Re: [Gdbheads] Let's resolve this quickly

[Andrew Cagney]

> I will just add that I am disappointed that the suggestion made by
> several people, that is to change Global Maintainers back into Blanket
> Write Maintainers, have simply been ignored by those who oppose the
> idea.

I think it was pointed out that the suggestion was not new, and people 
were refered to old discussion threads.  The other problem is that the 
originators of the proposal have been somewhat circumspect[sp?] when it 
comes to explaing their motivations.

"blanket write" isn't about improving the patch review process (although 
some are selling it as that), it is about speeding up the process where 
by ones personal patches get committed.

Anyway, lets for a moment assume that patch review time is the problem. 
 The originators of all this were JimB, KevinB, CorrinaV, MichaelS and 
EricB.  Three are _global_ _developers_ so they already have a great 
deal of freedom.  If Corrina needs something reviewed then one of her 
peers can certainly help.  Now while I guess this group can claim that 
they are thinking of "the greater good", I'm highly skeptical, remember 
as I've also shown, these are some of the least active developers in 
GDB.   Anyway, if you do the numbers, you'll find that the problem isn't 
so much patches not being reviewed, or their review is too slow(1), but 
rather that patches are being rejected or being asked to be rewritten 
(and note how I and Elena are frequent in that rejection rate).

Re-instate blanket write and these people have the perfect mechanism for 
forcing their code through.

Yes, we need to find better ways of distributing the patch review load, 
but lets not be fooled into adopting the wrong mechanism.

--

In a thread with Ian we talk about "power and glory".  If we adopt a 
mechanism that allows global developers to quickly and instantly approve 
any patch, we've also create a great demotivator.

How would you feel as an area maintainer if all the patches for your 
area were being approved by GDB most active global maintainer?  I mean 
what would be your point?  Looks good on a resume but other than giving 
you the ability to commit your own changes it would serve no purpose. 
When it comes to patches you dissagree with, instead of having the upper 
hand, you find that you're arguing over an already committed change. 
Over in GCC where there's this convolted patch reversion process that 
ends at their steering committee.  At least at present, the buck stops 
at the person that should be making the decision.

Look for instance at the current exchanges between my self and CGD.  As 
he observed, I dropped MIPS.  I did that knowing that we had technical 
disagreements, but I also did that knowing that even if we did disagree 
I could trust his final judgment.

Andrew

(1) The exception i can think of is File-I/O.  There I had a conflict -- 
I was heavily involved in the protocol design so I had a clear conflict. 
 There was no way I was going to move quickly on that one.