an `alpha' distribution, or
similar, containing things which people want to have tested before they
`go public', or have people set them up for private upload somewhere else
and post about it on the devel list. Neither seem to me to have many
advantages. The first is as near as makes no difference the current
situation, except with `experimental' changed to `alpha,' and the second
in my experience makes it highly unlikely that more than three people
will test it, as well as being inconvenient for people who don't run
their own FTP/HTTP servers.

Feel free to tell me I'm wrong, but this is how it seems to me.

&E, feeling vocal today

Andy Mortimer, andy.mortimer@poboxes.com
Finger asm21@asm21.emma.cam.ac.uk for PGP public key
Everything will be alright, everything will turn out fine.
Some nights I still can sleep, and the voices pass with time.

TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org . Trouble? 
e-mail to templin@bucknell.edu .

Reply to: