Subject: Re: [boost] Travis/Appveyor improvements for the superproject
From: John Maddock (jz.maddock_at_[hidden])
Date: 2017-09-20 16:20:09
On 20/09/2017 15:45, Peter Dimov via Boost wrote:
> Ion Gaztañaga wrote:
>> Shouldn't the hole boost repo be tested when a library changes?
> This isn't feasible as CI jobs have a time limit of 49 minutes on
> Travis, which is sometimes insufficient for testing even a single
I think this is the problem with this approach: any non-trivial library
will quickly run out of time on the CI servers, some of which are rather
slow. Certainly Math and Multiprecision libraries had to be split into
multiple jobs to get through CI testing. I'm sure there are many others
with similar issues.
--- This email has been checked for viruses by AVG. http://www.avg.com
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk