Re: [Linux-gpib-general] Python bindings (patch)

[Dan L. <dl...@gm...>]

Frank Mori Hess writes:

> On Monday 15 September 2008 12:44, Dan Lenski wrote:
>> * Function naming: some function names are expanded, e.g. ibwrt() ->
>> gpib.write, while others are left in their terse form, e.g. ibtrg() ->
>> gpib.trg().  Should these be named uniformly?
> 
> Yes
> 
>> Should we keep the terse 
>> names or the long names?
> 
> I slightly prefer the long names, but either way is fine.
> 
>> * Return values: some Python functions return their C return values
>> (e.g. the wrappers for ibdev(), ibfind(), ibwait()).  Others get the
>> interesting return value from a C pointer, e.g. the wrappers for ibrd(),
>> ibln(), ibask (), etc.  That's all well and good I think.  But many of
>> the Python functions return *nothing* for no apparent reason.  Should
>> these just return ibsta as in C?  There's no downside to this...
> 
> Ok

You want it, you got it.  Here's a patch on the current SVN.  A whole 
bunch of changes:

* fixed error handling to get errno from the right place
* __doc__ strings for every function (modeled on the ones Keith had written)
* a few function names made less terse:
    cmd->command
    tmo->timeout
    trg->trigger
  (I left ren, rsp, and ifc alone since remote_enable, request_serial_poll, 
  and interface_clear seemed unreasonable long to me, but those could be 
  changed as well.)
* integer constants for IbaXXXX and IbcXXXX added to the module
* all functions--that don't have something else useful to return--return 
  ibsta, just like the C versions
* the read() function now puts its result directly into the Python output
  string, with /no copying/ from a C string to a Python string.  This should 
  improve efficiency of large and/or frequent reads.

I'm itching to improve the Python wrapper class in Gpib.py, 
but decided to restrain myself from any further changes in this patch :-)  
Especially if some people are using it in its current form.

Dan