-
-
Notifications
You must be signed in to change notification settings - Fork 53
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Differing output for VM vs. IR codegen #340
Comments
and the vm-generated code:
|
Interesting. I'm not finding any errors:
Here's my info:
I'm running off of
|
Asking
|
If I modify the
|
Okay, if I add an I've tracked it down to how the VM IR is generated from I think the old behavior was correct, although that commit indicates that there's at least one corner case it didn't handle ( This then raises two questions:
This also raises an interesting potential peephole optimization. In the non-minimized VM IR code, when the code enters state 2, it will fail to match, but it first consumes all characters in the input and only then fails. This case can/should be detected and optimized to a |
For the following retest case:
I believe the IR-generated C is correct:
and the VM code (either interpreted, vmc or other VM-based codegen) is incorrect:
This is on the
kate/ci-retest-pcre-suite
branch, where 227ca72 introduced fsm_minimise() as well as fsm_determinise() for retest.The text was updated successfully, but these errors were encountered: