diff options
author | Ed Schonberg <schonberg@adacore.com> | 2021-12-14 15:44:16 +0100 |
---|---|---|
committer | Pierre-Marie de Rodat <derodat@adacore.com> | 2022-05-09 09:27:33 +0000 |
commit | f1231d9a0baab25e0a07b0d6146b91e59ecfadb8 (patch) | |
tree | 98c46cf63b1868891eadb818c92566203d3e7ac5 /Makefile.tpl | |
parent | 28fffc452eedb284d047b83adc7f8772b15bb0f8 (diff) | |
download | gcc-f1231d9a0baab25e0a07b0d6146b91e59ecfadb8.zip gcc-f1231d9a0baab25e0a07b0d6146b91e59ecfadb8.tar.gz gcc-f1231d9a0baab25e0a07b0d6146b91e59ecfadb8.tar.bz2 |
[Ada] Fix check for looking for user defined literals
Recall that Has_Possible_Literal_Aspects only comes in if a given node
does not have any visible interpretation. If one operand is a literal we
assume that there may be a user-defined literal defined for some type to
be determined during the downward pass of resolution, but that depends
on the existence of some user-defined function or operator that can take
that literal as an actual. If analysis did not find such, the call is
illegal. I think that the following must be added to Operator_Check to
recover the proper error.
Note that this breaks current partial implementation of user-defined
literals.
gcc/ada/
* sem_ch4.adb (Operator_Check): Fix condition.
Diffstat (limited to 'Makefile.tpl')
0 files changed, 0 insertions, 0 deletions