Code review comment for lp:~maddevelopers/mg5amcnlo/COLLIER

Hi Olivier,

On Thu, May 26, 2016 at 2:03 AM, Olivier Mattelaer <
<email address hidden>> wrote:

> Hi Valentin,
>
> This is so cool.
> I did not have the time to check yet but this is some first comment:
>
> 1) I would refrain to include this in 2.4.1, mainly because I want to have
> that version as bug fix only and be released very very soon (in max two
> weeks).
>

I understand. But I think that this implementation is fairly robust as it
doesn't modify anything else except MadLoop's output (to the exception of
the 'unique_id' additional option to generate_subrpocess_v4).
But still, I understand if you would rather delay it, just for safety; I'm
ok with this.

> 2) I do not understand the reason "UseQPIntegrandForNinja False -> True".
> You said "especially since we now have COLLIER first." This is actually not
> really True 99% of the user will not install collier and your option should
> reflect that scenario (or you update your option according to the tool
> available which can be really interesting.
>

I will have COLLIER installed by default from a local copy that can then be
updated at any point by the user from the online tarball using the command
'install collier'. (I'll ask D&D for their consent on this first, event
though the tool is public, just to be politically correct). Basically, the
same strategy as I used for Ninja.
And COLLIER has *no dependence*, is in plain fortran90 and only needs a
minute or so to install. So yes, everybody will have it :).

Regarding the order, I think I want to still keep Ninja first, then
followed by COLLIER and finally CutTools (though it should almost never get
there anymore).

Cheers,

>
> Cheers,
>
> Olivier
>
>
> --
> https://code.launchpad.net/~maddevelopers/mg5amcnlo/COLLIER/+merge/295787
> Your team MadDevelopers is subscribed to branch
> lp:~maddevelopers/mg5amcnlo/2.4.1.
>

--
Valentin