Campaigns Sprint 1 Tracking issue

[chrispine]

Plan

Note: We have moved to 2-week sprints! This one is our first.

The main theme of this cycle is quality of life improvements for our users. Campaigns could be a lot easier to use! Many of these improvements will be around debugging a campaign, to shorten the cycle of “tweak the spec, apply the spec, see what went wrong, repeat”.

We also have several tickets that continue the theme of the previous cycle: bug fixing and polishing.

User feedback (from customers and internal folks) is driving our focus. We are also finishing up a round of formal user testing to further validate that we are focusing on the most urgent needs.

Availability

Period is from Sep 21 to Oct 6, for 12 working day in total before branch cut.

If you have planned unavailability this iteration (e.g., vacation), you can note that here.

@chrispine 3d (Sep 23-25)
@mrnugget 1d (Sep 25)
@eseliger 0d

Tracked issues

@LawnGnome: 2.50d

Completed: 2.50d

• (🏁 18 days ago) Add backlink to PR description again (#13459) 0.50d
• (🏁 18 days ago) Decorating changeset body fails with "Namespace not found" (#14159)
• (🏁 11 days ago) Allow users to publish only a subset of changesets (#13139) 2.00d
• (🏁 11 days ago) doc: fix published array ordering (#14354)

@chrispine: 0.50d

Completed: 0.50d

• (🏁 21 days ago) Remove old code in 3.21 release (#13018) 0.50d 🧶

@eseliger: 5.00d

Completed: 5.00d

• (🏁 18 days ago) campaigns: cancelled GitLab pipelines don't take check state out of pending (#12667; PRs: #14129) 1.00d 🐛
• (🏁 13 days ago) Refactor internal campaigns package (#13174; PRs: #14188) 2.00d 🧶
• (🏁 13 days ago) Merge changesets.CreatedByCampaign and changesets.OwnedByCampaign fields (#12650; PRs: #14157) 1.00d
• (🏁 today) Expose author details on changeset spec description in GQL (#13645; PRs: #14181) 1.00d

@mrnugget: 5.00d

Completed: 5.00d

• (🏁 24 days ago) Documentation for configuring campaigns (scope for tokens) (#12888; PRs: #13948) 0.50d
• (🏁 14 days ago) Fix nil-pointer panic in go-diff (#13976) 🐛
• (🏁 13 days ago) Changesets are not re-opened and re-attached to a campaign if new apply adds them again (#13849) 2.50d 🐛
• (🏁 13 days ago) Split up ChangesetCountsOverTime integration test (#12835) 1.00d 🧶
• (🏁 11 days ago) Create integration test for src-cli (#13863) 1.00d

Legend

• 👩 Customer issue
• 🐛 Bug
• 🧶 Technical debt
• 🎩 Quality of life
• 🛠️ Roadmap
• 🕵️ Spike
• 🔒 Security issue
• Pull Request

[mrnugget]

Weekly update

Last week I focused on recording a new demo video for campaigns, which is now live, and thus spent a lot of time running and re-running campaigns, which resulted not only in a few small bug fixes, but also one of my main goals for this sprint: improve the feedback loop and debugging experience when building campaigns with src-cli.

That's what I want to focus on this week, too: improve the error reporting in src-cli, make the feedback loop tighter, increase performance, add tests, etc.

My other goals for this sprint include fixing a nasty bug in our reconciler/applyCampaign logic. I might do that once I reach my "eureka!"-moment with it. Oh, and I also found a tasty bug in go-diff that I'm also marinating in the back of my head.

[eseliger]

Weekly update

Last week I was out for vacation.

This week, I spent today on catching-up on mails, threads and discussions. Tomorrow, I will look deeper into the work to be done this iteration (or do we say sprint now?). Since my largest ticket, responsiveness, is dependent on design work, I'll be starting off with my other tickets first.

[LawnGnome]

Weekly update: 2020-09-14 to 2020-09-18

Last week was a bit of a struggle. I worked on implementing the changeset subset behaviour that we've been working on through the RFC process, and generally cleaning up RFCs, but got a bit less done than I'd hoped. I also pushed a few minor documentation updates for 3.20. In general, though, I was glad to have Friday off and reset.

This week, I'm going to be getting the subset behaviour landed come hell or high water (hopefully neither actually happen), trying to figure out the worker util issue I've been intermittently banging my head against for a while, and, most importantly, ensuring that campaign-created changesets have a link back to the campaign.

[mrnugget]

Weekly update

Last week I fixed multiple small things in src-cli and improved the error reporting. I also managed to fix the nasty bug that pops up when detaching/attaching changesets in applyCampaign, see https://github.com/sourcegraph/sourcegraph/pull/14099 for details, including a video walkthrough of the bugfix and how cool it demos the declarative model.

My goal for this week is to make progress on multiple smaller tickets, such as adding a src-cli integration test, splitting up a test in sourcegraph/sourcegraph, and fixing a bug in go-diff. Besides that I still want to make some progress improving error/debugging in src-cli.

[eseliger]

Weekly update: 2020-09-21 to 2020-09-25

Last week I first played catch-up from vacation the week before, then went onto tackling the non-design-related issues. Opened PRs for those, need to fix up some minor things and then they should be good to go.

This week, I'm planning to make all about responsiveness of the campaigns pages. The designs are ready for review and I'll be taking time today or tomorrow with design to go over it. Afterwards, fun implementation work. That should tick off most of my work items for this sprint, so currently seems to go according to plan.

[LawnGnome]

Weekly update: 2020-09-21 to 2020-09-25

Last week, I worked on a couple of key items in my sprint plan: adding backlinks back into campaign created changesets (which was a fun expansion pack project that came in way over its original time estimate, since I ended up cleaning up internal/db/namespaces.go fairly significantly as part of adding the functionality I needed specifically for this task), and updating RFC 228 to reflect what I really, really hope is the final version of changeset template specialisation, along with its associated PRs (which I think are more or less review ready, assuming we approve that RFC).

I also had a crack at GitLab integration tests, but had trouble with vcr, which isn't designed to integrate with how we use HTTP clients in the GitLab package. More thought required here, but I wasn't happy with the approach I was looking at, so I want to come back to this afresh at a later date.

This week, I want to get RFC 228 really, truly, done, but expect to spend most of my time working on support for using user tokens when interacting with code hosts. Around that, in my copious amounts of spare time, I may have another go at the GitLab integration tests, but I'll likely prioritise validating the reconciler logic above that as and when I have cycles.

[mrnugget]

Weekly Update

Last week I reached the goals I set for myself that week: wrote an integration test for src-cli, split up a test in sourcegraph/sourcegraph, fixed the bug in go-diff and added some more tests to go-diff. The rest of my time I spent in src-cli, trying to make it easier to debug.

The first result of that was unintended but, at least in my opinion, a possibly great way to add more power to campaigns on the src-cli: add templating support for campaign specs to allow run steps to access only the files yielded by search results and/or the files modified by a previous step. The design is not final and implementation wise it's still a prototype, but I'm looking for feedback on it and I want to spend more time on it this week.

The other result of my digging in src-cli was support for "status bars" next to a progress bar. Again, I think this is really handy and improves the feedback loop a lot, since you can see what's currently being executed in which repository. I also want to clean this up further this week.

Then we also have planning this week, which means I anticipate some backlog grooming or thinking about what we want to do.