Skip to content

Made casts explicit in the IRPrinter and StmtToHtml #9057

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
mcourteaux wants to merge 5 commits into
base: main
Choose a base branch
from
Open

Made casts explicit in the IRPrinter and StmtToHtml #9057

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
1c03712
Made casts explicit in the IRPrinter and StmtToHtml
mcourteaux Mar 17, 2026
34e01b6
IRPrinter: Let and LetStmt have explicit type: let varname : type = v…
mcourteaux Mar 17, 2026
14a649a
Fixup for correctness/callable_errors
mcourteaux Mar 17, 2026
db31812
Reflect changes in StmtToHtml.
mcourteaux Mar 17, 2026
79a83ba
Fixup test_internal for IRPrinter.
mcourteaux Mar 17, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 25 additions & 11 deletions src/IRPrinter.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,20 +31,20 @@ using std::vector;
ostream &operator<<(ostream &out, const Type &type) {
switch (type.code()) {
case Type::Int:
out << "int";
out << "i";
break;
case Type::UInt:
out << "uint";
out << "u";
break;
case Type::Float:
out << "float";
out << "f";
break;
case Type::Handle:
// ensure that 'const' (etc) qualifiers are emitted when appropriate
out << "(" << type_to_c_type(type, false) << ")";
break;
case Type::BFloat:
out << "bfloat";
out << "bf";
break;
}
if (!type.is_handle()) {
Expand Down Expand Up @@ -262,8 +262,8 @@ void IRPrinter::test() {
ostringstream source;
source << allocate;
std::string correct_source =
"allocate buf[float32 * 1023] in Stack\n"
"let y = 17\n"
"allocate buf[f32 * 1023] in Stack\n"
"let y : i32 = 17\n"
"assert(y >= 3, halide_error_param_too_small_i64(\"y\", y, 3))\n"
"produce buf {\n"
" parallel (x, -2, y + 2) {\n"
Expand Down Expand Up @@ -724,12 +724,16 @@ void IRPrinter::visit(const IntImm *op) {
if (op->type == Int(32)) {
stream << imm_int(op->value);
} else {
stream << typep(op->type) << imm_int(op->value);
stream << ansi_imm_int << op->value << "_i" << op->type.bits() << ansi_reset;
}
}

void IRPrinter::visit(const UIntImm *op) {
stream << typep(op->type) << imm_int(op->value);
if (op->type.bits() == 1) {
stream << ansi_imm_int << (op->value ? "true" : "false") << ansi_reset;
} else {
stream << ansi_imm_int << op->value << "_u" << op->type.bits() << ansi_reset;
}
}

void IRPrinter::visit(const FloatImm *op) {
Expand Down Expand Up @@ -796,10 +800,20 @@ void IRPrinter::visit(const StringImm *op) {
}

void IRPrinter::visit(const Cast *op) {
stream << type(op->type);
#if 0
// More explicit style of denoting a cast, which we did not yet agree upon.
// Leaving it in as commted out code, because it might be useful at some point.
stream << kw("cast<") << type(op->type) << kw(">");
openf();
print_no_parens(op->value);
closef();
#else
std::stringstream ss;
ss << op->type;
openf(ss.str().c_str());
print_no_parens(op->value);
closef();
#endif
}

void IRPrinter::visit(const Reinterpret *op) {
Expand Down Expand Up @@ -1084,7 +1098,7 @@ void IRPrinter::visit(const Let *op) {
if (!implicit_parens) {
stream << paren("(");
}
stream << paren("let ") << var(op->name) << paren(" = ");
stream << paren("let ") << var(op->name) << paren(" : ") << paren(op->value.type()) << paren(" = ");
print(op->value);
stream << paren(" in ");
if (!is_summary) {
Expand All @@ -1098,7 +1112,7 @@ void IRPrinter::visit(const Let *op) {

void IRPrinter::visit(const LetStmt *op) {
ScopedBinding<> bind(known_type, op->name);
stream << get_indent() << kw("let ") << var(op->name) << kw(" = ");
stream << get_indent() << ansi_kw << "let " << ansi_reset << op->name << ansi_kw << " : " << op->value.type() << " = " << ansi_reset;
Copy link
Member

@abadams abadams Mar 23, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I like this but I don't want to risk double-printing the type.

We should not do this if the RHS is going to immediately print the type (because it is a Call, VectorReduce, Variable, or immediate node). We should also skip it if the type is Int(32), for the same reason as we skip it on Variable nodes.

{
ScopedValue<int> reset_paren_depth(paren_depth, 0);
print_no_parens(op->value);
Expand Down
20 changes: 15 additions & 5 deletions src/StmtToHTML.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1429,19 +1429,25 @@ class HTMLCodePrinter : public IRVisitor {
void visit(const Cast *op) override {
std::string type_str = type_to_string(op->type);
print_opening_tag("span", "Cast");
print_opening_tag("span", "matched");
print_opening_tag("span", "matched keyword");
#if 0
print_text("cast&lt;");
print_type(op->type);
print_text("(");
print_text("&gt;");
#else
print_text(type_str);
#endif
print_closing_tag("span");
print_html_element("span", "matched", "(", type_str);
print(op->value);
print_html_element("span", "matched", ")");
print_html_element("span", "matched", ")", type_str);
print_closing_tag("span");
}

void visit(const Reinterpret *op) override {
std::string type_str = type_to_string(op->type);
print_opening_tag("span", "Reinterpret");
print_opening_tag("span", "matched Symbol", type_str);
print_opening_tag("span", "matched keyword", type_str);
print_text("reinterpret&lt;");
print_type(op->type);
print_text("&gt;");
Expand Down Expand Up @@ -1574,7 +1580,9 @@ class HTMLCodePrinter : public IRVisitor {
print_opening_tag("span", "matched");
print_text("(");
print_html_element("span", "keyword", "let ");
print_variable(op->name, op->type);
print_variable(op->name, op->value.type());
print_html_element("span", "Operator Assign", " : ");
print_type(op->type);
print_html_element("span", "Operator Assign", " = ");
print_closing_tag("span");
print(op->value);
Expand All @@ -1593,6 +1601,8 @@ class HTMLCodePrinter : public IRVisitor {
print_opening_tag("span", "matched");
print_html_element("span", "keyword", "let ");
print_variable(op->name, op->value.type());
print_html_element("span", "Operator Assign", " : ");
print_type(op->value.type());
print_html_element("span", "Operator Assign", " = ");
print_closing_tag("span"); // matched
print(op->value);
Expand Down
20 changes: 10 additions & 10 deletions test/correctness/callable_errors.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,9 +67,9 @@ void test_bad_untyped_calls() {
expect_failure(c(Buffer<uint8_t, AnyDims>(), 2, 1.0f, result1), "Buffer argument p_img is nullptr");
expect_failure(c(Buffer<void, 2>(), 2, 1.0f, result1), "Buffer argument p_img is nullptr");
expect_failure(c(Buffer<void, AnyDims>(), 2, 1.0f, result1), "Buffer argument p_img is nullptr");
expect_failure(c(42, 2, 1.0f, result1), "Argument 1 of 4 ('p_img') was expected to be a buffer of type 'uint8' and dimension 2");
expect_failure(c(in1, 2.25, 1.0f, result1), "Argument 2 of 4 ('p_int') was expected to be a scalar of type 'int32' and dimension 0");
expect_failure(c(in1, 2, 1, result1), "Argument 3 of 4 ('p_float') was expected to be a scalar of type 'float32' and dimension 0");
expect_failure(c(42, 2, 1.0f, result1), "Argument 1 of 4 ('p_img') was expected to be a buffer of type 'u8' and dimension 2");
expect_failure(c(in1, 2.25, 1.0f, result1), "Argument 2 of 4 ('p_int') was expected to be a scalar of type 'i32' and dimension 0");
expect_failure(c(in1, 2, 1, result1), "Argument 3 of 4 ('p_float') was expected to be a scalar of type 'f32' and dimension 0");
expect_failure(c(in1, 2, 1.0f, (const halide_buffer_t *)nullptr), "Buffer argument fn1 is nullptr");
expect_failure(c(in1, 2, 1.0f, (halide_buffer_t *)nullptr), "Buffer argument fn1 is nullptr");
expect_failure(c(in1, 2, 1.0f, Buffer<const uint8_t, 2>()), "Buffer argument fn1 is nullptr");
Expand Down Expand Up @@ -112,9 +112,9 @@ void test_bad_untyped_calls() {
expect_failure(c(&context, Buffer<uint8_t, AnyDims>(), 2, 1.0f, result1), "Buffer argument p_img is nullptr");
expect_failure(c(&context, Buffer<void, 2>(), 2, 1.0f, result1), "Buffer argument p_img is nullptr");
expect_failure(c(&context, Buffer<void, AnyDims>(), 2, 1.0f, result1), "Buffer argument p_img is nullptr");
expect_failure(c(&context, 42, 2, 1.0f, result1), "Argument 1 of 4 ('p_img') was expected to be a buffer of type 'uint8' and dimension 2");
expect_failure(c(&context, in1, 2.25, 1.0f, result1), "Argument 2 of 4 ('p_int') was expected to be a scalar of type 'int32' and dimension 0");
expect_failure(c(&context, in1, 2, 1, result1), "Argument 3 of 4 ('p_float') was expected to be a scalar of type 'float32' and dimension 0");
expect_failure(c(&context, 42, 2, 1.0f, result1), "Argument 1 of 4 ('p_img') was expected to be a buffer of type 'u8' and dimension 2");
expect_failure(c(&context, in1, 2.25, 1.0f, result1), "Argument 2 of 4 ('p_int') was expected to be a scalar of type 'i32' and dimension 0");
expect_failure(c(&context, in1, 2, 1, result1), "Argument 3 of 4 ('p_float') was expected to be a scalar of type 'f32' and dimension 0");
expect_failure(c(&context, in1, 2, 1.0f, (const halide_buffer_t *)nullptr), "Buffer argument fn2 is nullptr");
expect_failure(c(&context, in1, 2, 1.0f, (halide_buffer_t *)nullptr), "Buffer argument fn2 is nullptr");
expect_failure(c(&context, in1, 2, 1.0f, Buffer<const uint8_t, 2>()), "Buffer argument fn2 is nullptr");
Expand Down Expand Up @@ -162,16 +162,16 @@ void test_bad_typed_calls() {

// Calls to make_std_function fail
c.make_std_function<bool, int32_t, float, Buffer<uint8_t, 2>>();
expect_failure(-1, "Argument 1 of 4 ('p_img') was expected to be a buffer of type 'uint8' and dimension 2");
expect_failure(-1, "Argument 1 of 4 ('p_img') was expected to be a buffer of type 'u8' and dimension 2");

c.make_std_function<Buffer<uint8_t, 2>, bool, float, Buffer<uint8_t, 2>>();
expect_failure(-1, "Argument 2 of 4 ('p_int') was expected to be a scalar of type 'int32' and dimension 0");
expect_failure(-1, "Argument 2 of 4 ('p_int') was expected to be a scalar of type 'i32' and dimension 0");

c.make_std_function<Buffer<uint8_t, 2>, int32_t, bool, Buffer<uint8_t, 2>>();
expect_failure(-1, "Argument 3 of 4 ('p_float') was expected to be a scalar of type 'float32' and dimension 0");
expect_failure(-1, "Argument 3 of 4 ('p_float') was expected to be a scalar of type 'f32' and dimension 0");

c.make_std_function<Buffer<uint8_t, 2>, int32_t, float, bool>();
expect_failure(-1, "Argument 4 of 4 ('fn3') was expected to be a buffer of type 'uint8' and dimension 2");
expect_failure(-1, "Argument 4 of 4 ('fn3') was expected to be a buffer of type 'u8' and dimension 2");
}

// Test custom error handler in the JITUserContext
Expand Down
Loading