Re: GOP-PROP 4: lessons from 2.14

[Colin Campbell]

On 11-06-30 10:53 AM, Graham Percival wrote:
> By "our limitation", I'm referring to the lifetime of a patch.
> - patch gets written
> - patch is put on Rietveld.
> - ** James does the regtest comparison with his shiny new 4-core
>    8gb-ram machine
> - if no regtest errors, patch is tagged with patch-review, waits
>    2-6 days (estimated) to appear on patch countdown
> - 48-hour patch countdown
> - if no (serious) error, patch is pushed.
>
> I don't think that James' regtest comparison is a big problem.
> Granted, we'd like developers to run regtest comparisons before
> even uploading something to Rietveld, but I still think that the
> amount of time this takes (running in the background, right?) is
> not very significant when compared to the time it takes to write
> the patch in the first place.
>
> Cheers,
> - Graham

I'll jump in with a well-chewed bone:  life would be *much* simpler if 
we, at least for the time being, are disciplined about linking *every* 
patch to an issue on the tracker.  I'm still working through Reitveld, 
to identify those patches which are in active development, and I'm 
trying to get them tied to issues, with the intent that the bug/issue 
tracker be used as the way to record what we're working on.  Ideally, it 
would also tend to keep the higher-priority issues higher in the minds 
of the community.  If the developers would care to comb their own 
Reitveld open issues and let me know what connects to which, I'll be 
glad to handle the housekeeping of raising tracker issues as needed.  
Issues don't necessarily have patches, but all patches should have 
tracker issues.

Ultimately, of course, we need to look at alternate development 
environments, better suited to issue and patch management.

Cheers,
Colin

--
The human race has one really effective weapon, and that is laughter.
-- Mark Twain