test to ensure spurious state doesn't leak into incr sync responses when LLing #487
Conversation
…hen LLing tests matrix-org/synapse#3760 regression test for element-hq/element-web#7223
|
@matrixbot retest this please |
tests/10apidoc/09synced.pl
Outdated
| @@ -205,7 +205,34 @@ sub await_sync_presence_contains { | |||
| } | |||
|
|
|||
|
|
|||
| push @EXPORT, qw( assert_room_members assert_state_room_members_matches ); | |||
| push @EXPORT, qw( assert_room_members assert_state_room_members_matches assert_room_state); | |||
There was a problem hiding this comment.
nice to have: generally it seems helpful to have a push @EXPORT next to each func being exported, rather than one big one at the top
There was a problem hiding this comment.
the codebase is very inconsistent on this, but have moved them.
tests/10apidoc/09synced.pl
Outdated
| push @EXPORT, qw( assert_room_members assert_state_room_members_matches ); | ||
| push @EXPORT, qw( assert_room_members assert_state_room_members_matches assert_room_state); | ||
|
|
||
| # assert that the state body of a sync response is made up of the given state types. |
There was a problem hiding this comment.
nice to have: generally we seem to be using POD for this sort of thing. It'd be nice to be consistent, even if it's slightly more verbose.
tests/10apidoc/09synced.pl
Outdated
|
|
||
| # assert that the state body of a sync response is made up of the given state types. | ||
| sub assert_room_state { | ||
| my ( $body, $room_id, $state_types ) = @_; |
There was a problem hiding this comment.
some docs on what state_types should look like would be helpful, to save people reverse-engineering it.
tests/10apidoc/09synced.pl
Outdated
| push @EXPORT, qw( assert_room_members assert_state_room_members_matches assert_room_state); | ||
|
|
||
| # assert that the state body of a sync response is made up of the given state types. | ||
| sub assert_room_state { |
There was a problem hiding this comment.
this should probably be assert_room_state_types_match or something, because (a) it only checks for the types, not the whole state, and (b) it checks that they match exactly (as opposed to, for example, being a superset of the desired list)
There was a problem hiding this comment.
Also, should it be assert_state_types_match and take a state block, ie pull lines 214-215 up to the caller, making this equivalent to assert_state_room_members_matches rather than assert_room_members? That feels like it would be more intuitive to me.
There was a problem hiding this comment.
unconvinced about passing in a random list of state events rather than a sync body response, but whatever; have done so
tests/31sync/15lazy-members.pl
Outdated
| @@ -258,7 +267,6 @@ | |||
| }); | |||
| }; | |||
|
|
|||
|
|
|||
|
@richvdh ptal |
tests/10apidoc/09synced.pl
Outdated
|
|
||
| log_if_fail "Room", $room; | ||
| $state_types is an arrayref of arrayrefs, each a tuple of type & state_key. |
There was a problem hiding this comment.
an example might have been helpful.
|
retest this please |
tests matrix-org/synapse#3760
regression test for element-hq/element-web#7223