Byte-operator lowering: do not generate type casts to compound types [blocks: #2068] #4227
Conversation
tautschnig
requested review from
chrisr-diffblue,
kroening,
peterschrammel and
smowton
as code owners
| } | ||
| else if(target_type.id() == ID_string) | ||
| { | ||
| // this works, but is somewhat questionable |
There was a problem hiding this comment.
Why it works or why it is questionable?
There was a problem hiding this comment.
Because the only ID_string-typed expression we really deal with is a constant_exprt. But all tests pass when I do this 👼
There was a problem hiding this comment.
✔️
Passed Diffblue compatibility checks (cbmc commit: 175c12c).
Build URL: https://travis-ci.com/diffblue/test-gen/builds/101421426
| else | ||
| { | ||
| PRECONDITION_WITH_DIAGNOSTICS( | ||
| false, "bv_to_expr does not yet support ", target_type.id_string()); |
There was a problem hiding this comment.
That's not a precondition, use UNEXPECTEDCASE
There was a problem hiding this comment.
There is UNHANDLED_CASE, but the documentation suggests that this should not be used, and it wouldn't provide diagnostics either. So I'm inclined to leave this as is.
These aren't fully supported by the back-ends. Instead, bit-extract the components and construct an expression.
tautschnig
force-pushed
the
byte-op-conversion
branch
from
175c12c to
131fda0
Compare
|
Refactored and acted upon comments as far as possible. @owen-jones-diffblue @romainbrenguier @smowton could you please take another look? |
These aren't fully supported by the back-ends. Instead, bit-extract the
components and construct an expression.