On 01-06-14 22:03, Julian Rohrhuber wrote:

On startup, sclang tries to load the GeneralHID spec:
|GeneralHID.fromID( this.platform.defaultHIDScheme );|
It is not clear whether this is really necessary; in any case, it posts
a warning:

|WARNING: GeneralHID.fromID : The HID scheme 'nil' is not installed
The current scheme is still 'nil'!
RESULT = 0
|

Should the call just be removed and happen on demand?

yes, please!

I will need to look into removing GeneralHID, while still maintaining
possible advantages it had over plain LID interfacing (i.e. extending
the LID interface with the features of GeneralHID).
As a cross platform forwarding scheme GeneralHID should be deprecated,
now that we have HID.

But we should keep LID as it does provide access to some devices on
Linux that you do not get via HID.

sincerely,
Marije

Ok, I've added a branch with the removed classes: d7d0bb1
Would that be OK to merge? I have kept the PgeneralHID and GeneralHIDGui code for future usage, but I guess it may look different?

GeneralHID is still fully functional for Linux, so I would not like to
remove it completely yet.

Maybe for now just change this bit:

*fromID { arg id;

•  var newScheme = schemes[ id.asSymbol ];
  

•  if( newScheme.notNil, {
  

•      scheme = newScheme;
  

•  }, {
  

•      if(thisProcess.platform.name != \windows){ // on win we know it's
  

  not yet supported

•          ("GeneralHID.fromID : The HID scheme '" ++ id ++ "' is not
  

  installed\n" ++

•          "The current scheme is still '" ++ if( scheme.notNil, { scheme.id
  

  }) ++ "'!").warn;

•      }
  

•  });
  

•  ^scheme;
  

• }

To:

*fromID { arg id;
var newScheme = schemes[ id.asSymbol ];
if( newScheme.notNil, {
scheme = newScheme;
}, {
if(thisProcess.platform.name == \linux){ // only supported on linux
right now
("GeneralHID.fromID : The HID scheme '" ++ id ++ "' is not
installed\n" ++
"The current scheme is still '" ++ if( scheme.notNil, { scheme.id })
++ "'!").warn;
}
});
^scheme;
}

On 11-06-14 22:29, Julian Rohrhuber wrote:

Ok, I've added a branch with the removed classes: d7d0bb1
d7d0bb1
Would that be OK to merge? I have kept the PgeneralHID and GeneralHIDGui
code for future usage, but I guess it may look different?

—
Reply to this email directly or view it on GitHub
#1120 (comment).

Hm, it would really be nice not to have two implementations for the same thing. Couldn't GeneralHID be rewritten by redirecting to the new HID, and then be deprecated? Or is there any fundamental difference that needs support?

Hi,

the fundamental thing is that the new HID implementation on Linux does
not detect the internal keyboard, trackpad and such on Linux. So this is
why I do not want to deprecate LID.
GeneralHID as a layer should be removed in the future, but some features
that are in there should be ported to LID before doing so.
I will have to find some time in the future to do this; hopefully in
August, after I have had some holidays.

I think that any further layers to unify access to input devices should
be in an external toolkit like Modality.

sincerely,
Marije

On 28-06-14 00:06, Julian Rohrhuber wrote:

Hm, it would really be nice not to have two implementations for the same
thing. Couldn't GeneralHID be rewritten by redirecting to the new HID,
and then be deprecated? Or is there any fundamental difference that
needs support?

—
Reply to this email directly or view it on GitHub
#1120 (comment).

fully fixed in pull request #1573