diff options
| author | Bob Duff <duff@adacore.com> | 2019-08-14 09:52:20 +0000 |
|---|---|---|
| committer | Pierre-Marie de Rodat <pmderodat@gcc.gnu.org> | 2019-08-14 09:52:20 +0000 |
| commit | 2d1439c7ad59625fea5598dda6679c6f3be1fa1c (patch) | |
| tree | 4e7268760ed77b8d6f435ba9fb87ffb85aa23925 /gcc/ada/inline.adb | |
| parent | 4b96d3861e74b8df1032f4317230408248e4bf09 (diff) | |
| download | gcc-2d1439c7ad59625fea5598dda6679c6f3be1fa1c.zip gcc-2d1439c7ad59625fea5598dda6679c6f3be1fa1c.tar.gz gcc-2d1439c7ad59625fea5598dda6679c6f3be1fa1c.tar.bz2 | |
[Ada] Defer processing of unknown CTW/E conditions to the back end
2019-08-14 Bob Duff <duff@adacore.com>
gcc/ada/
* sem_prag.adb (Process_Compile_Time_Warning_Or_Error): Defer
processing to the back end in all cases where the pragma's
condition is not known at compile time during the front end
(except in generics), as opposed to detecting 'Size attributes
and the like. This ensures that we take advantage of whatever
can be compile-time known after running the back end, as opposed
to having the front end guess what the back end can do. Remove
a little duplicated code at the call site.
* gnat1drv.adb (Post_Compilation_Validation_Checks): Unlock the
Elists while in Validate_Compile_Time_Warning_Errors, because it
does analysis and name resolution, which sometimes involves
adding Elists.
From-SVN: r274466
Diffstat (limited to 'gcc/ada/inline.adb')
0 files changed, 0 insertions, 0 deletions