Hi Michal, On Tue, 2020-01-14 at 21:09 +0100, michal.lowas-rzechonek@xxxxxxxxxxx wrote: > Hi Inga, > > On 01/14, Stotland, Inga wrote: > > I wonder if it would be better to re-use "-i" option by changing it's > > meaning form "index" to "i/o". > > > > So that " -i hci<#>" will map to generic i/o on a specified controller > > and no "-i" option means any controller. > > > > Yes, we will loose some uniformity across all of the bluez in a sense > > that "-i <#>" won't work, but imo it's preferable to having two options > > with inter-dependecies. > > Hm, might be... The reason I added the "--io=<type>:<options>" was the > "<options>" part. > > For example, we have a non-HCI radio adapter that uses > "--io=uart:/dev/tty<n>" syntax, or remote radio using > "--io=tcp:<host>:<port>" syntax. > > So maybe I could parse "-i <n>" and "-i hci<n>" (where <n> is a > number) as a shortcut for "-i generic:hci<n>"? > > That would allow us to merge both options and rename "--index" to > "--io". > Yes, I'd be fine with this kind of change. Depending on the type, we would know how to parse the options. With the fallback to legacy cases: * no option given: generic, any controller * "-i <hci#>" or -i <#> : generic with the specified controller Thank you, Inga
