Conversation
|
sorry for the force pushes 😄 I haven't rebased in a while |
|
Oops, I left an extra field in the test table. Let me fix that |
this moves the unmarshaling of a remote actor out into a new helper which accounts for the possibility of a context being a list or a single entity. i.e. a string or an object. basics tests are provided for both situations also go fmt'd the file activitypub.go
|
Thanks! I think the code looks good overall, just need to test. Was this route better than changing the type of Also, @mrvdb could you take a look at this? |
|
I though so at first, after going with this instead I feel they would both be similar. Either way we need to check for a string or slice and convert any string contexts into a slice. Due to the append behaviour when adding a private key in the core library. So this could easily be moved there instead if preferred. |
|
Has this been this tested by someone already or is this a 'theoretical fix' at this time? (perhaps a good time to discuss at some point what is expected in terms of integration tests from contributors?) @thebaer I will have a look indeed. |
|
@mrvdb I have not had a chance to manually test the integration. But I agree this is a good time to start talking about integration tests in general. There would ideally be a suite that runs on pull requests to test automatically, as well as being able to run locally while developing. |
|
TravisCI does run on each commit on a pull request in these repos, to make sure the code's not broken. But that's not necessarily the same thing as "the code does what I want it to". ;) |
this moves the unmarshaling of a remote actor out into a new helper which accounts for the possibility of a context being a list or a single entity. i.e. a string or an object. basics tests are provided for both situations also go fmt'd the file activitypub.go
|
@robjloranger By the way, the best way to remove files that get accidentally committed like that would be with a new commit, instead of a force-push. Force-pushing is okay for single-user repos, but can mess things up when other people have pulled your branch. Another helpful hint, if you didn't know about it: using |
|
@mrvdb could you test by trying to follow I will try to set one up today and test as well. |
|
@robjloranger Thanks, I will certainly do so, but I'll only have the ability to do it in about a week or so. (I'm hiking next week) |
|
ok no trouble, I'll set up a local instance and try it out |
|
if anyone familiar with pubgate and writefreely could help, I can't figure out how to follow a writefreely user from pubgate. I assume it is an outbox follow on the pubgate side, but pointed at what endpoint on the writefreely side for a given user I do not know. |
|
Following The follow action works and the endpoint is correctly entered into the following list on my end. If you can start writing some posts on that endpoint I will monitor if the follow does actually work. ( @robjloranger I find it easiest to use https://www.getpostman.com/ and use the pubgate API that way) |
|
@mrvdb OK wrote two test posts. I see both from mast still. And can unfollow and follow from there as well still. |
|
I don't see any connection coming in, can you post again? (I've upped logging so I see more) |
|
All set |
|
Nothing on the pubgate side, my follow on mastodon gets the post. I think the error is on my side. |
|
Not sure what is going on. I get objects from my other subscriptions, so pretty sure the setup is ok now. I've unfollowed and re-followed, could you post again? |
|
This issue appears fixed, but we have a new interoperability issue: #116 |
|
all is working well now with pubgate @thebaer |
this moves the unmarshaling of a remote actor out into a new helper which
accounts for the possibility of a context being a list or a single entity.
i.e. a string or an object.
basics tests are provided for both situations
also go fmt'd the file activitypub.go
during the course of working on this I found a missing string format variable in
exports.go, in fixing that I also go fmt'd the file.Fixes #100
This also fixes #116 by using the followers Inbox if SharedInbox is not set.