Avoid conditional compilation on whether tracing is enabled #914
Conversation
There was a problem hiding this comment.
I assume these changes suggest the runtime impact of disabled tracing utilities is not significant enough to justify their conditional compile-time compilation and corresponding maintenance burden. It seems like the compilers are also gracious enough not to warn about the always true/false conditions predecated on MONGOC_TRACE_ENABLED. LGTM.
| @@ -85,10 +103,8 @@ mongoc_log (mongoc_log_level_t log_level, | |||
| bson_once (&once, &_mongoc_ensure_mutex_once); | |||
|
|
|||
| stop_logging = !gLogFunc; | |||
There was a problem hiding this comment.
Should this check be combined into _mongoc_log_trace_is_enabled()?
There was a problem hiding this comment.
Unsure. I left the logic as close to the original as possible.
| */ | ||
| #define MONGOC_TRACE @MONGOC_TRACE@ | ||
| #cmakedefine01 MONGOC_TRACE |
There was a problem hiding this comment.
This unexpectedly broke PHPC's build system (autotools). I've reported CDRIVER-4247 to request a more portable approach.
Macros that discard their arguments under certain configurations can lead to missed refactors and unknowingly breaking code. This changeset changes conditional tracing to use C-language-level conditions rather than preprocessor conditions.