ManualParameter and tests

[alexcjohnson]

@AdriaanRol @damazter @MarkusJacobsen @guenp this is re: #36 - pretty simple usage:

self.add_parameter('bias_resistor', parameter_class=ManualParameter)

with an optional initial_value

ManualParameter inherits the default vals=Numbers() from Parameter - I think this is better than Anything() as a default but I guess it depends on how it will get used.

[damazter]

I think Instrument should be an optional argument in ManualParameter,
This is related to #40

[alexcjohnson]

I think Instrument should be an optional argument in ManualParameter,
see also the pull request ill post in a few minutes

Yes, that's awkward - curious to see your solution 😄 - note that Parameter had better strip it out if it's a kwarg, so it doesn't get passed along to Metadatable

[alexcjohnson]

Just a github usage note (for everyone, but I noticed @damazter do it here) - it's better to make a new comment than to edit an existing one, unless you're just fixing a typo or something, because edits do not trigger email notifications

[AdriaanRol]

@alexcjohnson

I like the syntax and ease of use. I just pulled it in and will probably use it to replace some of the get-set constructs in my qubit object. Not all untill this is either merged in or we set on an alternative.

Yesterday I had some discussion with @damazter on this subject. (see also #40 ). I think that this functionality is applicable more generally than the instrument class but not as general as the bare instrument class. Therefore it makes sense to insert a different parameter in the inheritance tree between the parameter (base class) and the instrument parameter.

This also relates to another idea we were having (but did not agree on optimal implementation yet) which relates to wrapping of arbitrary functions. Suppose you have some function which you want to loop (and get as if it were a param). That could either be an extra possiblity in the syntax of the loop (increases complexity) or you create a new parameter for which the function is an input argument on instantiating. This is the exact same as when creating a par in an instrument (EDIT: and setting get_cmd = your function) except that there is no instrument there.

[damazter]

This looks perfect to me,
this indeed gives similar flexibility to what I had in mind but without the extra class involved (which is a style choice I guess)

[alexcjohnson]

Great, I'm going to take your comments @damazter and @AdriaanRol as a 💃 and I'll take care of sync/async pairing in the Parameter base class in another PR (that's going to take some testing, and I'd like to get #38 merged in before I do too much more with tests)

[akhmerov]

👍 from me as well, seems useful to the numerics applications.

[guenp]

if it makes any difference, 💃 👍 from me as well

[AdriaanRol]

Awesome 👍