Crossfire Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Whats the plan? (Was: Re: CF: Suggestions and bugs)

On Apr 28, 12:25pm, rhm-crossfire@math.utk.edu wrote:
> Subject: Re: Whats the plan? (Was: Re: CF: Suggestions and bugs)


> 0.94.x:  Stable distribution.
> 0.95.x:  Development distribution.
> 0.96.x:  Stable distribution with C/S play only.
> 0.97.x:  Developement tree for C/S only play.
> 0.98.x:  Stable C/S semi-balanced distribution.
> 0.99.x:  Developement tree adding last features/fixing bugs for 1.0 release.
> 1.0.x:  Stable public release.
> 1.1.x:  Development tree for new features for 1.2.x
>
> There is no crime in using multiple patch levels.  The currect scheme
> does not allow for clear distinction of what is considered stable and what
> is stable.  Features are only really added in the developement trees, and
> the stable distributions only really get bug fixes, and the occasional
> stable feature added.  Keep in mind that hopefully the development cycle
> on crossfire should be significantly shorter than the linux kernel :).

 I think you may be overestimating the amount of people actually contributing
stuff to crossfire.  Not to take away from those that do, but at best, there
are probably around 6 that actually contribute major new code segments, and not
just bug features.  And even those are pretty irregular.

 I also am not sure of the number of players (server admins probably more
appropriate here), but if there are not enough that the different versions get
much use, it may not be worth the effort.

 Ie, if there are 50 admins out there, and only 2 really want a stable version,
is it worth the effort?  Likewise, if there are the same number of admins out
there and the developement version is known to be very expiremental, are there
then enough that will actually use that and get it refined enough?

 On the other hand, right now, I try and make each release stable (try to only
incorporate stuff that shouldn't make things less stabled or play balanced.)
 Going to a definate developers/buggy version would allow much faster
incorporation of new features, which could be a good thing.


> In the 0.95.x series people could start ripping out X11 code right now
without
> fear of introducing problems into the 0.94.x series.  Having more than 1
> development tree can be a headache at first but I think is certainly worth
> the effort.

 Arguably, in the developers version, most anything could be changed, with the
implied assumption of anyone using it that it might very well be buggy or
nonplayable.  But the question does remain - how many people would actually use
a cutting edge version - if not enough, then it might never get stable.

 The 'question' is how does the stable release catch up to the developers
release.

 In a more real life example, lets go for the versioning as you desribed above
more or less.
 I bump up my versioning to 0.95.0, and start doing some major hacks.
 You get the 0.94.1 release, and start bug fixing.

 At what point does the 0.96 come out?  Do you start looking, and say '0.95.8
looks pretty stable - I am going to take and make it 0.96.0, and you should
move up your stuff to 0.97.0', or does some more refined method go on (ie, at
0.95.8, I stop taking new features, and work on stabilizing it?)  The point
does still remain - are there enough people are there that would download a
pretty suspect 0.95.8 release and thus help make it more stable?


>
> Does this sound like a good idea?

 Depends on how many people will play with the different versions.  It would
seem to me, that the person doing the stable releases can pretty much choose
whatever versioning he desires.

 I would really like to here input of the server admins however.  It seems a
lot want to download the latest version because they do want the new features -
if there was a 'stable' version with only bugfixes, how many would actually use
it?



> I would also like to formally ask permission to mirror the crossfire
> distribution on ogre.math.utk.edu.

 Permission granted. But under the gnu copyleft, which crossfire operates, you
could pretty much set up a mirror without permission, as the program is freely
available.


-- 

-- Mark Wedel
mark@pyramid.com
[to unsubscribe etc., send mail to crossfire-request@ifi.uio.no]
References: