[EMAIL PROTECTED] wrote:

> 
> Providing a capability often also imposes a policy.  Once a memory allocator
> is part of the standard toolset, it may work its way into e.g. semaphore
> allocation and stack allocation.  An IPC mechanism that relies on extending
> the thread structure imposes that additional cost on code that may not
> need the IPC mechanism.  "Many features" is a policy as much as "few features".

Providing mechanisms absolutely does not impose any policy, it just
eases one in implementing the one he/she preferes. A reacher toolbox can
just help in carrying out the job.
Steve/Pierre's RTAI real time mem manager is linked by default in its
schedulers, it is not part of them. So if one wants it can link without
it. In any case there is no cost being paid by those that do not need
it.
With that I/we agree it is usefull, but I/we are not imposing anything
to anybody.
RTAI users are free to use the wealth of RTAI provided mechanisms to
implement their own policies, the way they want, without any loss of
performance. RTAI aims at their utmost freedom not at teaching them what
to do.
 
> For RTLinux, if someone wants to write a malloc module that calls down to
> Linux kmalloc (this is absolutely trivial), then there is no problem and we
> would most likely include it in the distribution. But basic RTLinux
> services do not use dynamic allocators so that there will be as
> few  surprises as possible.

The same is true for RTAI and proves that a real time mem manager does
no harm but just adds flexibility.

Standard Victor's behaviour, already seen in the past :-).
Roughly stated it goes this way:
- no, no, no is wrong;
- you must/should be better doing this and that;
- if somebody wants to contribute...;
- ..... do we want to bet how long will it take to see a real time
memory manager in RTL?

Ciao, Paolo.
-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
---
For more information on Real-Time Linux see:
http://www.rtlinux.org/rtlinux/

Reply via email to