Underscore: _ variable in REPL
#2755
Conversation
src/libasr/codegen/asr_to_llvm.cpp
Outdated
| @@ -2709,11 +2712,20 @@ class ASRToLLVMVisitor : public ASR::BaseVisitor<ASRToLLVMVisitor> | |||
| } | |||
|
|
|||
| void visit_Variable(const ASR::Variable_t &x) { | |||
| if ((compiler_options.interactive) && | |||
There was a problem hiding this comment.
| if ((compiler_options.interactive) && | |
| if (compiler_options.interactive && |
Similarly others.
src/libasr/codegen/asr_to_llvm.cpp
Outdated
| @@ -2709,11 +2712,20 @@ class ASRToLLVMVisitor : public ASR::BaseVisitor<ASRToLLVMVisitor> | |||
| } | |||
|
|
|||
| void visit_Variable(const ASR::Variable_t &x) { | |||
| if ((compiler_options.interactive) && | |||
| (std::strcmp(x.m_name, "_") == 0) && | |||
There was a problem hiding this comment.
Why is it so that at some place we compare with (std::strcmp(x.m_name, "_") == 0) and at some we use (std::strcmp(x.m_name, compiler_options.po.global_underscore.c_str()) == 0)? What is the significance of compiler_options.po.global_underscore?
There was a problem hiding this comment.
The previous code block's last expression's value is stored in compiler_options.po.global_underscore, which is ___main__global_stmts_<int>__.
While generating IR for compiler_options.po.global_underscore, we store its hash in global_underscore_hash, then whenever we need a reference to _ variable, we use global_underscore_hash. So _ gets replaced with ___main__global_stmts_<int>__.
|
Looks good overall. Lets merge #2752 and rebase and resolve conflicts in this PR. After that I will have a look again. |
Vipul-Cariappa
force-pushed
the
underscore
branch
from
c015890 to
a4a2bc8
Compare
mostly I messed up the rebasing partially
Demo:
This PR is built on top of #2752. Therefore #2752 should be merged first.