Show errors on console under plugin development #1302
Conversation
Raising errors doesn't help situation because: * errors raised in threads will be raised in main thread by abort_on_exception=true * main thread of test-unit will rescues and ignore errors So we can't find any errors in unit tests without showing these on console.
tagomoris
force-pushed
the
show-errors-on-console-under-plugin-development
branch
from
fd8d86f
to
34dc637
Compare
|
@repeatedly could you review this change? |
| @@ -66,6 +66,12 @@ def thread_create(title) | |||
| yield | |||
| thread_exit = true | |||
| rescue Exception => e | |||
| if @under_plugin_development | |||
| STDERR.puts "\nError raised in thread from #thread_create, #{e.class}:#{e.message}" | |||
There was a problem hiding this comment.
Head \n is needed?
Previous log doesn't have \n?
There was a problem hiding this comment.
Yes, this code expects that leading text is '.' (means progress of test cases) from test runner, without tailing newline.
| if state.thread && state.thread.status # "run"/"sleep"/"aborting" or false(successfully stop) or nil(killed by exception) | ||
| state.thread.run | ||
| else | ||
| log.warn "thread is already dead" |
There was a problem hiding this comment.
If one thread is dead, we should replace dead thread with new active one?
There was a problem hiding this comment.
I don't think so. Dead flush thread means any kind of bug of output.rb.
There was a problem hiding this comment.
I see.
Off topic. Collecting live and dead threads information is good for debug.
If one thread is dead, it means other thread also becomes dead.
If this situation continues, all output threads are dead.
We should display the message when all output threads are dead.
|
Commented. Changes are LGTM. BTW, I first knew out_stdout supported async write ;p |
|
OK, let me merge this later. |
This change fixes code to show error/fatal logs in testing console.
Raising errors doesn't help situation because:
So we can't find any errors in unit tests without showing these on console.
I found that some
#try_writemethod call#commit_writein it, and it misses to remove the chunk from@dequeued_chunks. It's a kind of bug.Adding a chunk into
@dequeued_chunksafter calling#try_writehas thread-unsafe bug. (because plugin may commit the chunk in another threads just after#try_writecalled, and plugin's main thread may have NOT added that chunk into@dequeued_chunks.)So, I fixed it at the same time.