On 06/15/2015 06:31 PM, Jeremy White wrote:
<snip>
With that said, I think this is a hard problem; I think it boils down to
a request for Alexandre to be 'nicer', with varying shades of just what
that means, and it's been discussed enough through the years that it's
hard to discuss it constructively.
<snip>
Everyone seems to forget that Alexandre is doing *a lot*.
The simple patch queue management is not an easy task.
I guess he classifies patches, reads them, looks at bugzilla entries, installs demo programs, tests patches, verifies them, discusses with other developpers, waits for author feedback...
Understanding the *real* issue is also complex. He needs to know every windows component, how they are related..

People just want their patch committed because it-is-the-best-solution-ever but AJ has to keep the project moving the right way, avoiding regressions...
Let's imagine you're at the other side with a continuous flow of patches...

Yes, he has to make choices (rejecting patches...) and sometimes, we don't get any input on what's wrong.
But when he says 'doesn't compile here', it shows people don't even test their stuff.
Clearly, Alexandre is doing a lot and .. for years so be nice with him ^^

What are the issues here ?
It seems people want feedback.
What's wrong with my patch ? Why is it pending ? Is my patch reviewed at all ? How to improve my patch, I'm a newbie...

And why ?
I guess AJ does too many things, lacks time and manpower.

We can automate things:
- compilation (done with testbot but not perfect yet)
- formatting checks (spaces, if block on several lines or not...)
- common issues (LPcrap...)

I don't know how statuses are managed but maybe a tool can be created or improved ?
We can provide a mentor when a newbie posts a patch for the first time so that AJ can do other things in the meantime. (or delegate to wine-staging or a kind of sandbox but AJ doesn't like the idea)
We could also use the 'sign-off' feature to delegate make a patch verified by a trusted dev (= having a list with domain <-> trusted dev).
We can write standard commit rules and patch lifecycle rules so that everyone know where their patch is far from commit.

Any other thoughts ?