[tcf-dev] Services and the memory map

["Gherzan, Mircea" <mircea.gherzan@xxxxxxxxx>]

Hi Eugene,

Please, can you shed some light about the _proper_ use of the MemoryMap API? Here's the scenario: an user clicks a button in the Eclipse GUI, which results in the invocation of a custom TCF service command, with the final end-user result that some modules get automagically loaded in the Modules view for the selected context. The service command in the agent uses the context_() API to find some (PE/COFF, ELF) modules on the desired context. The resulted modules mappings should be editable by the user in the Eclipse GUI. My question, please: what is the "proper" TCF way to achieve this? I see two options:

A. The command reply provides an array of MemoryRegion objects, then some custom Eclipse GUI logic send a MemoryMap.set with them included.   
	Pro: pretty clean, no breaking of semantics/stacking
	Con: round-trip to the GUI and back

B. The service logic in the agent gets some "side-band" write access to the data container that "feeds" the context_get_memory_map() of the desired context, so the modules land in the "target" map.
	Pro: smaller round-trip (an event would still be needed though to tell the GUI client to "pull" the new mappings)
	Cons: 
		1. module entries not editable by the end-user because they are part of the "target" map. 
		2. breaking the API/stacking, as there's direct access from a service to a context-specific implementation detail
	 
Are there any other ways?

Thanks in advance,
- Mircea