You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
HDC-spec mandates all devices, also those compliant with future versions of HDC-spec, to reply to the MessageTypeID=0xF0 with a sting like "HDC 1.0". This was implemented in issue #4
This issue is about designing an (elegant?) mechanism for HDC-host implementations to cope with devices that are compliant to different HDC-spec versions.
Some approaches, sorted from simple to complex:
Refuse to communicate at all with devices compliant to a different HDC-spec version.
Tolerate older MINOR versions of HDC-spec and suppress proxies for their missing capabilities.
Use a different hdcproto.host.router.MessageRouter class, depending on the MAJOR version of the HDC-spec of a device.
The text was updated successfully, but these errors were encountered:
The mechanism, as it's now pretty obvious, will be through a proxy factory, like the one I'm about to commit as DeviceProxyBase.proxy_factory().
The factory is given the full device-descriptor which contains at the top-level the HDC-version string. Based on its value either parametrized proxy classes or completely different proxy-class-modules can be instantiated.
This is still a potentiality and not yet a reality, but the roadmap is pretty clear now. 😄
…on and customizable proxy factory.
First working version of fully automated proxy creation, based on the descriptors reported by a device.
Generic proxy classes get customized with information contained in descriptors.
Even more customization is possible via the proxy factory. Not only custom proxy classes can be instantiated, but also descriptors and constructors can be "tweaked".
Transitioning from the obsolete "introspection" architecture to the fully IDL based architecture is now complete!
The proxy factory is also the obvious way how proxies will in future be able to adapt to different HDC protocol versions.
Fixes#1Fixes#24
Work in progress on #11
HDC-spec mandates all devices, also those compliant with future versions of HDC-spec, to reply to the MessageTypeID=0xF0 with a sting like "HDC 1.0". This was implemented in issue #4
This issue is about designing an (elegant?) mechanism for HDC-host implementations to cope with devices that are compliant to different HDC-spec versions.
Some approaches, sorted from simple to complex:
The text was updated successfully, but these errors were encountered: