Allow a generic number of EKOs to be passed to pineappl evolve

[Radonirinaunimi]

This addresses #330.

Things to do:

• rename EkoNames to EkoPaths
• find a robust and stable way to parse multiple EKOs
• rely on information inside the EKOs to derive the types of convolutions
• add minor tutorial on the use of p,f,pf,fpmarkings in multi-convolution

[cschwan]

I'd make this a bit simpler by relying more on clap: replace EkoPaths with Vec<EkoPath> and make EkoPath a struct of a String and a ConvType. Furthermore, use #[arg(value_delimiter = ',', num_args = 1)]. Then clap does most of the parsing.

[Radonirinaunimi]

I'd make this a bit simpler by relying more on clap: replace EkoPaths with Vec<EkoPath> and make EkoPath a struct of a String and a ConvType. Furthermore, use #[arg(value_delimiter = ',', num_args = 1)]. Then clap does most of the parsing.

So you want to remove EKOs as a positional argument? Because I don't think it would work otherwise.

[cschwan]

No, it should remain a positional parameter. It's exactly one argument, but internally clap fills the Vec by splitting the argument by ,.

[cschwan]

On second thought, the information in the +p, +f and +pf syntax should be encoded in EKO's yaml files. We should rely on them instead of the information provided by the user.

[Radonirinaunimi]

No, it should remain a positional parameter. It's exactly one argument, but internally clap fills the Vec by splitting the argument by ,.

But this, I think, doesn't work. #[arg(value_delimiter = ',')] is functionally equivalent to .num_args(1..) so it can only be the last positional argument and no positional arguments can come after.

The alternative would be to use named optional argument #[arg(short, long, value_delimiter = ',')], but I am not sure we also want this.

On second thought, the information in the +p, +f and +pf syntax should be encoded in EKO's yaml files. We should rely on them instead of the information provided by the user.

Yes, I will double check this to be certain, but if that's the case then that should definitely be the way to do it (added in the TODO of the description).

[felixhekhorn]

On second thought, the information in the +p, +f and +pf syntax should be encoded in EKO's yaml files. We should rely on them instead of the information provided by the user.

they are of course

just to say: if only we had someone work on NNPDF/eko#97

[Radonirinaunimi]

With respect to how the EKOs are parsed, our only option is to pass everything a one big STRING. As mentioned above having something like Vec<String> is not working.

Also, in the end, I didn't add a documentation on how to use pineappl evolve as this might not be used by many people. As currently present in the documentation, that'd be referred in the extra-subcommands.

And with these, this PR is ready.

[cschwan]

But this, I think, doesn't work. #[arg(value_delimiter = ',')] is functionally equivalent to .num_args(1..) so it can only be the last positional argument and no positional arguments can come after.

Yes, I didn't realize this sooner, had to see it for myself. Thank for trying it out!