Fortuitous inverse

[Phil Last]

I always keep an if operator in the ws:

      if←{
          1∊⍵⍵⊣⍵:⍺⍺⊣⍵
          ⍵
      }

This allows me to do, among an awful lot of other stuff:

      ⊂if≡

which I understand is going to be an i-beam in 16 as "enclose if simple".
What's not immediately obvious is that

      ⊃if≢

is in many cases its direct inverse "disclose if single".
Sometimes APL is just right even without meaning to be!

[Michael|Dyalog]

Hi Phil,
a minor detail: it's not going to be i-beam, it will have its own symbol!

[JohnS|Dyalog]

If - nice!
"Nest" (formerly known as enclose-if-simple) will be monadic ⊆.
Dyadic ⊆ will be APL2(⎕ML=3)-style "partition". We're hoping this will encourage people towards writing their code in migration-level 1.

[paulmansour]

Dyadic ⊆ will be APL2(⎕ML=3)-style "partition". We're hoping this will encourage people towards writing their code in migration-level 1.

I've been meaning to do this for a while, and that will certainly help. I feel somewhat duped by Dyalog ... I had always thought that one should be using the highest migration level...

Anyway, I want to change a large code base from ML 3 to ML 1. Can this be done programatically or is it a painful line-by-line inspection? I haven't thought that much about it yet...

[Phil Last]

If up-arrow were only either mix or first, right-shoe were only first or mix and if equal-underbar were only depth then yes, you could write a function to do it.

If you could find a sure way to distinguish a monadic from a dyadic call to the functions represented by those glyphs you could still write a function to do it that would work most of the time. More if you could identify the exceptions to your default ⎕ML=3. .

Otherwise ...

There again there are only about eight hundred occurrencies of each of equal-underbar and right-shoe and about a thousand of up-arrow in FlipDB.

But to go back to John's thesis I think Dyalog might hope more realistically that the new primitive dyad will encourage users to go back to ⎕ML=2 which is the point after which dyadic left-shoe gained its new definition.

To go back to ⎕ML=1 might require a run-time intervention, a temporary bolt-on to the tracer (say), to flag the hits that count as the interpreter can only ascertain whether it's got a monad or a dyad at that time.

[Phil Last]

This replaces a second identical copy of the above post. Dyalog was slow responding last evening and I seem to have hit "Submit" twice.