This repository has been archived by the owner on Nov 7, 2020. It is now read-only.
Test inconsistency #267
Comments
|
I've worked around this problem by splitting the test suite in two during runs (see 9400319 ). Leaving this open as I'd prefer to fix it without resorting to hacks like that. |
yourcelf
added a commit
that referenced
this issue
Apr 4, 2014
yourcelf
added a commit
that referenced
this issue
Apr 9, 2014
yourcelf
added a commit
that referenced
this issue
Apr 9, 2014
yourcelf
added a commit
that referenced
this issue
Apr 9, 2014
yourcelf
added a commit
that referenced
this issue
Apr 11, 2014
yourcelf
added a commit
that referenced
this issue
Apr 15, 2014
yourcelf
added a commit
that referenced
this issue
Apr 15, 2014
yourcelf
added a commit
that referenced
this issue
Apr 17, 2014
|
After a long chain of fixes, I think most of the myriad timing issues have been identified and fixed. At this point, tests run consistently enough that I think we can regard any new inconsistencies as separate issues, and this catch-all can be retired. |
|
Reopening due to failures in #340. |
yourcelf
added a commit
that referenced
this issue
Sep 13, 2014
yourcelf
added a commit
that referenced
this issue
Sep 27, 2014
throttleSync previously used a key of "roomId" + "type" to determine whether a message was the same type of message, and could therefore be dropped. However, since the message transport refactor last April, all the "types" are simply "state", with paths, operators, and other crucial details in the args. Change this method to use a caller-defined key instead. Sessions were not being persisted server-side on changes to connectedParticipants due to a missing return value. Superagent requests seems to interact poorly with sinon fake timers -- in particular, clock restoration in the body of a request.end callback seems to not work as expected, leading to at least some of the inconsistency contributing to #267. Symptom is that "throttleSync" was triggering functions, but never executing (the setTimeout was never executing). Fix this by re-establishing global timers manually.
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
test/test.session-joining.selenium.jsseems to fail fairly often, but inconsistently. It reliably passes when executed individually, but occasionally when executed with the whole test suite, it fails about half the time like these:https://travis-ci.org/drewww/unhangout/builds/20160219
https://travis-ci.org/drewww/unhangout/builds/20159670
https://travis-ci.org/drewww/unhangout/builds/20146641
https://travis-ci.org/drewww/unhangout/builds/20095832
https://travis-ci.org/drewww/unhangout/builds/20095636
This seems to be either some sort of global state issue, some sort of timing issue (which only manifests in the whole-test-suite context), or some sort of socket/open files type issue. Not sure what's going on.
The text was updated successfully, but these errors were encountered: