Servers by jDrones

Massive backlog of PRs and Issues

Topic: reviewing PRs and Issues is not sexy work ,and we have a big backlog of open PRs and Issues.
https://github.com/ardupilot/ardupilot/pulls ( over 330)
https://github.com/ardupilot/ardupilot/issues ( over 1200)

Proposal type: Hardware [ ] , Software [ ] , Other [ X] : _________________

Description: I believe we currently partially-fund Francisco’s time on a fixed $Z/mo or X hrs per week basis for him to do as much of this as he can given the $/time constraints, but it’s clearly not enough, and this proposal is for us to spend more $.

Planned amount $$ (USD): my suggestion is that we spend an additional 20k/yr on top of our existing expenditure to Franscisco. This proposal is NOT to just give Franscisco double the work ( i don’t even know if he wants it at this stage). The proposal is about the work that needs doing, and getting a budget to do more of it.
Later on, after the funds are approved and the potential candidate/s to do the work highlighted, if there is more than one, the dev team may need to take the candidate/s to a second vote to decide who gets funded to do this.

Estimated time for completion: ongoing at $20k/yr.

3 Likes

It could be a good idea !
I tried to push forward a lot of PR, generally they only miss something like 30min rework to be ready for merge. The main blocking point I see on PR is the final validation that is generally delayed forever … I don’t blame anybody, our maintainer are already doing an awesome job.

Maybe schedule some merge session/night once a month an pay for pizza and beer for participant ?

1 Like

Thanks for the proposal Buzz.

I’m leaving my role so that will be open for someone else to take, but I’ll comment on your request to extend that into a full-time job (or two part-times for different people).

This is something I thought about asking for myself. However, it’s not viable on the long term yet - we don’t have enough Partners to guarantee that the following year we’ll have that amount of money to pay.
If we have someone (qualified for the job) that is willing to take that risk, then I would support the idea.

The value is a bit above what I got, but I’m not shocked by it, I think it’s within an acceptable range. More important though might be finding people willing to work for that amount.

Thanks you for your inputs Francisco !

I am bit split on a full time proposal. Besides that it is quite risky on long term, IMHO having two reviewers will give us more flexibility and broader range of competencies.

I would be interested in a half time reviewers position. Even if I don’t have Francisco reviewer skills or knowledge on some part of the codebase, I have been doing it often on the past years and know well SITL and autotest features.

1 Like

Francisco is no longer on the payroll, so this is a little out of date
now.

I’ll put my hand up to take on the role that Buzz is suggesting here.

1 Like

So, just to be clear… Now that francisco has stepped down from his role, this proposal should really be for two expenditures of $20k USD per-annum, and just like how franscisco’ was setup, they would each expire yearly and be subject to review/renew each year, one should be to essentially replace the work Francisco was doing, and the second one to expand on his effort/s and aim to get ‘more’ done than was able to be done by Franscisco alone… ie a total of $40k usd. These assignment of $ need not necessarily go to the same person, but there’s no reason i can see to say that should be denied either. finally, i suggest that when it comes time for a vote to be taken by the group, that it should be voted on as two separate but related item/s in any any vote taken…
eg:
item one: Do you support an expenditure of $20kusd to go to paying someone to continue doing essentially what Francisco was doing previously( review and triage of open issues and PRs , merging where authorised and/or obvious, giving feedback, etc )?
item 2: Do you support an additional expenditure of $20kusd to do the same thing as item 1, but to hopefully extend the amount we get done )?

Buzz, this needs to go to the Dev mailing list etc.
I think it might be wise to go with a single proposal first, and vigorously debate expanding it during the Conf in March.

To expand on that: there isn’t enough detail in this proposal to make a decision.
To engage someone, a contract must be struck with SPI, so this proposal needs to have the necessary details to create that contract. As the amount is greater than what the Funding Committee can approve, it needs to go to a full team vote. Defining the role/responsibilities/deliverables has to happen before that vote, so that the team has sufficient information to actually approve it, and I really think that the dev mailing list is better than here to get the right engagement and feedback on that.

Servers by jDrones