[#4656] Revert PR #3633 so that mux expressions will be constant folded #4660
Conversation
fruffy
added
core
|
Looks like this is supceded by #4657, so this can be closed? |
|
@ChrisDodd It is an alternate approach to #4657. I plan to close one after the other is merged. |
kfcripps
force-pushed
the
4656-revert-3633
branch
from
ec4d304
to
c2ac874
Compare
|
As mentioned in #4657, I'm actually preferring this PR at this point. |
| } | ||
| apply { | ||
| switch(baz()) { | ||
| 1 + 2 == 3 ? 1 : 2 : { foo(); } |
There was a problem hiding this comment.
I am just surprised this is not syntactically ambiguous, the 2 : look quite weird :-). (This is not an issue with the PR/test, I'm just surprised the grammar allows it).
There was a problem hiding this comment.
A single token of lookahead in Bison is apparently enough for it to figure it out, especially from the context that this is inside of a switch statement.
|
p4lang/p4-spec#1281 determines whether we may merge this particular PR, too. |
OK, so should this one be closed now? |
|
Sure, it will need to be rebased even if the outcome of p4lang/p4-spec#1281 is to remove the error. |
Reverts 4a74084, as discussed as an alternate fix for #4656 in #4657.