2024-09-18 release planning meeting¶
Attendees: Adam (partial), Anish, Ghislaine, Kailash, Kriti, Pavish, Sean
Current status of work¶
Sean’s high-level status report¶
-
Database Page: ✅
- RPC changes done
- Ready for QA except:
- Permissions features (which I can’t comment on)
-
Schema Page: ✅
- RPC changes done
- Ready for QA except:
- Explorations features
- Permissions features (which I can’t comment on)
-
Import page: ✅
- RPC changes done
- Ready for QA (I think we should perform extensive QA here)
-
Explorations: ❌
- RPC needs extensive front end work still, and likely some minor backend work too
- It’s difficult to accurately estimate the remaining time. Best case: ready for QA Sept 25. Worst case: ready for QA Oct 8.
-
Record selector: ✅
- RPC changes done
- Ready for QA
-
Table page: 🟡
- Most front end work done.
- Tiny bit of front end work blocked:
- “split table” & “move columns” features in draft PR with further front end progress blocked by backend issue Return value needed from data_modeling.split_table
- Some QA could potentially begin, with knowledge of above limitations
-
Record page: 🟡
- Front end work likely done.
- QA blocked by backend issue: Make records.get work with stringified PK values
Pavish’s status report - permissions¶
- Database page
- Settings tab - Complete, ready for QA
- Permissions modal
- ‘Share’ tab - Complete, ready for QA
- ‘Transfer Ownership’ - pending
- Schema page
- Permissions modal
- ‘Share’ tab - Complete, ready for QA
- ‘Transfer Ownership’ - pending
- Disable actions based on user’s privileges - pending
- Permissions modal
- Table page
- Permissions modal
- ‘Share’ tab - Complete, PR open
- ‘Transfer Ownership’ - pending
- Disable actions based on user’s privileges - pending
- Permissions modal
ETA for getting PRs open for the above pending functionality.
- 20th EOD.
- My current buffer is exhausted, so I might end up using the weekend as a buffer period if something comes up.
Anish’s update¶
- All issues blocking Sean are resolved.
- Only 4-5 backend endpoints left for permissions, ETA Friday unless there’s more blocking work from Sean.
Ghislaine’s update¶
- QA script for permission is in review.
- Visual consistency / improvements work PRs are in progress or review, should be done by early next week.
- Hasn’t started on visual improvements for permissions, planned for next week.
Adam’s availability¶
- Adam is available to help with QA next week.
- Either in a session, or independently with instructions.
Release work plan and timeline¶
Here’s the rough work plan and check-in points for the release.
We won’t have a specific date we’re aiming for, since the work is pretty well defined but has some unknowns. - Instead, we’ll set up several checkpoints, completing each checkpoint will also trigger beginning other work. - We’ll also check in at the daily progress meetings and evaluate if we can cut scope.
Individual tasks we’re working on, other than below checkpoints¶
- QA script
- Visual improvements work
- Other small issues that may come up
- Kriti will continue to look through the notes and flag anything we missed.
- No need to track record summary UI changes or hiding public shares, they are alrwady done.
1. All RPC frontend work merged except explorations.¶
All balls for this currently in Sean’s court. Once this is done:
- Ghislaine will start (unstructured) QA on the
develop
branch locally.- Mathesar should be set up in production mode for QA
- Pavish will help Ghislaine set it up.
2. All planned work on permissions is merged.¶
Pavish and Anish are working on this. This triggers:
- Kriti & Ghislaine will do a visual review of permissions.
- Anish will deploy a QA server.
- Kailash will run through the QA script (except explorations).
- Adam will help with QA.
- Ghislaine will work on updating user documentation.
3. Explorations RPC work is done.¶
- QA on explorations begins,
- Everyone participates in QA.
4. Installation documentation work is done.¶
- Administrator-focused QA begins.
- Triage of QA issues to figure out what to fix / what to defer.
- We fix critical QA issues.
- Review and fix issues with user documentation
5. QA + critical fixes complete¶
- Release process begins
Priorities for each person¶
Adam¶
- Perform QA when ready.
Anish¶
- Backend endpoints for permissions
- Unblock Sean with backend fixes as needed
- Installation testing & docs
- Talk to Kriti to break down task before getting started.
- Perform QA when ready.
Ghislaine¶
- Wrap up visual improvements PRs ASAP to make room for other work.
- Finish QA script work so QA is unblocked.
- Review and visual improvements for permissions.
- User docs for permissions.
- Talk to Kriti to break down task before getting started.
- Perform QA when ready.
Kailash¶
- Review QA script from PoV of someone who will be using it.
- Perform QA when ready.
Kriti¶
- Continue release planning and organization
- Perform QA when ready.
Pavish¶
- Get permissions done.
- Perform QA when ready.
Sean¶
- Merge non-explorations RPC work
- Explorations RPC frontend work
- Perform QA when ready,
Release logistics¶
Name¶
What do we call the release?
- “release candidate” is probably not good, implies too much polish.
- Language like “testing build” probably makes more sense.
- “Pre-beta testing build 1” works for people.
Version¶
- Do we even need a version?
- We’re not supporting upgrades or tagging as
latest
- We’re not supporting upgrades or tagging as
- Where do we use the version?
- Someone should chase this down by looking at the code.
- Off the top of our heads:
- GitHub release “object”
- Docker build
- Docs, marketing updates, blog post, social media, etc.
- GitHub tag
- We need something to name the GitHub tag, etc.
- Shouldn’t be
0.1.7
since this is a very different Mathesar from last release, same version number doesn’t indicate that. - Should not be
0.1.8
, this is not an upgrade (yet) - Should not follow
number.number.number
pattern to make it clear it’s not a usual version.
- Shouldn’t be
- Beta will probably be
0.2.0
- Not
1.0-beta
because we don’t have a fixed plan for 1.0.
- Not
- We could do something like
0.2@next.1
if we went with how Svelte does things. They have a4.2.19
version but also a5.0.0@next.X
for next version testing builds.- The
@
is weird. Other options- 0.2.0-testing-1
- 0.2.0-testing.1
- 0.2.0.testing-1
- The
- DECISION:
0.2.0-testing.1
wins!
Communication¶
- Do we need to make a GitHub release?
- No, but GitHub does support marking releases as “pre-releases” and not showing as latest.
- Should work for us without any code changes.
- Having a GitHub release will make the build more visible.
- Let’s do it.
- Sean to do some investigation to ensure that this doesn’t screw up anything for
0.1.7
BEFORE we make the actual GitHub release changes.
- No, but GitHub does support marking releases as “pre-releases” and not showing as latest.
- We should do:
- Matrix post
- Blog post
- Dev mailing list post
- Twitter post
- We need to specify “DO NOT UPGRADE” prominently in all comms
- Communication ideas
- We can use the concept of a “calf” release (riffing off our elephant logo) to convey that the release is still a bit unsteady on its feet and needs to grow up. :)
- Good motif to use in comms
- Can also use “mammoth” motif for indicating final / stable version.
Publication¶
- Where do we host testing instructions?
- Blog post seems more appropriate than docs.
- But we need a place to host docs for the testing build.
- New URL?
- mkdocs version switcher? (we’ve been kicking that can down the road)
- Django has development docs hosted this way.
- Let’s do mkdocs version switcher unless it proves untenable.
- Sean to try setting it up.
- Instructions can live on the docs like the other releases, but the docs site will default to
0.1.7
. - We can link to appropriate versioned docs for testing build when we make our comms about the release.
- Docker repo
- Can we use Mathesar debug, to indicate it’s a testing build?
- Yes, everyone’s in agreement.
- We should investigate turning on the debug logging like we have for other debug images
- Only if it doesn’t affect performance.
- We will investigate.
- Aside: for final beta version, we should consider setting up
mathesar/mathesar
repo instead ofmathesar-prod
- Can we use Mathesar debug, to indicate it’s a testing build?
Community engagement¶
- What to ask community for feedback on?
- Performance improvements
- Permissions, how does this work for your use case, etc.
- Usability feedback?
- Give them some hints / specific questions to guide feedback.
- Need to work on what those are.
- That’s enough, don’t want to overwhelm.
- How to collect feedback?
- Multiple channels to make it easier for different kinds of users.
- Consider issue template on GitHub
- plus issue form
- for people who prefer structured feedback (like Kriti)
- Also encourage email or Matrix feedback that’s unstructured for people who prefer that (like Sean)
- Can also open a single GitHub discussion thread for reactions on comments about the release.
- We should reach out to specific users that we’ve corresponded with and we think would be good fits to test.
- Can look through our internal CRM.
- Kriti has some idea who to start with, should reduce work needed to look through CRM
Task management¶
We did not have time to look through GitHub & Basecamp in the meeting, Kriti will follow up via email.
Kriti’s notes on this topic¶
- Ensure 22 open items on Pre-beta test build #1 milestone belong there.
- Should we split items we’re not doing for this release into new issues?
- Walk through work tracked in Basecamp, see if we need to put anything in GitHub
- Need to create issues for:
- table / schema
PUBLIC
issue, needed for permissions to work
- table / schema
- Could be more smaller issues, still going through all the meeting notes.