remove event subscription sleeps on linux

[sergirubio]

I'm currently testing these changes using HDB++ and Taurus GUI's.
I have not detected any case yet in which removing the sleeps has caused subscription to fail.
Please, try it on your test suites to see if it has any unwanted consequence.

[bourtemb]

pipe_event automated test is failing when we remove this sleep:

pipe_event: /home/tango/src/cpp_test_suite/event/pipe_event.cpp:238: int main(int, char**): Assertion `cb.cb_executed == 2' failed.

[sergirubio]

Hi Reynald, I have modifed the sleeps to 5ms and 1ms (compared to 20-10 in original sources).

Could you crosscheck if now tests are passed?

[bourtemb]

@sergirubio You can easily see it by clicking on the red cross (or green tick) on the right of your commit...
or at the end of this PR page.
With your latest commit, the Travis CI tests passed successfully! :-)
appveyor is not yet completely configured so that's why it is reporting a failure. This will change when Sébastien Gara's PR will be merged.

[sergirubio]

Ok, once we detected that pipes need them, I'm trying to found the "lower limit" for the sleeps

[sergirubio]

As the last commit failed some checks, we decided to use the previous and build our own Tango packages with it:

18802e5

It's been deployed and tested with most of our clients and it works ok

[sergirubio]

AppVeyor has passed the tests for windows ... is it good or bad?

[bourtemb]

Appveyor is currently "only" building the libraries and creating some windows installers.
It is not running any test yet... So to answer to your question, it is good the appveyor check passed but it does not mean the sleep values you used on Windows are good. It just means, it managed to compile successfully on Windows.