Skip to content

test: seed retries on robur; conformance on Bedework 5, gmx and robur - #58

Open
tobixen wants to merge 6 commits into
mainfrom
test/robur-seed-retry
Open

tobixen wants to merge 6 commits into
mainfrom
test/robur-seed-retry

Conversation

@tobixen

@tobixen tobixen commented Nov 17, 2026

Copy link
Copy Markdown
Member

Disclaimer: AI-generated code and text below the line, generated by Claude Opus 5.5, from the prompts:

[pick from the task list: "calendaring: remaining conformance failures on Ox, Zimbra and others"] (...) One of the priorities should be to ensure the compliance tests passes (sic) on all the servers, was that task completed yet?

Continue. Robur should pass, if just retrying hard enough. I suppose it is possible to run several times with pytest --last-failed or something like that?


Robur leaves about one request in 60 unanswered (a 70-second hang, then a reset) and answers some PUTs with a 500. The listing test seeds 120 items, so it failed nearly every run on a write it is not about, however often it was rerun. The CalDAV suite's seed() now tries a failed write up to three times, taking the item as stored when it can be read back, and raises a clash that no retry explains. With it, and --last-failed for the rest, every conformance test passes on robur.

docs/caldav-servers.md replaces the stale 2026-11-03 "known failure" with today's results: Bedework 5 passes (caldav now drops the ATTENDEE it stores twice), gmx passes, robur passes on rerun; the servers not run are named. Also a one-clause fix to the #29 changelog fragment, so it matches API_DESIGN on when native is None.

Tests: the unit suite (3306 passed); live runs on Bedework 5 (sync suite), gmx and robur.

🤖 Generated with Claude Code

Robur leaves about one request in 60 unanswered (a 70-second hang,
then a reset) and answers some PUTs with a 500, so the listing test,
which seeds 120 items, failed nearly every run on robur on a write it
is not about.  The CalDAV suite's seed now tries a failed write up to
three times, taking the item as stored when it can be read back.

Prompts:
> [pick from the task list: "calendaring: remaining conformance
> failures on Ox, Zimbra and others"] (...) One of the priorities
> should be to ensure the compliance tests passes (sic) on all the
> servers, was that task completed yet?

> Continue.  Robur should pass, if just retrying hard enough.  I
> suppose it is possible to run several times with `pytest
> --last-failed` or something like that?

Assisted-By: Claude Opus 5.5
The 2026-11-03 known failure, added ATTENDEE parameters, has been
handled since; the section now gives the latest results per server.

Prompts:
> [pick from the task list: "calendaring: remaining conformance
> failures on Ox, Zimbra and others"] (...) One of the priorities
> should be to ensure the compliance tests passes (sic) on all the
> servers, was that task completed yet?

> Continue.  Robur should pass, if just retrying hard enough.  I
> suppose it is possible to run several times with `pytest
> --last-failed` or something like that?

Assisted-By: Claude Opus 5.5
The changelog fragment said a written item's `native` stays None until
it is read back.  The design document is more precise: only when the
server answers the write with an ETag, until get(), search() or
reload() reads it.  Fragment of
#29.

Prompt: [fix issues found during the review process]

Assisted-By: Claude Opus 5.5
Infomaniak stores URL as URL;VALUE=URI.  URI is URL's default type
(RFC 5545 section 3.8.4.6), so nothing was lost, yet verify reported a
difference and the conformance round trip of the time-zone series
failed there.  A VALUE parameter now compares as absent when it names
the property's default type, taken from icalendar's own type table.

Prompt: [Continue working on getting the conformance tests passing]

Assisted-By: Claude Opus 5.5
Reviewed-by: Tobias Brox <tobias@redpill-linpro.com>
Results of the 2026-11-17 runs; ecloud is still being run.

Prompt: [Continue working on getting the conformance tests passing]

Assisted-By: Claude Opus 5.5
Reviewed-by: Tobias Brox <tobias@redpill-linpro.com>
Prompt: [fix issues found during the review process]

Assisted-By: Claude Opus 5.5
Reviewed-by: Tobias Brox <tobias@redpill-linpro.com>

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant