RE: [lwip-users] sys_timeout

[Curt McDowell]

Hi Peter,

> You just don't implement the timeouts, you implement 
> "semaphores with timeout" as required by lwIP. This is the 
> given interface for a multitasking O/S, and nobody keeps you 
> from implementing them with whatever means you like.

Thanks; that makes sense, and is documented pretty well in sys_arch.txt.  So
why is it so confusing, as evidenced by the number of questions arising here?

Here's a proposal to gain a cleaner 'apparent' separation without affecting the
architecture at all.  The idea is to rename those portions of sys_arch that are
in core, so they're more obviously part of core and not sys_arch:

(1) Reserve the 'sys' prefix exclusively for functions and defines
    to be provided by platform ports.
(2) Rename sys.c to task.c.
(3) Rename sys_sem_wait() to task_sem_wait().
    Rename sys_mbox_fetch() to task_mbox_fetch().
(4) Rename sys_arch_sem_wait() to sys_sem_wait().
    Rename sys_arch_mbox_fetch() to sys_mbox_fetch().
(5) Rename sys_timeout() to task_queue(), deconfuzzling it with timeout.
(6) Rename sys_timeouts() to sys_thread_data() and redefine it to return a
    per-task block of specified-size opaque context.

Regards,
Curt