C: Support header files in ccall#1731
Conversation
| if (AST::is_a<AST::ConstantStr_t>(*call_d->m_keywords[i].m_value)) { | ||
| std::string header_name = AST::down_cast<AST::ConstantStr_t>( | ||
| call_d->m_keywords[i].m_value)->m_value; | ||
| bindc_name = s2c(al, header_name); |
There was a problem hiding this comment.
I'm using bindc_name to test the prototype and use it as a header file name which in LLVM is a function name to be called from C. Once this works, do we want to add a new argument in the function type to check which header file it is contained in?
There was a problem hiding this comment.
I think we do. Probably not in FunctionType but in Function.
There was a problem hiding this comment.
I don't think Function has a mechanism to store the c header file. It just has bindc_name for its interop with C.
| Function(symbol_table symtab, identifier name, ttype function_signature,
identifier* dependencies, expr* args, stmt* body, expr? return_var,
access access, bool deterministic, bool side_effect_free)
Should we go with this?
diff --git a/src/libasr/ASR.asdl b/src/libasr/ASR.asdl
index 34af5be3c..3f4b29b4a 100644
--- a/src/libasr/ASR.asdl
+++ b/src/libasr/ASR.asdl
@@ -363,7 +363,7 @@ ttype
| CPtr()
| TypeParameter(identifier param, dimension* dims)
| FunctionType(ttype* arg_types, ttype? return_var_type,
- abi abi, deftype deftype, string? bindc_name, bool elemental,
+ abi abi, deftype deftype, string? bindc_name, string? c_header, bool elemental,
bool pure, bool module, bool inline, bool static, ttype* type_params,
symbol* restrictions, bool is_restriction)
|
Thanks a lot @Smit-create for working on this! It looks like I found a workaround with #1732 and #1734 so that I am not blocked today anymore. This buys us time to design this well. We definitely need this feature. |
src/libasr/ASR.asdl
Outdated
| | TypeParameter(identifier param, dimension* dims) | ||
| | FunctionType(ttype* arg_types, ttype? return_var_type, | ||
| abi abi, deftype deftype, string? bindc_name, bool elemental, | ||
| abi abi, deftype deftype, string? bindc_name, string? c_header, bool elemental, |
There was a problem hiding this comment.
c_header I think should go into Function, not be part of the type (neither logical nor physical: lfortran/lfortran#1601).
certik
left a comment
certik
left a comment
There was a problem hiding this comment.
With those two changes this should be fine.
certik
left a comment
certik
left a comment
There was a problem hiding this comment.
There is one more thing that we should fix.
| #include <math.h> | ||
| #include "complex.h" | ||
| #include "inttypes.h" | ||
| #include "math.h" |
There was a problem hiding this comment.
I think this change is a problem, it should use <, not ".
There was a problem hiding this comment.
Yep, thanks. Reverted that.
| for (auto s: headers) { | ||
| to_include += "#include <" + s + ".h>\n"; | ||
| for (auto &s: headers) { | ||
| to_include += "#include \"" + s + "\"\n"; |
There was a problem hiding this comment.
For now let's use <, that is closer to correct.
We probably should distinguish between a header file that should be referenced with < (intrinsic ones) and " (user ones).
certik
left a comment
certik
left a comment
There was a problem hiding this comment.
I think this is good enough now.
I noticed there are some default header files and macros always included and we should eventually handle them in the same way (put the macros into a header file and only include it if needed), but that's for later.
I think this is good to go in. Thanks for this @Smit-create, very important new feature.
3 participants
#1613