I can jump in on this question, as these changes are only relevant for our Cloud products and not Server you’ll have total control of your app to update all of your customers at once as your apps will automatically update when you update it.
Hi @rwhitbeck, I have some users that are still on older versions of my plugin (I think this was where the permissions changed) and they aren’t updated (I assume they have to accept the new permissions). What happens if I need to change the user id or key passed into screens through the atlassian-connect.json
Thanks
Paul
@paul - are these users on private versions of your cloud apps?
@akassab they aren’t, I am the only one that uses the private versions
Thanks
Paul
Hi @ElementML,
It should be fairly straight forward. Just update the username in question to accountId. Like this:
project = “yourproject” and assignee=pweinberger ORDER BY priority DESC
to
project = “yourproject” and assignee=8ad65abc53def45456dead3 ORDER BY priority DESC
Could you please post an example jql and I can take a look at it?
Is there an API we can submit legacy JQL to and have it converted into the accountId version? We have many stored JQL strings that need updating. Unless Atlassian provide something, every vendor is going to have to implement the same thing.
Also we have the workflow post function problem. Data stored in postfunctions contains userKeys which need migrating to accountId, but we can’t do it.
@jmort, I have the postfunction problem as well, haven’t managed to get an answer on what to do yet though
Hi @pweinberger,
It should be fairly straight forward. Just update the username in question to accountId. Like this:
project = “yourproject” and assignee=pweinberger ORDER BY priority DESC
to
project = “yourproject” and assignee=8ad65abc53def45456dead3 ORDER BY priority DESC
Could you please post an example jql and I can take a look at it?
Thanks for the reply. I do understand how the queries need to change in order to work with the accountId, my question was more along the lines of whether Atlassian would be providing a more automated migration path for actually making the changes in question.
If not, every Marketplace vendor that stores user-supplied JQL is going to have to implement their own solution for parsing the queries, identifying the fields that contain usernames or keys (including custom user picker fields, which are of course different for each Jira tenant), looking up the accountId for each parsed username, validating the new query, etc. This is a significant development effort, and it seems like a huge waste of resources for the ecosystem if everyone has to roll their own code, particularly given that Atlassian obviously has a method for migrating the queries automatically. What I was really hoping for was an endpoint that we could hit with a valid JQL query and receive a valid, migrated query in the response.
(EDIT: Looks like @jmort beat me to it.)
Cheers,
Mark
Hi @rwhitbeck,
I can jump in on this question, as these changes are only relevant for our Cloud products and not Server you’ll have total control of your app to update all of your customers at once as your apps will automatically update when you update it.
This isn’t completely true. See this page from the Marketplace docs for the relevant details on manual upgrades of Atlassian Connect apps. Users can have old versions installed forever if there are changes to required data scopes in the app descriptor.
Cheers,
Mark
Such a service is in our roadmap and we plan to do it as quickly as possible.
Repeating my answer to @jmort here.
Such a service is in our roadmap and we plan to do it as quickly as possible.
Thanks @pweinberger, that’s great news – appreciate the clarification.
Hi @jbevan
No, you are correct. These controls haven’t been made public yet. I don’t have an exact date for when they will be made public, but we’re working on it.
Thanks for all the extra info, its much appreciated.
I don’t doubt that Atlassian are working on it, but deprecating things without providing alternatives or the ability to test the relevant changes is particularly frustrating for vendors. I don’t mind how long it takes for the privacy controls and API changes to be released, but I do mind if those changes aren’t completed with enough time for vendors to react to those changes before the deprecation deadline (29th March) that has already been set.
Additionally, we’ve still not had any confirmation or clarification on how we’re supposed to migrate post-function configuration that contain userkeys?
Hi,
Migration guide specifies that only GET methods for search endpoints are impacted. Is that correct?
Will the POST methods accept username and userkey in JQL queries?
- /api/3/search
- /api/3/issue/picker
Methods
- GET
Request query parameter changes
- JQL will not allow
username
oruserKey
in queries. UseaccountId
instead.
Hi,
As per the connect API migration guide here: https://developer.atlassian.com/cloud/jira/platform/connect-api-migration/, we are testing our add-on with the “apiMigrations”: {
“gdpr”: true
}
The JWT seems to honour this flag, but the Rest APIs are still returning the old responses with username and userkey included. We have also tried setting the visibility to private (Screenshot), but it does not change the API response.
The avatarUrls field will not be removed. The migration guide examples are being updated to reflect that. The only change is that the links in the avatarUrls field will be different, but shouldn’t affect any logic for third party apps.
@pweinberger When we opt into the GDPR apiMigration, does this apply only to API calls on the Connect server, or does it also apply to API calls made from the browser?
I’m also not clear on what endpoints are affected by opting into to the migration, and whether it applies to v2 API responses?
As far as I understand things, both v3 and v2 of the REST APIs are affected by GDPR changes, regardless of how those APIs are accessed (via AP in the browser/using curl/from a connect app server etc).
The primary change between v2 and v3 of the REST API is that v3 supports the new Atlassian Document Format. Please anyone correct me if I’ve misunderstood.
These GDPR privacy changes, forces us to do our own mapping table of users, when reading project permissions/roles between platforms on the REST API.
Today I use email-prefix(before@) -> username lookup, hence they map 1:1. Easy Peasy! Works without ANY manual intervention!
How can we now do this after April 1st? - We will be forced to maintain a mapping of the unique accountID-> Internal domain account from admin perspective.
Any way of extracting a full list of users/accountIDs on the API or extract per user for usage towards other platforms/integrations.
Currently our integration/functionality is bound to broken by March 29st unless there is a fully automatic alternative to mapping tables.
And sorry to say…Atlassian’s Access AD integration is still ridiculously expensive! (and probably does not read out the AccountId anyways for usage from a central AD perspective)
/Morten