UDS/Maverick

Not logged in - Log In / Register

This page contains notes and links related to Launchpad from UDS Maverick (Belgium, May 2010).

The idea is to collect raw data here and then quickly process it into a more useful form.

Relevant sessions

Note that this deliberately excludes translations stuff, since jml doesn't feel qualified to filter translations gobby notes 

  -rw-r--r-- 1 jml jml  3459 2010-05-18 10:54 arm-m-archive-features
  -rw-r--r-- 1 jml jml  3193 2010-05-18 10:54 arm-m-automated-bootstrap
  -rw-r--r-- 1 jml jml  2791 2010-05-18 10:54 arm-m-automated-testing-framework
  -rw-r--r-- 1 jml jml  1928 2010-05-18 10:54 arm-m-derived-archive-rebuild
  -rw-r--r-- 1 jml jml  1150 2010-05-18 10:54 arm-m-development-environment
  -rw-r--r-- 1 jml jml  4800 2010-05-18 10:54 arm-m-image-building
  -rw-r--r-- 1 jml jml  1523 2010-05-18 10:55 bazaar-m-agenda
  -rw-r--r-- 1 jml jml  4250 2010-05-18 11:26 bazaar-m-healthcheck
  -rw-r--r-- 1 jml jml  2472 2010-05-18 10:55 community-m-adopt-upstream
  -rw-r--r-- 1 jml jml  3680 2010-05-18 10:55 community-m-development-workflow-review
  -rw-r--r-- 1 jml jml  2643 2010-05-18 11:20 community-m-document-daily-builds
  -rw-r--r-- 1 jml jml  1111 2010-05-18 10:55 community-m-harvest
  -rw-r--r-- 1 jml jml  3570 2010-05-18 10:55 community-m-launchpad-bugs
  -rw-r--r-- 1 jml jml  3209 2010-05-18 10:55 community-m-launchpad-patches
  -rw-r--r-- 1 jml jml  3319 2010-05-18 10:55 community-m-patch-review-process
  -rw-r--r-- 1 jml jml  3061 2010-05-18 10:55 desktop-maverick-desktop-as-IDE
  -rw-r--r-- 1 jml jml  1916 2010-05-18 10:55 foundations-m-cycle-review
  -rw-r--r-- 1 jml jml  3054 2010-05-18 13:47 foundations-m-distributed-development-review-and-planning
  -rw-r--r-- 1 jml jml  1608 2010-05-18 10:55 foundations-m-robust-python-packaging
  -rw-r--r-- 1 jml jml  2856 2010-05-18 10:55 foundations-m-software-center-roadmap
  -rw-r--r-- 1 jml jml  2401 2010-05-18 11:23 kernel-maverick-bug-process
  -rw-r--r-- 1 jml jml  2604 2010-05-18 10:55 kernel-maverick-launchpad-bug-requests
  -rw-r--r-- 1 jml jml  1901 2010-05-18 10:55 qa-maverick-improving-communication
  -rw-r--r-- 1 jml jml  2827 2010-05-18 10:55 qa-m-launchpad-test-cases
  -rw-r--r-- 1 jml jml   626 2010-05-18 10:55 server-m-daily-builds

Ad-hoc feature requests

These are requests that jml got in corridors and side conversations at UDS.

Other notes

Soyuz/UDS-Maverick has some stuff relevant to Soyuz.

TODO

Hallway Conversations

Bryce

I talked a lot about the 'Bug Q&A' concepts with various distro people, all of whom thought it sounded brilliant and potentially a good solution to the issue of so many low quality bug reports. Specifically, we discussed the idea of a rating score that indicates 'completeness' of the report and encourages reporter to provide more detail; such a score could better enable developers to focus on high quality bug reports (by excluding bugs below a threshold), and enable triagers to identify and programmatically close the lowest quality reports at the end of a release to clean up the tracker, and give bug reporters some clearer clue about why their report may not be getting attention.

I was a bit surprised at how positively it was received when I mentioned I was getting my feet wet by working on improving the API docs. It seems I've not been alone at finding the API docs a bit too vague in parts.

Session Notes

desktop-maverick-desktop-as-IDE

Fairly high level / conceptual discussion. Launchpad really didn't come up except in the context of groundcontrol. I don't recall any specific Launchpad action items or ideas.

kernel-maverick-bug-process

Planning on an online 'triager summit' in near future to train triagers. Planning a focus on bugs with patches (or SHA1's). Discussed improvements to various launchpadapi scripts shared with X and other teams, and expanded use of tagging to sort their pile of bug reports. Anticipates expiring about 2/3rds of the kernel bugs this month.

Kernel team has requested to turn off dupe suggestion by launchpad for bug reporters. (See next session)

kernel-maverick-launchpad-bug-requests

1. Bug duplication. Hardware-specific bug reports tend to get set as duplicates improperly by bug reporters, because they match based on similarity of *symptoms* when in reality they need to be matched on *symptoms AND hardware*. This is resulting in bug reports falling through the cracks and not getting fixed. For maverick the kernel team wants to switch off any automated or tool-driven mechanism that sets or suggests dupes, and leave this work to triagers.

2. Comment Hiding. Needed for cleaning up 'me-too storms' to make launchpad bugs more palatable to upstream developers, by hiding less relevant comments.

3. Fix Released locking. Disallow re-opening of bugs marked Fix Released except by bug supervisors. Was suggested this be done only for HW-specific packages (although I mentioned this in a Desktop team session and this idea got applause and ooh's from everyone there, so there may be wider interest in this).

4. Lock down. Problem is that certain bugs turn into 'me-too storms'. These bugs get looked into by upstream developers, who are put off by the excessive dogpiling and thence get a dim impression of launchpad bugs and become less likely to help with Ubuntu bugs. Desire is for a mechanism for triagers to switch off comments and status changes from random people other than the original reporter and bug supervisors.

ACTIONS FOR LAUNCHPAD TEAMS:

  1. Disable dupe suggestion for all source packages the kernel team is a supervisor for (i.e.: https://edge.launchpad.net/~ubuntu-kernel-team/+packagebugs )

  2. Implement mechanism for hiding comments (LP #522741, #557386, #58670) 2.1. Near term - Change from displaying the first 80 comments to the first 40 and last 40 (LP #335120)
  3. Implement (possibly per-package) way to control who can re-open fix released bugs
  4. Implement a 'lock down' functionality (LP #73122)
  5. Get greasemonkey scripts higher profile on LP web docs

desktop-maverick-gpu-hang-reporting

Plans involve turning on higher level of debugging info with certain types of X failure modes. Want to investigate having apport detect if debugging was not on, and offer to reconfigure system to rerun with it switched on.

A concern is that this may result in *much* larger logs (GiB's), so X/kernel team has some actions around avoiding or minimizing that. No action for Launchpad except maybe to be aware of this potential.

The kernel team's request to disable dupe suggestions was discussed. The X.org team generally feels this is not as significant a problem for them (we think because the team updates bug titles pretty aggressively to include chipset and detailed descriptions, so false dupes occur less often). So they would prefer to not opt-in to having dupes disabled.

desktop-maverick-kms-bug-collaboration

X team uses a bunch of launchpadapi scripts for capturing and processing bug reports, this discussion involved a bunch of ideas for extending and improving them. Nothing particularly Launchpad-specific, but I did offer to help on three tasks that are considerably easier for me to do than would be for anyone else.

dx-m-qa-workflow

DX team changes impact Desktop and are quite visible. DX team is more like an upstream in this context. DX needs better follow up on testing, triaging and fixing bugs. No Launchpad action items.

community-m-launchpad-bugs

1. Dupe detection should search fedora bugs, bts bugs, xfce, upstream bugs -- every relevant bug -- as well as Launchpad bugs.

2. Full import and synchronization of external bug trackers into Launchpad

3. Don't want bugs to be re-opened when they are marked closed by someone who is a bug supervisor (already works for wontfix) [Was also discussed in kernel session, above, by same folk.]

4. Locking down comments by the bug supervisor [Was also discussed in kernel session, above, by same folk.]

5. Moderation / cleanup of bug comments -- or a selective view [Was also discussed in kernel session, above, by same folk.]

6. Make it easier for people to say that the bug affects me too (possibly auto-detecting "me too" comments)

7. Disable duplicate bug finder for hardware-driven projects / packages [Was also discussed in kernel session, above, by same folk.]

9. Dupe search when you are on the "Mark as duplicate" form

10. "Fixed elsewhere" shows a lot of false positives (LP: 232545, 200890, 239604, 239606, 239608, 239609)

11. UI review for linking a bug to an external upstream => being worked on right now

12. Status listing for upstream bug tracker synchronization so Jorge can figure out who he needs to talk to

13. Easy forwarding of bugs upstream? Maybe, but only for competent humans. Bryce gave a demo of a tool he uses. He's written a CGI script that currently runs on his laptop that pretty much populates a bug form ready for sending upstream. We want this in Launchpad.

ACTIONS FOR LAUNCHPAD TEAM

community-m-launchpad-patches

Mainly discussed the new patches view and some stray issues related to it.

For projects that are hosted on Launchpad, it's easy to miss patches that are on the upstream or on the package. Launchpad should help you find the upstream patches if you are on the package page, or the package patches on the upstream page.

We refer to 'patches' but really what we're talking about are 'fixes'. Not all patch files are actually 'fixes', and some fixes are not 'patches' (e.g. branches).

Ultimately, if we want something more fully featured, we should integrate the patches view with Launchpad's existing merge proposal.

ACTIONS FOR LAUNCHPAD TEAM

UDS/Maverick (last edited 2010-05-18 19:41:48 by bryce)