Re: Pike object assembler code

On Tue, 14 Oct 2014, Stephen R. van den Berg wrote:

> Say I want to see what byte code a certain piece of Pike code compiles
> into.  What would be the best way to obtain a readable output?

Getting debug output including byte code mnemonics from the 
assembler/peep-hole optimizer is done by raising the assembler 
debug level to 2 or higher (done by calling Debug.assembler_debug() 
in a pike compiled with rtl-debug). Note that at level 3 you also get 
to see the parse tree, the result of the peep-hole pass, code compiled 
to evaluate constant expressions and the resulting symbol table.

> Ideally, there would be some symbolic output which resembles
> "Pike-assembler-code" intermingled with the Pike source code it
> came frome.

Each line of debug output is prefixed with the linenumber (but not the 
file name) in the corresponding source file.

> The reason I ask is mostly educational.  I'd like to roughly understand
> what code results in what low-level construct, so that I don't have
> to run a benchmark everytime to decide between construct a or b
> (deciding on saving space or time or both).

Enabling the assembler debug and/or the optimizer debug is a good way to 
get an understanding of how the Pike compiler works.

-- 
Henrik Grubbström					[email protected]
Roxen Internet Software AB