The change introduced in #2150 introduces a bug where a Team would be re-created every time a user logs in, as the check if a team already exists was based on both the unique `oidcID` and the `name`. This PR proposes to only base the check on the ID, as this should be unique.
Co-authored-by: Daniel Herrmann <daniel.herrmann1@gmail.com>
Reviewed-on: https://kolaente.dev/vikunja/vikunja/pulls/2152
Reviewed-by: konrad <k@knt.li>
Co-authored-by: waza-ari <daniel.herrmann@makerspace-darmstadt.de>
Co-committed-by: waza-ari <daniel.herrmann@makerspace-darmstadt.de>
This PR stems from issue #2150, in turn following up on PR #1393
It adds additional details around the OIDC authentication feature, as well as details about how the team assignment works.
Co-authored-by: Daniel Herrmann <daniel.herrmann1@gmail.com>
Reviewed-on: https://kolaente.dev/vikunja/vikunja/pulls/2151
Reviewed-by: konrad <k@knt.li>
Co-authored-by: waza-ari <daniel.herrmann@makerspace-darmstadt.de>
Co-committed-by: waza-ari <daniel.herrmann@makerspace-darmstadt.de>
Testing this locally resulted in improved response times from ~50ms to ~20ms when creating a project. It looks like even though the code running these sql queries uses different go routines, they affect each other (caused by IO or context switching?)
This change ensures already loaded projects are passed down when fetching their subscription instead of re-loading each project with a single sql statement. When loading all projects, this meant all projects were loaded twice, which was highly inefficient. This roughly added 25ms to each request, assuming the per page limit was maxed out at 50 projects.
Empirical testing shows this change reduces load times by ~20ms. Because the request is already pretty fast, this is ~30% of the overall request time, making the loading of projects now even faster
Since Vikunja's docker image is now based on the scratch image, the root ca certificates are not included by default anymore. This meant Vikunja could not check if the certificate presented by a remote host was valid, thus failing the connection. This meant it was impossible to use features communicating with external hosts such as webhooks, openid auth or gravatar.
Before this change, applying a filter and then scrolling a bucket would not use that filter when lazy loading the tasks in that bucket. That resulted in all tasks being loaded, regardless if the filter applied to them.
