GDB's roles

GDB's roles

[Andrew Cagney]

Ignoring the upper levels, I think the people listed in the gdb 
MAINTAINERS file find themselves filling one or more of the following roles:


Elders:

These are the keepers of the `GDB wisdom'.  They are largely uninvolved 
in day to day GDB development - very hands off - very much don't given 
opinions except where to point out problems and flaws.


Adminstrator:

They do the day to day stuff: bug report unreviewed patches, ping 
maintainers that are asleep at the wheel, try to keep the political 
(RMS, et.al.) separate from the technical (this groups focus), web 
pages, ...


Architect:

Ensure that GDB is `heading in the right direction'.  The actual 
direction is determined by the group but this person gets to watch for 
things going off the rails.  Also, very occasionally, make architectural 
judgment calls.

The architect is woried about key interfaces such as the architecture 
and target vectors.


Core Maintainers:

These are the people on which everyone else depends.  They put 
themselves to the grindstone ensuring that the GDB wheel keeps turning. 
  These are the people that do the hard work of reviewing / approving 
patches.  These people need to be relatively reliable.  These people 
need to be willing to do the dirty work (such as restructuring) that 
can't reasonably be expected of a contributor.

While a core maintainer might also be responsible for certain specific 
areas (symtab, threads, remote), they won't cherry pick the patch list 
and definitly won't fall asleep at the wheel.


Specific Maintainers:

These are responsible for specific areas.  Native, target, host and 
language maintainers come to mind.  Their responsabilities are pretty 
clear, while needing to be responsive, they are not on the critical path 
like core maintainers.  The thing I really like about the target 
maintainers is how they, every so often, pop up to do some maintenance 
(eliminate something deprecated), and then pop back down again.


Developers/Contributors:

These are the people that have `fun'.  They have the luxury of poping 
up, contributing a new feature, and then simply disappearing again 
(typically though, a contributor ends up being the maintainer).  New 
development, of course, needs peer review (by a maintainer) as, in the 
long run, it will be the [core] maintainers that have to support that code.


Using the above as a reference (the last thing this list needs is a long 
irrelevant discussion about the semantics of each of the above :-)) it 
might be helpful for global maintainers to look at this so that they can 
see how their activities contribute to GDB.


Me?  Architect/Adminstrator + core.  While I'm a maintainer for a few 
specific area's (remote and mips) that is very very low down, same for 
development.

Andrew

Re: GDB's roles

[Daniel Jacobowitz]

On Fri, Feb 21, 2003 at 01:48:06PM -0500, Andrew Cagney wrote:
> Core Maintainers:
> 
> These are the people on which everyone else depends.  They put 
> themselves to the grindstone ensuring that the GDB wheel keeps turning. 
>  These are the people that do the hard work of reviewing / approving 
> patches.  These people need to be relatively reliable.  These people 
> need to be willing to do the dirty work (such as restructuring) that 
> can't reasonably be expected of a contributor.
> 
> While a core maintainer might also be responsible for certain specific 
> areas (symtab, threads, remote), they won't cherry pick the patch list 
> and definitly won't fall asleep at the wheel.
> 
> 
> Specific Maintainers:
> 
> These are responsible for specific areas.  Native, target, host and 
> language maintainers come to mind.  Their responsabilities are pretty 
> clear, while needing to be responsive, they are not on the critical path 
> like core maintainers.  The thing I really like about the target 
> maintainers is how they, every so often, pop up to do some maintenance 
> (eliminate something deprecated), and then pop back down again.

One of my concerns from my previous thread about this, which seems to
have died, is that the above doesn't completely describe current
structure.  The set of core maintainers breaks down into specific
maintainers for core components.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: GDB's roles

[Andrew Cagney]

> On Fri, Feb 21, 2003 at 01:48:06PM -0500, Andrew Cagney wrote:
> 
>> Core Maintainers:
>> 
>> These are the people on which everyone else depends.  They put 
>> themselves to the grindstone ensuring that the GDB wheel keeps turning. 
>>  These are the people that do the hard work of reviewing / approving 
>> patches.  These people need to be relatively reliable.  These people 
>> need to be willing to do the dirty work (such as restructuring) that 
>> can't reasonably be expected of a contributor.
>> 
>> While a core maintainer might also be responsible for certain specific 
>> areas (symtab, threads, remote), they won't cherry pick the patch list 
>> and definitly won't fall asleep at the wheel.
>> 
>> 
>> Specific Maintainers:
>> 
>> These are responsible for specific areas.  Native, target, host and 
>> language maintainers come to mind.  Their responsabilities are pretty 
>> clear, while needing to be responsive, they are not on the critical path 
>> like core maintainers.  The thing I really like about the target 
>> maintainers is how they, every so often, pop up to do some maintenance 
>> (eliminate something deprecated), and then pop back down again.


> [...] The set of core maintainers breaks down into specific
> maintainers for core components.

The paragraph:

 >> While a core maintainer might also be responsible for certain specific
 >> areas (symtab, threads, remote), they won't cherry pick the patch list
 >> and definitly won't fall asleep at the wheel.

is trying to express this.

core different to maintainers of sub-systems such as the target, native, 
language or host.  The critical nature of a core maintainer isn't there 
though.  Typically maintainers of nat, tdep, et.al. code are in a `if it 
ain't broke / deprecated, don't fix it' situtation.  While fixes/patches 
should make it into the next release, they are not so much on the 
critical path.

Contrast that to my current core activity of rewriting frames.  I'm on 
the hook for everything.

Anyway, the main thing is that my post is a `white paper', a discussion 
`puff piece'.

Andrew

Re: GDB's roles

[Andrew Cagney]

> 
> One of my concerns from my previous thread about this, which seems to
> have died, ...

`shit happens'.

If you check your mailbox you'll notice an e-mail I sent in november to 
the global maintainers that identified a number of problem areas (from 
memory, I delayed sending it until you'd been made a global maintainer). 
  You'll note that Java was top of that list and I've since managed to 
resolve it -  Java patches have just started to flow again.  Your 
discussion has helped clarify a few things and re-prioritize accordingly.

As the `adminstrator', I get to follow through all of this.

Andrew