Re: Outreachy - Guix Data Service: implementing basic json output for derivation comparison page

[Christopher Baines]

Luciana Lima Brito <lubrito@posteo.net> writes:

> Hi,
>
> On Sun, 18 Apr 2021 17:34:13 +0100
> Christopher Baines <mail@cbaines.net> wrote:
>
>> I think it might be good to do something, just to narrow the scope.
>> The outputs binding is valid for the whole let*, and all the code in
>> it, but is only used on three lines, in one single place. Maybe there
>> could be a let there that just defines outputs (maybe named
>> output-groups so you can use the outputs binding for the overall
>> thing).
>
> I did it.

Great :)

>> That's a good question, I'd look at the database schema, assuming the
>> type of the field is a boolean, the question is whether the field is
>> nullable?
>
> I looked on the database schema, and the "recursive" field is not
> nullable, and it is a boolean, so the test I'm doing is working for
> this.

Cool.

>> Hmm, I'm not sure why that is on the HTML page, but I'd generally try
>> and keep most bits in the JSON, since it's not as helpful to omit bits
>> if they're not that important.
>
> I added the "common" field for inputs.

Ok.

>> One other thing I noticed is that the alist for the args is being
>> picked apart then reconstructed. Like for the inputs, outputs and
>> sources, I'd map over the arguments alist and transform it to the way
>> you want it to be.
>
> This part was a bit more complicated for me to understand. You mean I
> should build a function similar to outputs, inputs and sources to map
> the arguments, wouldn't it be a lot just to show a vector?

As an example, if you have this structure.

  '((foo . 1)
    (bar . 2))

and you want the numbers to be strings, you can do something like:

  (map (match-lambda
        ((name . number)
         (cons name (number->string number))))
       data)

and I think a similar approach for the transformation you're trying to
perform to the arguments will more cleanly represent what you're trying
to do.

One different thing I noticed:

  ,@(if (not (string? hash-alg))
        '()
        `((hash-algorithm . ,hash-alg)))

I'd suggest simplifying this by flipping the different parts of the if,
and removing the not.

signature.asc
Description: PGP signature