a few thoughts;

Using containers adds a dep (already mentioned) and subdividing test
runs across containers will require a mechanism for aggregating test
result reports which may get a bit complicated. Maybe container arch
could just be restricted to travis-ci and those who are ok with
containers (I am but wary of unintended architectural 'seepage' with
their usage).

As a poor mans optimisation, runtests.pl could be amended to fork
tests ... for completeness, I see that ctest can execute tests in
parallel though I have no experience with it.

It would be cool to see Dan's proposed changes as a branch.

Jim

On Tue, 2 Oct 2018 at 09:54, Daniel Stenberg via curl-library
<curl-library_at_cool.haxx.se> wrote:
>
> On Mon, 1 Oct 2018, Kamil Dudka wrote:
>
> > If you do, please keep the possibility to run at least some basic sequential
> > test-suite in environments where containers are not available. I think such
> > restriction is currently common for many build services.
>
> Agreed. In many cases it will make sense to run them serially. I also imagine
> that when we want to debug a single test failure for example, we rather just
> be able to run them like today with -g etc.
>
> --
>
> / daniel.haxx.se
> -------------------------------------------------------------------
> Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
> Etiquette: https://curl.haxx.se/mail/etiquette.html
-------------------------------------------------------------------
Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
Etiquette: https://curl.haxx.se/mail/etiquette.html
Received on 2018-10-02