Re: commit access policies

[Alfred M. Szmidt]

   If you want to suggest the creation of a specific branch for a
   specific purpose, I'm all for it.

I atleast don't see any need for a specific branch right now, I wanted
ams-branch to have a generic name (devel-branch), but Roland disliked
that.  In either case, I consider it generic.

   >    The advantage of this model is that people who are trusted to
   >    follow the commit policy, who can be trusted not to hose the
   >    CVS server or raise major headaches for maintenance, can be
   >    given commit access in Tier Two.
   >
   > I think this should be asked from anyone with commit access, be
   > it tire three, tire two or tire one.

   You must have misunderstood.  People in Tier Three do *not* need to
   be trusted with these things, because they cannot abuse cvs access,
   not having any.  (By "hose the CVS server" I meant things like
   checking spam in, or checking in ginormous files, not just a run of
   the mill DoS attack.)

I wasn't thinking about Tier Three people, I was thinking about Tier
Two.  It was simply a typo to mention Tier Three in there, since they
can't do anything anyway to the CVS tree.

   >    The result is that people who have not yet earned the
   >    confidence that Tier One implies have to be denied commit
   >    access entirely.
   >
   > You put a to strong weight on what Tier one should imply.  Not
   > hosing the tree, and causing trouble for other is quite good
   > cirteria for anyone who has commit access no matter what Tier
   > they belong to.

   No, if you do not check things in to the main branch, then you
   cannot break the main branch, no matter what disastrous things you
   do elsewhere.

Uhm, but Tier Two people can do exactly that.  They have earned the
confidence not to screw up the main branch (or any other branch),
unless someone approved the change.  They can still screw it up, but a
limb or two might be missing after the act.

   I do not want people getting their "asses chewed".  Gnucash
   development is occasionally hosed because maintainers with commit
   access check things in that break builds for other people

They should have a stable vs development branch.  libc head likes to
break on occasion to and I'm not refering to Hurdy bits there.  I
don't actually mind if the tree doesn't build on ocassion in CVS.  It
is a bit different for us since we don't have a release we can point
people to.

If it wasn't clear, by `ass chewed' I really meant a nice poke that
they should fix whatever they broke.

   > So yeah, I prefer a two tier based system over three tier based
   > systems, since in tier three based system tier one is utterly
   > small, lazy, and unresponsive.

   If you are in Tier One of the gnumach source, would you be small,
   lazy, and unresponsive?  I hope not...

Depends on who you compare to.

   So I'm happy to put you into Tier One, if Roland agrees (I'll poke
   him myself if he doesn't pop his own head up shortly), but a
   condition of that is that you don't start declaring Tier Two empty
   on your own hook.

I'm still not happy about it.  The reason is that I will get _all_ the
burden (simply because I am the most responsive).  I actually trust
people to check that something works before I commit it (I do look
that the overall patch is OK).  And would expect that this would be
done if they commit it themselfs.  The only time I think the
intervention from a tier one is really needed is if when one wishes to
commit something that is big, changes interfaces or something that one
is unsure of.  I'm also not sure what kind of responsibility a tier
one has.

So I'd rather see something like a Tier One and a Half group of people
who can commit stuff, but they should send a message to bug-hurd with
the patch, what it does etc before commiting it (atleast for gnumach,
I don't care much for hurd HEAD since the creation of ams-branch).
And if the fix isn't correct (since a Tier One looked at the patch),
they should fix it ASAP.  So they should be more carefull than a Tier
One, but not require explicity approval as a Tier Two.  So for
example, a Tier Two person who has had commit access for say a year
(and has actually gotten patches in as is by a Tier One on several
occasions) automaticlly get pushed into the Tier One and a Half group.


By the way, could you/Roland give Thomas Schwinge (savannah user name
tschwinge) commit access? I trust him enough not to fuck things up,
and he has done most of the dirty work when it comes to patches and
would be a immense help for me atleast.