bug#2816: regexp-opt: does not optimize 'Abc' and 'abc'

bug#2816: regexp-opt: does not optimize 'Abc' and 'abc'

[Jari Aalto]

A)

    (regexp-opt
      '("Abc"
        "abc"))

    => "\\(?:[Aa]bc\\)"

B)

    (regexp-opt
      '("Abc"
        "def"
        "abc"))

    => "\\(?:Abc\\|abc\\|def\\)"

Shouldn't regexp-opt optimize B case as well?

Jari

bug#2816: regexp-opt: does not optimize 'Abc' and 'abc'

[Stefan Monnier]

> A)

>     (regexp-opt
>       '("Abc"
>         "abc"))

>     => "\\(?:[Aa]bc\\)"

> B)

>     (regexp-opt
>       '("Abc"
>         "def"
>         "abc"))

>     => "\\(?:Abc\\|abc\\|def\\)"

> Shouldn't regexp-opt optimize B case as well?

The `opt' of regexp-opt should be understood in the sense of
optimization used in compilers, not in the formal mathematical sense of
"optimal".  I.e. it does some effort to improve things, but without any
guarantee about what is and what isn't optimized.

So, yes, it would be good if it did do what you suggest, but it's not
a bug if it doesn't.


        Stefan