Hi Stephen,

On 02/06/11 20:56, Stephen Warren wrote:

Liam,

UCM syntax is e.g.:

SectionModifier."Capture Voice".0 {

This maps to a modifier->name of "Capture Voice.0".

However, when find_modifier searches for modifiers, it strips off the index (".0") from modifier->name, and just compares base name. As such, the index doesn't seem useful.

Your right, the index is probably not very useful for modifiers atm.

I propose:

a) In parse_modifier, fail if index!=0.

b) In parse_modifier, store just the name in modifier->name.

c) In find_modifier, remove all the special handling of "." in the name, and just strcmp(modifier->name, modifier_name).

Perhaps the index should be completely removed from the file syntax too?

Yeah, I would go with that for modifier.

If an index was actually useful in any case, the author of the file could simply add it within the name string, i.e.:

typical: SectionModifier."Capture Voice" {

unusual: SectionModifier."Capture Voice 0" {

Of course, that wouldn't be backwards-compatible file syntax. Perhaps we could make the index optional, but always validate it was 0 if present, as I said above.

we do have some users using the modifier.0 syntax so maybe best to deprecate it.

I actually wonder if the index on the SectionDevice makes any sense either; functions like is_modifier_supported would be a lot simpler if devices simply never had an index, such that is_modifier_supported was a simple strcmp too. The worst fallout from that change might be a requirement to list more entries in SupportedDevice/ConflictingDevice.

ok.

Second, when find_modifier is called from get_value, it'll fail with a valid modifier name that happens not to be supported on any currently active devices. Is that intended? I'd expect to be able to retrieve values from any/all devices and modifiers irrespective of whether those devices or modifiers could actually be activated given the currently active configuration.

I propose: Adding a parameter to find_modifier to indicate whether to call is_modifier_supported or not, this new parameter being false for get_value and true in other cases.

This issue will also be relevant to find_device when I modify that to call new function is_device_supported.

For find_device, we may also have to add a parameter for a device name that gets ignore when checking SupportedDevice/ConflictingDevice, which can be passed when switching devices.

Thanks for your thoughts.

Sounds reasonable to me.

Liam