Will start_lmm become a standard library function?

mindrobots

I'm starting to document the second Toggle Tutorial which is lmm_c_toggle - starting a second LMM engine in its own COG. This example uses the lmm_start function which seems to be a perfect candidate for a standard library routine(provided it is modified to return a COG ID or Error indicator instead of void).

Is that a safe assumption for future releases so the user doesn't need to be burdened with

extern unsigned int _load_start_kernel[];

or the stack manipulation to put the function address onto the stack before calling cognew()?

It would be much nicer to just be able to

start_lmm(do_toggle, cog1_stack + STACK_SIZE);

without needing to put the start_lmm function into your code each time.

This isn't something I want to explain in the tutorials!

Thanks!

ersmith

Yes. The next release will have a _start_cog_thread function which will do what start_lmm does. It will also have an (incomplete) pthreads library so that you can start threads with standard pthreads functions. As you might imagine, the pthread_create function uses _start_cog_thread internally.

Eric