On 06/18/2015 10:39 AM, David Malcolm wrote:
On Thu, 2015-06-18 at 11:41 -0400, Patrick Palka wrote:
On Tue, Jun 9, 2015 at 1:31 PM, Patrick Palka <patr...@parcs.ath.cx> wrote:
This patch refactors the entry point of -Wmisleading-indentation from:
void
warn_for_misleading_indentation (location_t guard_loc,
location_t body_loc,
location_t next_stmt_loc,
enum cpp_ttype next_tok_type,
const char *guard_kind);
to
struct token_indent_info
{
location_t location;
cpp_ttype type;
rid keyword;
};
void
warn_for_misleading_indentation (const token_indent_info &guard_tinfo,
const token_indent_info &body_tinfo,
const token_indent_info &next_tinfo);
The purpose of this refactoring is to expose more information to the
-Wmisleading-indentation implementation to allow for more advanced
heuristics and for better coverage.
(I decided to keep the usage of const references because nobody
seems to mind. Also I added a new header file, c-indentation.h.)
gcc/c-family/ChangeLog:
* c-indentation.h (struct token_indent_info): Define.
(get_token_indent_info): Define.
(warn_for_misleading_information): Declare.
* c-common.h (warn_for_misleading_information): Remove.
* c-identation.c (warn_for_misleading_indentation):
Change declaration to take three token_indent_infos. Adjust
accordingly.
* c-identation.c (should_warn_for_misleading_indentation):
Likewise. Bail out early if the body is a compound statement.
(guard_tinfo_to_string): Define.
gcc/c/ChangeLog:
* c-parser.c (c_parser_if_body): Take token_indent_info
argument. Call warn_for_misleading_indentation even when the
body is a semicolon. Extract token_indent_infos corresponding
to the guard, body and next tokens. Adjust call to
warn_for_misleading_indentation accordingly.
(c_parser_else_body): Likewise.
(c_parser_if_statement): Likewise.
(c_parser_while_statement): Likewise.
(c_parser_for_statement): Likewise.
gcc/cp/ChangeLog:
* parser.c (cp_parser_selection_statement): Move handling of
semicolon body to ...
(cp_parser_implicitly_scoped_statement): .. here. Call
warn_for_misleading_indentation even when the body is a
semicolon. Extract token_indent_infos corresponding to the
guard, body and next tokens. Adjust call to
warn_for_misleading_indentation accordingly. Take
token_indent_info argument.
(cp_parser_already_scoped_statement): Likewise.
(cp_parser_selection_statement, cp_parser_iteration_statement):
Extract a token_indent_info corresponding to the guard token.
Pinging this series.
FWIW, they look reasonable to me; I'm not a reviewer.
But as the implementer of the warning, your comments/thoughts are
definitely helpful in the review process.
We've never worked too hard to find a way to formalize this into a set
of policies and procedures, which is probably a mistake.
jeff