Biweekly Triaging module meeting for planning and coordination.
Folks on the payroll should attend if possible, others are invited as well.
In these meetings, we can keep us up-to-date on everything triaging, discuss recent important issues and see if we can bring the module forward.
I believe it is also good to see human faces once in a while :), so there is always room for some personal stuff if appropriate.
The meeting will be on 2025-06-05T11:00:00Z. It is open for everybody interested to join the video call (link below).
Links
- Video call google meet for now, we can try jitsi as well if people prefer
- #module-triaging matrix chat room on chat.blender.org
- #module-triaging matrix chat room link for other matrix clients
- live meeting notes
Attendees
Topics
- Announcements
- 4.5 is Beta now
- Untriaged reports are slowly growing (109 at the time of meeting)
"Pending TODOs/followups (wont get an active topic if no news have to be announced)"
- Open PR list growing, can Triagers help in that?
- Usage of LLMs
- this doc has been forwarded to both Sergey and Dalai
- talk about Proposal - Label Changes
- Alaska has made a tool to automatically notify people of old
Needs information from user
reports: Sign in - Blender ID - blender.org - Feedback regarding our triaging infrastructure
- We use a script to help write our weekly reports. There have been a few issues with it’s output recently (E.g. Duplicate data, missing data from PRs merged by others)
- Yimming was planning to look into this
- We use a script to help write our weekly reports. There have been a few issues with it’s output recently (E.g. Duplicate data, missing data from PRs merged by others)
- Wizard
- Gather a list of two or three examples from “out there” that work similar to what we want to achieve (add that to Sign in - Blender ID - blender.org) is still pending, some more discussion in Sign in - Blender ID - blender.org though regarding Richards experiments
- Could we add a feature that lets users know the location of crash logs after Blender crashes so they can more easily provide it in a bug report?
- There has been progress on the crash window pull request
- Discussion about exposing dependency cycles to the end user (an idea we have discussed in previous meetings)
- Design task: Sign in - Blender ID - blender.org
- PR: Sign in - Blender ID - blender.org
- Discussion about Tests:
- reply in devtalk
- Status of Open pull request numbers
Meeting Notes
- TODOs/followups
- Discussion about exposing dependency cycles to the end user
- As part of the project, Yiming was looking at other tools they add to the dependency cycle viewer. An example being Mesh validation tools for things liike 3D printing. However most of this is already covered by third party addons.
- Status of Open pull request numbers
- Pratik has been pinging people on some pull requests to help bring the number down, but it hasn’t really shifted (Open pull requests are at 1139)
- Maybe an announcement should be made about this issue to encourage developers to go over their pull request backlog?
- Sometimes triagers make reports about issues that can only be reproduced on specific hardware that other’s don’t have. Should we mark these reports as confirmed as triagers are generally reliably people?
- It was agreed upon that these can probably be marked as confirmed.
- Quickly checked to see how triagers were using LLMs for triaging.
- For most people, they haven’t found the time or a need for it yet.
- In their spare time, Alaska has been running experiments on areas where a LLM could be used for triaging. Examples include:
- Use a LLM to write out the steps to do a common thing in Blender faster than a human could.
- In their experience, the output is either wrong, or too long and convoluted.
- Adjustments to the prompts helped with the “too long and convoluted” issue, but didn’t resolve the occasional incorrect information.
- Giving the LLM access to search tools helped fix the incorrect information, but the source for the correct information just ended up being Alaska commenting the steps in other reports.
- At this point it seems a better use of time to add these common “steps to do something” to the triaging playbook.
- Sometimes when encountering a issue, triagers either know, or have the vague feeling that things are working as expected. So they should close the report and explain why it’s working as expected.
- Alaska gave this a try with some known issues/unintuitive behaviour in Cycles and the LLM typically gave incorrect information. So these can’t be relied on 100% for this type of task. Especially if it’s a vague feeling that things are working as expected and you believe the LLM will be able to clearly and correctly explain whether or not it is or isn’t expected.
- Use a LLM to write out the steps to do a common thing in Blender faster than a human could.
- Discussion about exposing dependency cycles to the end user
Following meeting
The following meeting will be on 2025-06-19T11:00:00Z. Again it will be open for everybody who’s interested.