Add Descriptor::iter_pk
#823
Conversation
This commit was -not- AI-generated, though probably it should've been.
apoelstra
force-pushed
the
push-slnpvpouumpw
branch
2 times, most recently
from
d3fab75 to
82f86e1
Compare
This one -was- AI-generated, though I reviewed it.
apoelstra
force-pushed
the
push-slnpvpouumpw
branch
from
82f86e1 to
aa42cd3
Compare
| use crate::{miniscript, Miniscript, MiniscriptKey}; | ||
|
|
||
| /// Iterator over all the keys in a descriptor. | ||
| pub struct PkIter<'desc, Pk: MiniscriptKey> { |
There was a problem hiding this comment.
This is fine. But Ideally, we want this to be a sum type instead of a product type? But then we get whole loads of matches everywhere?
There was a problem hiding this comment.
The main reason to use a product type was that Taproot descriptors have both single keys and (optionally) a taptree iterator.
And in PkIter::next I have a construction where the taptree iterator yields new Miniscript pkiters.
I can try to refactor this to use a sum type but I think it would result in more repeated/redundant code.
There was a problem hiding this comment.
I can try to refactor this to use a sum type but I think it would result in more repeated/redundant code.
Yeah, that would be messy. Ideally, we want to stick the mantra of "make invalid states unrepresentable"; but it might be overkill for this small change
| use crate::{miniscript, Miniscript, MiniscriptKey}; | ||
|
|
||
| /// Iterator over all the keys in a descriptor. | ||
| pub struct PkIter<'desc, Pk: MiniscriptKey> { |
There was a problem hiding this comment.
I can try to refactor this to use a sum type but I think it would result in more repeated/redundant code.
Yeah, that would be messy. Ideally, we want to stick the mantra of "make invalid states unrepresentable"; but it might be overkill for this small change
|
Let's revisit this after we get rid of the |
Fixes #821.
Should backport to 12.x.