On 6/2/24 23:45, Andi Kleen wrote:
No intentional semantics change.gcc/cp/ChangeLog: * cp-tree.h (struct cstr): Add structure. (get_cstr): Declare. (extract_cstr): Declare. (free_cstr): Declare. * semantics.cc (finish_static_assert): Factor out constant string expression extraction code and move to... (get_cstr): Here. (extract_cstr): Dito. (free_cstr): Dito. --- gcc/cp/cp-tree.h | 17 +++ gcc/cp/semantics.cc | 292 +++++++++++++++++++++++++------------------- 2 files changed, 184 insertions(+), 125 deletions(-) diff --git a/gcc/cp/cp-tree.h b/gcc/cp/cp-tree.h index 565e4a9290e2..25b8033db788 100644 --- a/gcc/cp/cp-tree.h +++ b/gcc/cp/cp-tree.h @@ -9015,6 +9015,23 @@ struct push_access_scope_guard } };+/* Extracting strings from constexpr */+ +/* Temporary data for extracting constant string. */ +struct cstr +{ + tree message; + tree message_data; + tree message_sz; + char *buf; +}; + +bool get_cstr (tree message, location_t location, const char *what, const char *what2, + cstr &cstr); +bool extract_cstr (const char *what, const char *what2, location_t location, + cstr &cstr, const char * & msg, int &len); +void free_cstr (cstr &cstr);
get_ and extract_ are confusingly similar names. Let's make cstr more of a class, initialized at least from 'message', and perhaps from the other context information.
Then get_ can be something like cstr::type_check. And then free_cstr is the destructor, and the copy constructor is deleted.
And please don't use the same name for the class and parameter/variables of that type.
+ error_at (location, "%qs %s must be a string "
From ABOUT-GCC-NLS: "Avoid using %s to compose a diagnostic message from multiple translatable strings; instead, write out the full diagnostic message for each variant. Only use %s for message components that do not need translation, such as keywords."
Also, if you are passing an English string for a diagnostic to a function that isn't from diagnostic.h, you need to use the intl.h macros to mark it for translation.
Jason
