[heat] Addressing the large patch backlog.

Zane Bitter zbitter at redhat.com
Tue Dec 3 19:50:55 UTC 2019


On 12/11/19 12:21 pm, David Peacock wrote:
> Hi all,
> 
> For those interested in the workings of the Heat project, I'd like to 
> kick off a call to action.
> 
> At the time of writing there are approximately 200 open patches against 
> the core Heat project repo alone, not counting the other Heat repos.  
> Recently I started going through and triaging the patches I'd consider 
> "historical" with an arbitrary cut off for this definition of August 1st 
> of this year.
> 
> There are 148 patches which meet this definition, dating all the way 
> back to 2015.  I have gone through them all and placed them into a 
> spreadsheet [0] which I'd invite you all to check.  Provided is a link 
> to the patch in question, initial upload date, last meaningful update 
> date, primary author, and a high level summary of the patch.

Thanks for doing this David!

I added a column for "Core Reviewer comments" as well.

> Additionally I've broken the patches down into three recommended states 
> based on a high level first pass.
> 
> *_Abandon_*
> 
> 34 patches are candidates to be abandoned; they usually are of 
> debatable utility, have significant outstanding concerns, or have no 
> followup from the original developer in a very long time.  In many 
> cases, all of these conditions. *Without good reason or explanation from 
> the original developer, these patches may ultimately be cleared out.*

I went through and abandoned 26 of these patches.

There were a handful of patches recommended for abandoning that I think 
still have value to be salvaged from. I added comments in the 
spreadsheet, prefixed with (ZB).

> *_Rebase + Merge_*
> 
> 38 patches are with a high level look in reasonably good shape, perform 
> a stated goal, and may be trivial to core review and ultimately rebase 
> and merge. *If you're the original developer or otherwise interested in 
> these patches and wish to see them through the merge process, please 
> rebase the patch.*

Most of the original developers here are long gone, and those that 
aren't have probably long since given up on these patches.

Would you have time to start going through these and rebase or recheck 
them (as required), then provide a list somewhere of priority reviews 
(i.e. those that rebased easily and passed tests) for cores to go 
through? It should be pretty easy to merge a good number of these quite 
quickly.

(Maybe wait for the gate to be unbroken first though.)

> *_Research_*
> 
> 76 patches are sufficiently complex that they'll need a much closer 
> look.  Some of these patches are in a seemingly "finished" state, some 
> are a way off.  Some have unanswered concerns from core review and have 
> been left dangling. *If you're the original developer or otherwise 
> interested in working these patches through to completion, please do get 
> involved.*

As a first step it'd be great if we could extract from this a list of 
stuff that looks promising but hasn't been reviewed by a core. Those 
would be the next highest priority to review I'd expect.

> When I started this little mission I wasn't quite sure what to expect.  
> What I have found is that as much as there was anticipated cruft to 
> clear out, there is a great deal of very good work lurking here, waiting 
> to see the light of day, and it would be so good to see this work 
> realised.  :-)

+1

> If you have anything to say, feel free to write back on list, and if 
> you'd like to coordinate with me any efforts with these patches I can be 
> found by email or on Freenode in the #heat channel; I'm dpeacock.
> 
> Based on feedback of this idea, and indeed on each individual patch, I 
> hope we can get this backlog under control, and harvest some of this 
> excellent code!
> 
> Thank you,
> David Peacock
> 
> [0] https://ethercalc.openstack.org/b3qtqyhkg9g1 Please be mindful of 
> accidental edits.
> 
> 




More information about the openstack-discuss mailing list