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

Re: releases



On Tue, Feb 08, 2000 at 10:17:12PM +0800, Hilaire Fernandes wrote:
> Although we have written a lot of new code since last august - at
> least in the Dr Geo code - we have been quite lazy in term of release
> or any other kind of communication, and we were stick to a 0.5.0
> release since we merge. Going ahead to 0.5.5 instead if 0.5.1 was to
> show that the project was quite active and alive.  I think we should
> careful to release more often as people cas see progress in the project.

I was just a bit worried about increasing version numbers too fast.  We'd run
out of numbers real quick that way :)

>  > 2) Also hilaire could you announce a pending release a couple of days in
>  > advance, in case there is some broken code or something that needs to be
>  > fixed or other issues with a release.
> 
> Yes, your are totally right, let say two days. So anyone leading one of
> the main part of Dr Genius can propose for a release when she/he has
> achieved a notable improvement and if no complain in two days I
> publish the next release.

perfect.

>  > 3) Why are the files always ending with -current instead of the version
>  > number.  I think it confuses people.  make distcheck already gives you a nice
>  > filename why not use that and only make a soft link on the site to the newest
>  > one with current.  Also there should be a way to download older releases.
>  > Sometimes people can find bugs by looking at older releases to see when they
>  > were introduced or people can revert to older versions if new versions are
>  > broken.
> 
> ok, I don't know how far we should go with older release ? one ? Two ? All ?

As long as diskspace is not an issue, why not all.  We can stick old releases
in an "old" subdirectory so it makes looking up the current version easier.
Also leave the version number in the filename, I've had a couple of people
wonder about the tarball on the site being the right version since the
filename doesn't indicate anything.

George