Re: Fwd: Get rid of uniChr

Robin Becker <[email protected]> Fri, 31 Dec 2021 08:41:23 +0000

Newsgroups	gmane.comp.python.reportlab.user
Message-ID	<[email protected]>

On 30/12/2021 17:11, [email protected] wrote:
.........
> 
> You should probably make the tests repeatable by using a fixed seed for the random generators. If there are additional  
> truly random tests, they should be clearly marked as such, and output the input data they used so that possible failures 
> can be reproduced.
> 
> -- Christoph
........
I think that's the answer. For the simple tests I don't think we set up rl_invariant; for the comparison tests where we 
look at images derived from the pdf outputs we always have to use

export RL_invariant=1

in the script which forces all/most of the random sequences to be the same for each run.
-- 
Robin Becker