Should double negation be eliminated?

[StreetStrider]

I believe it should. positive path + !A + !B. If first negative glob inverts positive path or pattern, then why second does not?

[amobiz]

Hi @StreetStrider, sorry for not responding immediately.

Could you please provide an actual use case?

The initial idea was that, regardless of how many components need to be excluded, they are all the parts the user wants to exclude. So, as long as there is a negative glob present, the entire glob should become negative.

[StreetStrider]

Consider setup: we have two subsystems A and B. A is responsible for transforming, say, .js, .ts files. We may have also other systems A2, A3 that are responsible for transforming other types of files. B is responsible for transforming the rest of the files that are not under any of A. (B, say, can copy the rest of the files 1-to-1.)

So, if A = [ '*.js', '*.ts' ] then B = [ '**/*', '!*.js', '!*.ts' ].

If one of As involves negation, we would have something like that:
A = [ '*.js', '*.ts', '!*.d.ts' ], so B should now include negated thing B = [ '**', '!*.js', '!*.ts', '*.d.ts' ]. So B should double negate and make glob positive, because it was previously negated by some of As. (Considering later globs override former negations)

[StreetStrider]

The second variant of the case: we have positive glob with negations and we have something like ignore option in other subsystem. If we want to ignore the opposite of the glob. If negations are involved they're need to be un-negated to become actually ignored.