Remove constraints blocks #652
Conversation
Schaeff
force-pushed
the
remove-constraints-blocks
branch
from
3500770 to
8af2582
Compare
| *body = self.expand_macros(std::mem::take(body)) | ||
| } | ||
| InstructionBody::CallableRef(..) => { | ||
| machine.statements = machine |
There was a problem hiding this comment.
This was changed to a functional approach, as a statement can yield many, and iter_mut does not allow that. Hence the usage of flat_map here.
| _ => {} | ||
| }); | ||
| MachineStatement::Pil(start, statement) => self | ||
| .expand_macros(vec![statement]) |
There was a problem hiding this comment.
It would make more sense to me for expand_macros to take a single statement rather than a list, but I tried to keep this diff small.
| @@ -40,7 +40,7 @@ impl<T: Display> Display for Machine<T> { | |||
| write_items_indented(f, &self.registers)?; | |||
| write_items_indented(f, &self.instructions)?; | |||
| write_items_indented(f, &self.callable)?; | |||
| write_items_indented(f, &self.constraints)?; | |||
| write_items_indented(f, &self.pil)?; | |||
There was a problem hiding this comment.
Is it a problem that these might be ordered differently? I mean can we always re-parse stuff that is printed here?
There was a problem hiding this comment.
Can you elaborate here? Within constraints, the order is preserved throughout the compiler, but eventually this should not be a requirement, right? There are no tests which parse outputs of this display afaik, this could make sense to add but it seems unrelated to this change.
There was a problem hiding this comment.
I'm just wondering if it might make sense to move the pil a bit further up or what happens if you mix constraints and instruction declarations and such, but yes, unrelated to this PR.
Schaeff
force-pushed
the
remove-constraints-blocks
branch
2 times, most recently
from
fa493ad to
71b1dc1
Compare
Schaeff
force-pushed
the
remove-constraints-blocks
branch
from
71b1dc1 to
0950ff5
Compare
|
still conflicts. |
Schaeff
force-pushed
the
remove-constraints-blocks
branch
from
0950ff5 to
a3ebf46
Compare
Becomes
I did not change
analysis/README.mdas the diff looks quite different now.I also tried to update the book in a way that made sense, pointing to the docs for powdr-pil when mentioning pil code in machines.