krkeegan / insteon-mngr Goto Github PK

Links that would otherwise not appear in the group links page such as:

• Controller links on a device for a group that DNE

Vertical responsive page doesn't really work for small screens.

Start with device level then global

Call the specific i1 or i2 sequences from that. Get rid of specialized calls everywhere else.

I don't think this solution is good either, or is at least incomplete. Simultaneous queued sequences would still bind to the same address.

Bind needs to do something so that subsequent items see the address as taken

Likely need to make better use of message failure callbacks as well, at minimum look in aldb sequences, should be catching and sending to failure

Does this add a lot of overhead in a big system?

When calling the check_aldb_delta function, call the success function immediately if the delta value has been checked recently

Also when calling the check_aldb_delta function, check to see if the scan function is already running, and then just append the success function onto that sequence.

I think we need to add a timeout to the triggers, and a failure callback when the timeout is hit.

Otherwise, the ALDB Sequence could stall

Right now when creating an InsteonDevice the flow resembles

1. on init set to generic
2. init calls functions.initialize_device which creates the groups and at least adds the root as group 0x01
3. init also calls send_handler.intialize_device which runs a bunch of inits steps and calls Insteon_Device.update_device
4. update_device uses the dev_cat data to re-init the various handlers to their proper type

This seems circuitous. Some of this is needed, for example we need a generic send and rcvd handler to do a dev_cat query. But if we could condense this down, I am sure we would avoid some gnarly errors.

A responder can only have one link to a controller. Removing the used ones from the list would likely make things easier to understand

These are unnecessary I think we can move everything into the groups

Start with single device level then global level.

Rather than passing a record location in the ALDB, we should return the actual object which itself has all the necessary methods

A few things to test:

• Do i2_modem_links still use group 0x00?
• What do device->modem links look like

Or should this just be added automatically whenever it is discovered?
Should there be some button to force a re-initialization? Maybe just a rescan button?

I don't know if this could reuse the existing state_machine or not, or maybe all state_machine entries should be a queue?

But currently sending multiple aldb_Record deletes causes them all to try and run at the same time, which results in generally only one working.

In an ideal world, subsequent calls would skip the check_aldb_delta call and perhaps other things like the set_msb and such.

The PLM keys are not fixed and so may change on a scan.

Some links such as unknown and undefined should be fine.

But duplicate modem links and possibly defined links are obscured

The organization is a mess, many of the functions are too complicated.