Opened 15 years ago
Closed 15 years ago
#2857 defect closed fixed (fixed)
test_stdio doesn't correctly pass the reactor to subprocesses
Reported by: | therve | Owned by: | |
---|---|---|---|
Priority: | highest | Milestone: | |
Component: | core | Keywords: | |
Cc: | Branch: |
test-stdio-reactor-2857
branch-diff, diff-cov, branch-cov, buildbot |
|
Author: |
Description
It uses reactor.__class__.__module__
, which is at least false for the glib2reactor. Of course, the problem is that we don't actually have access to the selected reactor for now in trial, so we probably need to put the config somewhere.
Attachments (1)
Change History (8)
Changed 15 years ago by
comment:1 Changed 15 years ago by
comment:2 Changed 15 years ago by
Branch: | → test-stdio-reactor-2857 |
---|---|
Keywords: | review added |
Owner: | therve deleted |
Priority: | high → highest |
This is ready to review in test-stdio-reactor-2857. No great, but it does the job for fixing the fedora32 slave.
comment:3 Changed 15 years ago by
Keywords: | review removed |
---|---|
Owner: | set to therve |
I'll tryo something else.
comment:4 Changed 15 years ago by
Keywords: | review added |
---|---|
Owner: | therve deleted |
OK it's better, fedora32 slave is totally green.
comment:5 Changed 15 years ago by
Keywords: | review removed |
---|---|
Owner: | set to therve |
Beautiful, please merge.
comment:7 Changed 11 years ago by
Owner: | therve deleted |
---|
Note: See
TracTickets for help on using
tickets.
Mmm. Global variables. :P
I just want to point out that the strategy of selecting a reactor using the trial command line and then having that reactor globally shared by all the tests, as a manner of testing that reactor, is and always has been bogus. We may want to make the trial reactor selection available to unit tests, but we should also be thinking about writing real unit tests for the reactors.