[Coco] Linux RBF filesystem support

[Steven Hirsch]

On Tue, 21 Oct 2008, Chuck Youse wrote:

> On Tue, 2008-10-21 at 07:48 -0400, Steven Hirsch wrote:
>> On Mon, 20 Oct 2008, Boisy Pitre wrote:
>
>> The threaded model always sounds good on
>> paper, but it can be pathologically complex to assure correctness and
>> avoid deadlocks and race conditions.
>
> Eh, mostly due to the fact that most programmers don't seem to grok that
> threads are useful but should be used sparingly.  Seems, too, that some
> newer synchronization schemes (e.g., serializing tokens) are completely
> ignored.  We're still in a mutex/condition-variable world.

I didn't mean to suggest that threads should never be used, only that they 
are often (as you seem to agree) misused.

>> I've done a reasonably large amount of cross-platform coding in my day and
>> find that with contemporary ANSI compliant compilers, C and C++ are at
>> least as portable as Java applications if you stay away from questionable
>> constructs.  More so, if you take into account that the language is not a
>> moving target.
>>
>
> I have to agree on this one; I haven't actually done too much poking
> under the hood with Toolshed and I am at a loss for where the system
> dependencies might lie, aside from word order and word size (and those
> are straightforward to cope with).
>
> Heh, I still think that C++ is a bit of a moving target.  The compiler
> implementations are far more varied in my experience, requiring that one
> keeps to a subset and allow for wide behavioral differences.  As such
> I'm not a huge fan and would rather use Ada for very large, complex
> systems.  (Hey, did I ever mention that I dated Bjarne's secretary's
> niece for a few years?  Claim to fame!)

That was certainly true up until perhaps 4 or 5 years ago, but things are 
much, much better now.  I'm working on the development of an extremely 
complex EDA application that makes heavy use of templates. 
Traditionally, this was nightmarish when moving between platforms but 
we've had surprisingly few problems.  With the release of g++4, I'd have 
to say even the GNU compiler is marching to the same drummer as, e.g. xlC 
(IBM RISC compiler).

And, hey, if it had been up to me the world would have settled on Modula-2 
or one of its object-oriented successors (Modula-3 or Oberon).  As it 
turns out, it wasn't up to me :-).  They pay me well to work in C++, so I 
smile and go with the flow.

Sorry to pull this off-topic.  I'll hold my peace...


Steve


--