Skip to content

Latest commit

 

History

History
184 lines (126 loc) · 14.9 KB

CHANGELOG.md

File metadata and controls

184 lines (126 loc) · 14.9 KB

Changelog

The format of this file should adhere to Keep a Changelog.

Changelogs prior to v1.2 follows other formats and are available in the v1.2-release.

This project should adhere to Semantic Versioning, though some earlier releases may be incompatible with the SemVer standard.

[2.0.0] - [Unreleased]

In version 2.0, the requests library will be replaced with niquests or httpx. See #457. Master branch is currently running niquests.

(Also, in version 2.0, support for python 3.7 will be dropped, possibly also python 3.8. Master branch supports both as for now).

[1.5.0] - [Unreleased]

Changed

Refactoring

  • Minor code cleanups by github user @ArtemIsmagilov in #456

Test framework

  • Python 3.13 is officially supported by github user @ArtemIsmagilov in #454
  • Functional test framework has been refactored in #450
    • code for setting up and rigging down xandikos/radicale servers have been moved from tests/test_caldav.py to tests/conf.py. This allows for:
      • Adding code (including system calls or remote API calls) for Setting up and tearing down calendar servers in conf_private.py
      • Creating a local xandikos or radicale server in the tests.client-method, which is also used in the examples-section.
      • Allows offline testing of my upcoming check_server_compatibility-script
    • Also added the possibility to tag test servers with a name

Added

  • By now calendar.search(..., sort_keys=("DTSTART") will work. Sort keys expects a list or a tuple, but it's easy to send an attribute by mistake. #449

[1.4.0] - 2024-11-05

  • Lots of work lifting the project up to more modern standards and improving code, thanks to Georges Toth (github @sim0nx), Matthias Urlichs (github @smurfix) and @ArtemIsmagilov. While this shouldn't matter for existing users, it will make the library more future-proof.
  • Quite long lists of fixes, improvements and some few changes, nothing big, main focus is on ensuring compatibility with as many server implementations as possible. See below.

Fixed

  • Partial workaround for #401 - some servers require comptype in the search query -
  • At least one bugfix, possibly fixing #399 - the accept_invite-method not working - #403
  • Fix/workaround for servers sending MAILTO in uppercase - #388, #399 and #403
  • get_duration: make sure the algorithm doesn't raise an exception comparing dates with timestamps - #381
  • set_due: make sure the algorithm doesn't raise an exception comparing naive timestamps with timezone timestamps - #381
  • Code formatting / style fixes.
  • Jason Yau introduced the possibility to add arbitrary headers - but things like User-Agent would anyway always be overwritten. Now the custom logic takes precedence. Pull request #386, issue #385
  • Search method has some logic handling non-conformant servers (loading data from the server if the search response didn't include the icalendar data, ignoring trash from the Google server when it returns data without a VTODO/VEVENT/VJOURNAL component), but it was inside an if-statement and applied only if Expanded-flag was set to True. Moved the logic out of the if, so it always applies.
  • Revisited a problem that Google sometimes delivers junk when doing searches - credits to github user @zhwei in #366
  • There were some compatibility-logic loading objects if the server does not deliver icalendar data (as it's suppsoed to do according to the RFC), but only if passing the expand-flag to the search-method. Fixed that it loads regardless of weather expand is set or not. Also in #366
  • Tests - lots of work getting as much test code as possible to pass on different servers, and now testing also for Python 3.12 - ref #368 #360 #447 #369 #370 #441 https://github.com/python-caldav/caldav/pull/443a
  • The vcal fixup method was converting implicit dates into timestamps in the COMPLETED property, as it should be a timestamp according to the RFC - however, the regexp failed on explicit dates. Now it will take explicit dates too. #387
  • Code cleanups and modernizing the code - #404 #405 #406 #407 #408 #409 #412 #414 #415 #418 #419 #417 #421 #423 #430 #431 #440 #365
  • Doc - improved examples, #427
  • Purelymail sends absolute URLs, which is allowed by the RFC but was not supported by the library. Fixed in #442

Changed

  • In #366, I optimized the logic in search a bit, now all data from the server not containing a VEVENT, VTODO or VJOURNAL will be thrown away. I believe this won't cause any problems for anyone, as the server should only deliver such components, but I may be wrong.
  • Default User-Agent changed from Mozilla/5 to python-caldav/{__version__} - #392
  • Change fixup log lvl to warning and merge diff log messages into related parent log by @MrEbbinghaus in #438
  • Mandatory fields are now added if trying to save incomplete icalendar data, #447

Added

  • Allow to reverse the sorting order on search function by @twissell- in #433
  • Work on integrating typing information. Details in #358
  • Remove dependency on pytz. Details in #231 and #363
  • Use setuptools-scm / pyproject.toml (modern packaging). Details in #364 and #367
  • Debugging tool - an environment variable can be set, causing the library to spew out server communications into files under /tmp. Details in #249 and #248
  • Comaptibility matrix for posteo.de servers in tests/compatibility_issues.py
  • Added sort_reverse option to the search function to reverse the sorting order of the found objects.
  • It's now possible to specify if expand should be done on the server side or client side. Default is as before, expanding on server side, then on the client side if unexpanded data is returned. It was found that some servers does expanding, but does not add RECURRENCE-ID. #447

Security

The debug information gathering hook has been in the limbo for a long time, due to security concerns:

  • An attacker that has access to alter the environment the application is running under may cause a DoS-attack, filling up available disk space with debug logging.
  • An attacker that has access to alter the environment the application is running under, and access to read files under /tmp (files being 0600 and owned by the uid the application is running under), will be able to read the communication between the server and the client, communication that may be private and confidential.

Thinking it through three times, I'm not too concerned - if someone has access to alter the environment the process is running under and access to read files run by the uid of the application, then this someone should already be trusted and will probably have the possibility to DoS the system or gather this communication through other means.

Credits

Georges Tooth, Крылов Александр, zhwei, Stefan Ollinger, Matthias Urlichs, ArtemIsmagilov, Tobias Brox has contributed directly with commits and pull requests included in this release. Many more has contributed through reporting issues and code snippets.

Test runs

Prior to release (commit 92de2e29276d3da2dcc721cbaef8da5eb344bd11), tests have been run successfully towards:

  • radicale (internal tests)
  • xandikos (internal tests)
  • ecloud.global (NextCloud) - with flags compatibility_issues.nextcloud + ['no_delete_calendar', 'unique_calendar_ids', 'rate_limited', 'broken_expand'] and with frequent manual "empty thrashcan"-operations in webui.
  • Zimbra
  • DAViCal
  • Posteo
  • Purelymail

[1.3.9] - 2023-12-12

Some bugfixes.

Fixed

  • Some parts of the library would throw OverflowError on very weird dates/timestamps. Now those are converted to the minimum or maximum accepted date/timestamp. Credits to github user @tamarinvs19 in #327
  • DAVResponse.davclient was always set to None, now it may be set to the DAVClient instance. Credits to github user @sobolevn in #323
  • DAVResponse.davclient was always set to None, now it may be set to the DAVClient instance. Credits to github user @sobolevn in #323
  • examples/sync_examples.py, the sync token needs to be saved to the database (credits to Savvas Giannoukas)
  • Bugfixes in set_relations, credits to github user @Zocker1999NET in #335 and #333
  • Dates that are off the scale are converted to min_date and max_date (and logging en error) rather than throwing OverflowError, credits to github user @tamarinvs19 in #327
  • Completing a recurring task with a naïve or floating DTSTART would cause a runtime error
  • Tests stopped working on python 3.7 and python 3.8 for a while. This was only an issue with libraries used for the testing, and has been mended.
  • Bugfix that a 500 internal server error could cause an recursion loop, credits to github user @bchardin in #344
  • Compatibility-fix for Google calendar, credits to github user @e-katov in #344
  • Spelling, grammar and removing a useless regexp, credits to github user @scop in #337
  • Faulty icalendar code caused the code for fixing faulty icalendar code to break, credits to github user @yuwash in #347 and #350
  • Sorting on uppercase attributes didn't work, ref issue #352 - credits to github user @ArtemIsmagilov.
  • The sorting algorithm was dependent on vobject library - refactored to use icalendar library instead
  • Lots more test code on the sorting, and fixed some corner cases
  • Creating a task with a status didn't work

[1.3.8] - 2023-12-10 [YANKED]

Why do I never manage to do releases right ..

[1.3.7] - 2023-12-10 [YANKED]

I managed to tag the wrong commit

[1.3.6] - 2023-07-20

Very minor test fix

Fixed

One of the tests has been partially disabled, ref #300 , #320 and #321

[1.3.5] - 2023-07-19 [YANKED]

Seems like I've been using the wrong procedure all the time for doing pypi-releases

[1.3.4] - 2023-07-19 [YANKED]

... Github has some features that it will merge pull requests only when all tests passes ... but somehow I can't get it to work, so 1.3.4 broke the style test again ...

[1.3.3] - 2023-07-19

Summary: Some few workarounds to support yet more different calendar servers and cloud providers, some few minor enhancements needed by various contributors, and some minor bugfixes.

Added

  • Support for very big events, credits to github user @aaujon in #301
  • Custom HTTP headers was added in v1.2, but documentation and unit test is added in v1.3, credits to github user @JasonSanDiego in #306
  • More test code in #308
  • Add props parameter to search function, credits to github user @ge-lem in #315
  • Set an id field in calendar objects when populated through CalendarSet.calendars(), credits to github user @shikasta-net in #314
  • get_relatives-method, #294
  • get_dtend-method

Fixed

  • Bugfix in error handling, credits to github user @aaujon in #299
  • Various minor bugfixes in #307
  • Compatibility workaround for unknown caldav server in #303
  • Google compatibility workaround, credits to github user @flozz in #312
  • Documentation typos, credits to github user @FluxxCode in #317
  • Improved support for cloud provider gmx.de in #318

Changes

  • Refactored relation handling in set_due

[1.3.2] - 2023-07-19 [YANKED]

One extra line in CHANGELOG.md caused style tests to break. Can't have a release with broken tests. Why is it so hard for me to do releases correctly?

[1.3.1] - 2023-07-19 [YANKED]

I forgot bumping the version number from 1.3.0 to 1.3.1 prior to tagging

[1.3.0] - 2023-07-19 [YANKED]

I accidentally tagged the wrong stuff in the git repo