smm broken

M

Thread Starter

Mario de Sousa

Hello all,

Jiri Baum wrote:

> Hello Mario,
>
> we should really be holding these conversations on the list (though not
> sending code) - please reply on-list if you agree.
>

I have been chating with Jiri off-list for some time, but as he is going offline, it'd probably be better to start getting everybody involved.

The smm code is presently broken if --PLCisolate is used. I have sent a correction to Jiri, but as it included my way of dividing up the smm_lib.c code into several files he will probably not have time to look into it and put it up. (Jiri?)

> > The problem is that you redefine globmap_ , privmap_ , confmap_ , ...
> > variables in several .c files. What is happening is that these variables,
> > even though they have the same name, have multiple copies of each, one
> > for each definition.
>
> Yes, that was intentional - but now that I think of it I realize that the
> smm_remote module uses the smm_local version of pt_by_name. Oops!
>

(...)

> > Try running ./plctest -s P0.0 --PLCisolate to see the smm crash.
>
> I haven't tried it, but with pt_by_name() the way it is, it has to crash.
>
> > I am including my version of dividing up the smm.
>
> I'll have a look at it later...

(...)

> > Future work:
> > What would you like me to do?
> > I was thinking of either implementing the multi-bit points in the smm, or
> > else implementing a PID controller. Obviously this last would require
> > multi-bit points.
>
> Multi-bit points will be required by just about everything, so it'd be good
> to have them implemented soon. I won't be doing anything much over the two
> days remaining to me, so go ahead!
>
> For the PID controller, send a note to the mailing list, just in case
> somebody else has dibs on that.
>

I have just finished a VERY rough draft of the PID controller. I will now start with the multi-bit points so I can test it. The PID controller
cannot work on single bit points...

Is anybody else interested in writing the PID controller? (Please reply to the list). I can send them the code if they wish to continue. I won't put it up until the multi-bit points are
supported and the pid controller has been tested.


Ken: Do you think I could have write access to the CVS? I have never used cvs before, so I would first be uploading a few files to the /test directory to make sure I will not muck up anything.


Mario.

_______________________________________________
LinuxPLC mailing list
[email protected]
http://linuxplc.org/mailman/listinfo/linuxplc
 
Hello,

Mario de Sousa:
> The smm code is presently broken if --PLCisolate is used. I have sent a
> correction to Jiri, but as it included my way of dividing up the
> smm_lib.c code into several files he will probably not have time to look
> into it and put it up. (Jiri?)

I'm going through it now; I'll see how I go. Shouldn't be a problem moving the variables that plc_pt_by_name() needs into smm_private.h, though.

I'll be on-line while I'm doing that, just for a change.


Jiri
--
Jiri Baum <[email protected]>
"Coffee machine out of paper" (error message missing from Coffee mini-HOWTO)

_______________________________________________
LinuxPLC mailing list
[email protected]
http://linuxplc.org/mailman/listinfo/linuxplc
 
Top