feat: Introduce REF= and REFERENCE_TO
#1251
Conversation
volsa
changed the title
REFERENCE_TOREF= and REFERENCE_TO
…into volsa/referenceto
volsa
force-pushed
the
volsa/referenceto
branch
2 times, most recently
from
1ac7bfd to
7772a3b
Compare
volsa
force-pushed
the
volsa/referenceto
branch
2 times, most recently
from
b60bc18 to
c5680bd
Compare
|
I think I addressed all your comments @mhasel , should be ready for a second round 🤞 |
| * `foo : ARRAY[...] OF REFERENCE TO (* ... *)` | ||
| * `foo : REF_TO REFERENCE TO (* ... *)` | ||
|
|
||
| Furthermore `REFERENCE_TO` variables must not be initialized in their declaration, e.g. `foo : REFERENCE TO DINT := bar`. |
There was a problem hiding this comment.
fyi, this is not entirely correct but will be addressed in the aliasing PR later on
| foo REF= bar; | ||
|
|
||
| // CHECK-NOT: ^5$ | ||
| printf('%d$N', foo); // We expect some random value here because foo isn't pointing to any variable yet |
There was a problem hiding this comment.
we are auto-dereferencing an uninitialized pointer here - this is undefined behaviour. Should we initialize these pointers with null like we are doing for VAR_INPUT {ref}? Ideally we would check for use before assignment but that feature feels a bit far off still.
There was a problem hiding this comment.
Discussed offline, generally in favor of NULL pointers here but I'd probably also address this in another PR. To have a second opinion here though cc @ghaith
There was a problem hiding this comment.
I think null makes sense here so that we don't have random values. I'm in favour of not having such tests tbh. We can't test undefined behaviour (by design). Also what happens if the random value is 5? I am more in favour of having a test that verifies the pointer value of foo is null. Not sure how to do this with the auto deref rules.. maybe it's enough to have an IR check that verifies the initial values of REFERENCE TO and Pointer fields in general.
There was a problem hiding this comment.
Sounds good, I've created an issue and deleted the test here. I'll address it in another PR since this one bloats up way too much if you'd ask me.
This PR introduces two new keywords, namely
REF=andREFERENCE TO:REF=is essentially syntactic sugar for an assignment where the right-hand side is wrapped in aREF()function call. Thereforefoo := REF(bar)andfoo REF= barare equivalent.REFERENCE TOis identical toREF_TOwith the exception of being auto-deref by default. A variablefoodeclared asREFERENCE TOwill therefore auto-deref on assignments, i.e.foo := 5is equivalent tofoo^ := 5.More information on CodeSys' REF= and REFERENCE TO documentation pages.