De-facto for example trice( "hi" ); is a function call and TRICE( "hi" ); a code insertion. It is not obvious to see but you can follow up like in this example:

Find definition of trice( brings you to line 33 in triceOn.h:

#define trice(  tid, fmt, ...) TRICE_CONCAT2(trice_,   TRICE_COUNT_ARGUMENTS(__VA_ARGS__))(tid, fmt, ##__VA_ARGS__)

This ends up in trice_2 when you have 2 arguments, what ends up in trice32_2, when TRICE_DEFAULT_PARAMETER_BIT_WIDTH == 32.
In file triceOn.h line 354 we see #define trice32_2(tid, fmt, v0, v1) trice32fn_2(tid, (uint32_t)(v0), (uint32_t)(v1)) and that finally is a defined function in trice32.c line 339:

void trice32fn_2(uint16_t tid, uint32_t v0, uint32_t v1) {
	trice32m_2(tid, v0, v1);
}

trice32m_2 is a macro adding the trice code into the function body, so each trice( "hi" ) is a function call. When you follow the code of TRICE in a similar way you will see the macro invocation directly and not a function call. That is a very bit faster but includes the Trice code on each TRICE invocation, where using trice just calls the appropriate function.

Historically there where only TRICE macros (for max speed), but then I realized too much memory usage, when you have many times TRICE in the code and added trice macros, which are in fact macros, but are finally function calls.

You should be able to write trice32fn_2( id(123(, "hi %d %d", 1, 2 ); and get exactly the same code as with trice( id(123(, "hi %d %d", 1, 2 );. But the trice parser will ignore such code during ID management, I guess.

Uh oh!

Feature: function calls instead of inline #579

Description

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions