- Oct 20, 2015
-
-
Erik Johnston authored
-
Daniel Wagner-Hall authored
This is done by using the same /leave flow as you would use if you had already accepted the invite and wanted to leave.
-
- Oct 19, 2015
-
-
Paul "LeoNerd" Evans authored
-
Paul "LeoNerd" Evans authored
-
Paul "LeoNerd" Evans authored
-
Paul "LeoNerd" Evans authored
Another signing test vector using an 'm.room.message' with content, so that the implementation will have to redact it
-
Paul "LeoNerd" Evans authored
-
- Oct 13, 2015
-
-
Mark Haines authored
-
Erik Johnston authored
-
Erik Johnston authored
-
- Oct 12, 2015
-
-
Mark Haines authored
Apply the filter the 'timeline' and 'ephemeral' keys of rooms. Apply the filter to the 'presence' key of a sync response.
-
- Sep 23, 2015
-
-
Mark Haines authored
-
- Sep 22, 2015
-
-
Mark Haines authored
-
Daniel Wagner-Hall authored
SYN-287 This requires that HS owners either opt in or out of stats reporting. When --generate-config is passed, --report-stats must be specified If an already-generated config is used, and doesn't have the report_stats key, it is requested to be set.
-
- Sep 10, 2015
-
-
Mark Haines authored
-
- Sep 09, 2015
-
-
Mark Haines authored
-
- Sep 01, 2015
-
-
Erik Johnston authored
-
Erik Johnston authored
-
Daniel Wagner-Hall authored
-
- Aug 26, 2015
-
-
Daniel Wagner-Hall authored
This reduces our ~8 second sequential test time down to ~7 seconds
-
Daniel Wagner-Hall authored
-
Daniel Wagner-Hall authored
A couple of weird caveats: * If we can't validate your macaroon, we fall back to checking that your access token is in the DB, and ignoring the failure * Even if we can validate your macaroon, we still have to hit the DB to get the access token ID, which we pretend is a device ID all over the codebase. This mostly adds the interesting code, and points out the two pieces we need to delete (and necessary conditions) in order to fix the above caveats.
-
- Aug 25, 2015
-
-
Daniel Wagner-Hall authored
-
Daniel Wagner-Hall authored
Removes device_id and ClientInfo device_id is never actually written, and the matrix.org DB has no non-null entries for it. Right now, it's just cluttering up code. This doesn't remove the columns from the database, because that's fiddly.
-
- Aug 20, 2015
-
-
Daniel Wagner-Hall authored
This allows refresh tokens to be exchanged for (access_token, refresh_token). It also starts issuing them on login, though no clients currently interpret them.
-
Daniel Wagner-Hall authored
We're about to have two kinds of token, access and refresh
-
Daniel Wagner-Hall authored
I prefer the auth handler to worry about all auth, and register to call into it as needed, than to smatter auth logic between the two.
-
- Aug 19, 2015
-
-
Daniel Wagner-Hall authored
-
- Aug 18, 2015
-
-
Daniel Wagner-Hall authored
-
Daniel Wagner-Hall authored
I made the non-test seconds instead of ms, but not the test
-
Daniel Wagner-Hall authored
-
Daniel Wagner-Hall authored
This just replaces random bytes with macaroons. The macaroons are not inspected by the client or server. In particular, they claim to have an expiry time, but nothing verifies that they have not expired. Follow-up commits will actually enforce the expiration, and allow for token refresh. See https://bit.ly/matrix-auth for more information
-
- Aug 11, 2015
-
-
Erik Johnston authored
-
Daniel Wagner-Hall authored
-
Daniel Wagner-Hall authored
-
- Aug 07, 2015
-
-
Erik Johnston authored
-
Erik Johnston authored
-
- Aug 06, 2015
-
-
Erik Johnston authored
-
- Aug 05, 2015
-
-
Erik Johnston authored
-
- Aug 04, 2015
-
-
Erik Johnston authored
-