[CF-Devel] About CVS server stability.

[yann.chachkoff at softhome.net]

>
     
      > I repeat again what I've said a lot: if we do not agree on a clear 
     
     >
     
      > submission/test/update protocol, the problems will get worse and will *not* 
     
     >
     
      > solve by themselves.
     
     >
     
     
     >
     
      IMHO, alot of this is semi-rediculous.
     
     >
     
     
     >
     
      1) Branching.  I wholeheartedly agree with Mark.  If we are going to have
     
     >
     
      branches, they need to be stable branches.
     
     <snip>
Did I ever said the contrary ?

>
     
      2) It's CVS.  It's a live development copy.  I don't understand why people seem
     
     >
     
      to have it in thier heads that CVS is allways going to be stable, and happy.
     
     <snip>

Correct me if I'm wrong, but where did I say CVS should be the "stable"
repository ? What I simply say is: we need to provide stable releases on a
regular basis. It can be from the CVS, although I think there are much
better solutions available.
 
>
     
      3) Patches are only useful for shaking out bugs if you have a critical mass of
     
     >
     
      users.  We don't have that many servers out there.  If none of them run the
     
     >
     
      patch, it doesn't get tested.
     
     I agree that if no one uses the patches, they are useless. But what do you
call "the critical mass" ? Do you think we need 20 servers to test a piece
of new code ?

>
     
      Don't get me wrong..  I don't think we should just throw code willy nilly into
     
     >
     
      the server, nor do I think we should be stuffing wholly untested code into it,
     
     >
     
      or nasty hacks.  But lets have some scope here.  This is a game.
     
     I never said you shouldn't have fun with CF coding. Where did I say such a
thing ? And if you don't agree with the idea of testing the code before
making it available for the users, what solution do you suggest ?

Besides this, don't you think there any developer has some kind of
responsability over all those users who just play crossfire and do not know
anything about its internals ? Coding for your pleasure is indeed a good
motivation, but what about all the users behind ? Should we ignore them and
their possible complaints just because 'it is not fun to test code' ?

>
     
     
     >
     
      However.. just about every time anyone has committed anything larger than a
     
     >
     
      bugfix in the past few months, someone turns around and threatens to quit.  I'm
     
     >
     
      truly terrfied that eventually, we will reach this stage where everything has
     
     >
     
      to be proposed in a formal proposal, given 9 months of discussion, and has to
     
     >
     
      be unanimously agreed upon before hitting the tree with a host of crazy
     
     >
     
      regression suites.  Thats NOT FUN.  I want to have fun, and write code.  I
     
     >
     
      don't want to be embroiled in political process for every line of code I write.
     
     
Who spoke about political discussions about a piece of code ? Do you think
documenting code and letting other coders say what they think about it is
'political' ? Who spoke about 'crazy regression suites' ? I just suggested
a test period (maybe one or two weeks) ! Of course, testing and debugging
is *not* the funniest thing to do - but it is a necessary think do do. 

>
     
      Maybe we do need a formal procedure, but I implore whoever writes it, to
     
     >
     
      consider that we aren't writing space shuttle navigation code, it's a video
     
     >
     
      game.
     
     Who spoke about writing a space shuttle navigation code ??? Is it too
difficult to understand that we need to *test* code at least a little
before putting it into the main source of updates ?

To make my suggestion clearer, since it looks like I was misunderstood:
- Make stable releases available much more often than now;
- Don't put untested code in the stable releases;
- Every developer should take some time testing others' stuff.
If you think it is 'no fun' and 'political procedure' then maybe I'm too
concerned about users and what they want to get. Maybe I should think less
about them and more about my own satisfaction.

(And to make things clearer: I do not blame AV's new submission in that
case - there was enough discussions about the new class IMO, and testing it
was indeed not a bad idea)

Y. Chachkoff