Previously, you (Holger Kruse) wrote:
>> Although striping the /dev prefix is somewhat handy it at first
seemed limiting in QNX where a serial port on another node has a node number or name
in front of /dev. ie: //4/dev/ser5 refers to serial port 5 on node 4 and can be opened
accross the net just like /dev/ser1 on whatever the local node is. Fortunately, QNX
allow one to symbolic link (prefix) a remote node into the local namespace. One could
thus prefix //4/dev/ser5 -> /dev/ser4.5 and open ser4.5 using REBOL - so REBOL does
not to be QNX net aware to benefit.
> At the moment REBOL does not directly support any QNX-specific
> features. It accesses QNX through standard Posix functions, which is
> why any QNX-specific extensions to the serial mechanism (or other
> features) probably won't work.
You miss my point here I think. What I mean is that since QNX allows building symbolic
links accross network nodes, REBOL does not need to know about the QNX network
features - it just works.
> This is also why some users have problems setting the baud rate from
> REBOL in QNX. Apparently some versions of QNX have some bugs in their > Posix
>emulation code. This is not a bug in REBOL. The same code works > fine for "regular"
>Unix flavors.
I hope that you'vde reported this to QSSL. I do find it hard to believe though. The
GNU stuff I've played with tends to handle the serial ports correctly if it compiles.
The QNX 4 posix functions seem to work well in code I've written with the exception
that flow control has a funny "lock" bit which is not obvious.
>> The crazy thing that REBOL does is make you open the port with a path name
>serial://port[N] where N is the index? of the port name in system/ports/serial. Thus
>in serial-path-to above, I find the index? or append the passed name and use the
>length? to get port[N]. Of course, REBOL then has to change this back into an open(
>"/dev/serX", ..) so to me it would make sense to use the port name instead of index.
> One of the main goals of /Core is interoperability across platforms.
> Introducing a standardized naming scheme for serial ports means that
> scripts do not have to be changed when moving them across platforms.
> With the current scheme the only part of a script that has to change
> is the port number, as opposed to changing full port specs.
How is changing a port number any easier that changing a name? Don't you still need to
either edit hard coded port numbers of build some cover function which creates a port
spec given a number parameter?
>> and REBOL would figure out that you want 2400 baud, 7 data bits, even parity and 2
>stop bits. I don't think that REBOL provides a mechanisim to change the option after
>an open.
> It does. Just change the parameters in the port structure and call
> 'update on the port.
Great I had assumed that there must be a method, just hadn't a clue on where to find
it. This is probably my biggest gripe re REBOL - it does a lot of cool things, but
figuring out how is a mystery. Yes, I've read the official guide, but port coverage is
pretty skimpy. Of course serial ports are new so I know that I'm out on a limb there.
Dave
--
David L. Hawley D.L. Hawley and Associates 1.503.274.2242
Software Engineer [EMAIL PROTECTED]
--
To unsubscribe from this list, please send an email to
[EMAIL PROTECTED] with "unsubscribe" in the
subject, without the quotes.
