Re: Outreachy - Guix Data Service: questions about improving the data for derivation comparisons.

[Christopher Baines]

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

> On Mon, 26 Apr 2021 22:21:50 +0100
> Christopher Baines <mail@cbaines.net> wrote:
>
>>
>> Rather than writing:
>>
>>   (match-lambda
>>     ((alist ...)
>>
>> I'd just use
>>
>>   (lambda (alist)
>>
>> as I think that's equivalent right?
>
> Right, I did this.

Great. Unfortunately I missed a few things similar to this when I last
looked.

For this bit:

  (match-lambda
    ((label args ...)
     `(,label . ,(list->vector args))))

Given the thing being matched is a pair in an alist, I'd use (label
. args). I'd also perhaps use cons rather than `(, . ,) for creating the
pair.

>> >> I'd consider these options first probably:
>> >>
>> >>  - Could the data coming from derivation-differences-data have
>> >> vectors where appropriate already? The HTML code would probably
>> >> need to be adjusted, but I think that's fine.
>> >
>> > I tried this for days but with no success. Maybe the only way would
>> > be to tweak group-to-alist, but it touches many places, and I
>> > didn't want to mess with it.
>>
>> Maybe add another procedure that combines group-to-alist but generates
>> an alist with vectors as the values? (group-to-alist/vector maybe).
>
>> I think using let is OK, but I think just unpacking data-groups as
>> you've called it directly in to the alist is fine (so ,@data-groups),
>> rather than picking out the elements. JSON objects are unordered, so
>> the ordering isn't something that really matters.
>>
>> If you do go down this route though, I'd probably add a comment saying
>> what things are being added to the outer most alist, just to make the
>> code quicker to read.
>
> Well, I went down the second route, now I'm calling the ,@data-groups
> and I added a comment explaining its use.
> The main point here is, the code is working and it looks nice, but to
> get the data with the vectors seems to be right too. I'm sending the
> new patch for your review and I'll wait for your call, if you think I
> should try the first route or not.

This looks OK now.

The other thing I forgot to mention last time was using empty objects
(() in the code) for the hash and hash-algorithm for derivations where
those aren't applicable. I'd probably suggest using the symbol null, or
not including those fields in the object.

Hopefully that's all the bits to fix up, apologies for not mentioning
them last time.

Chris

signature.asc
Description: PGP signature