Repository navigation
Replies: 2 comments 9 replies
|
The reason is external to git-bug: to push an event to an external bug tracker, you need credentials for that specific user. For example, Github will not accept you pushing a comment for me. However, I'd like to correct something: git-bug doesn't restrict pushing to the active user, it restricts pushing to what you have credentials for. This means that you can push every event with git-bug, as long as you have the credentials for it. It may be that the target bug tracker accept some kind of all-powerful admin credential, and in that case we would just need some logic for git-bug to understand that fact. Another option would be some kind of bridging Saas where every user grant permission and would do the transfer on their behalf. |
|
Actually, looking at those bugs for M2Crypto, I have actually forced all those tickets to be submitted by myself (with the original author in comments). That would be probably a solution. See for example https://todo.sr.ht/~mcepl/m2crypto/303 |
|
That's a solution, but it's not the normal behavior of a git-bug bridge. If you implement that, it should be an opt-in behavior IMHO, a flag or something. |
|
Absolutely, and I am not sure how to make it not to corrupt local database and/or other bridges. |
|
@MichaelMure @sudoforge What do you think about #1501 ? |
|
Thanks for the explanation that clears it up. So the limitation isn’t in git-bug itself but in the fact that external trackers require per-user credentials and won’t let you push events on behalf of other authors. That makes sense. It might help to clarify this in the documentation, since the current description of “full synchronization” suggests projects can be fully migrated, which isn’t possible unless the target tracker supports admin impersonation or similar privileges. If impersonation APIs are available on some platforms, supporting them in bridges could make full migrations possible. Happy to help test if that moves forward. |
|
Would you be willing to review the plan in #1501 or would you be more comfortable if I ask Gemini to generate the code and review that? |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
While developing bridge for https://todo.sr.ht (#1499), we came to the seeming limitation of the current
git-bugarchitecture: the bridge documentation promises “bi-directional gateways” to “move to another platform” and “synchronise changes”, but all bridge implementations (GitHub,GitLab,Jira,TodoSRHT) only export bugs where the user is an actor.$ git bug bug|wc -l 434 $and
and really I have to create one bug myself to have https://todo.sr.ht/~mcepl/git-test/1
This means:
TodoSRHTAll exporters use
HasAnyActor()check that ignores bugs with "not an actor" message, contradicting the documented behaviour of full synchronisation.Expected: Export all imported bugs (with origin metadata) regardless of author, only restrict locally-created bugs.
Actual: Only export bugs where user is author/commenter.
This makes bridges nearly useless for one obvious use case for
git-bug: migrating between issue tracking systems.Is this intentional design, a bug that should be fixed, or (most likely) our misunderstanding of
git-bugcode?All reactions