Re: Tests and randomness

On Thu, 19 Dec 2019, Frank Gevaerts via curl-library wrote:

> Is there any way to keep state between runs? If so, and assuming test
> failures tend to be fixed reasonably quicly, maybe only change the seed if
> there are no failures? You'd probably still not want to change it at every
> run even when no tests fail (test success can be as intermittent as test
> failure), so maybe change it weekly or monthly whenever there are no
> failures?

We run tests in a large number of different places and in different setups and
I don't think we can assume that we can save anything locally between test
runs. For example, most of the CI builds start over on completely clean
machines every time.

My current system with "year + month + md5(uname)" is the best approach I've
come up with in the absense of stored state.

-- 
  / daniel.haxx.se | Get the best commercial curl support there is - from me
                   | Private help, bug fixes, support, ports, new features
                   | https://www.wolfssl.com/contact/
-------------------------------------------------------------------
Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
Etiquette:   https://curl.haxx.se/mail/etiquette.html