[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
Re: RES: [dsdp-mtj-dev] Extension point for defining devices - Back onTrack
|
Gustavo,
You say:
But to be honest, I only see this working if the whole SDK is
delivered
via p2 and placed in some "pre-defined" folder. Then it makes sense to
have to have a plugin that automatically says where the sdk is
installed
I have to disagree here. A very simple example of how this could work
on a Windows machine is via the registry. A standard Windows
installer could write information into the registry that the
registered ISDK could find and use. It would not require a path in
the extension point and it would not require P2 installation. The old
Siemens extension for EclipseME used to work this way... digging out
the information from the registry rather than the non-existent UEI
interface in that SDK.
It seems like we have UEI on the brain! <grin> There are lots of ways
that software can be installed and lots of ways that we can discover
that software. Limiting things to only be on a specific path seems
unnecessarily constraining to me.
Craig
On Jan 30, 2009, at 11:17 AM, Paula Gustavo-WGP010 wrote:
Hi craig,
I looked at JDT extension point. Roughly what is does is define a way
that eclipse can automatically find a JRE location (path on the file
system) and "import" it into JDT.
So if we transfer this concept to MTJ, the extension point will also
have some like:
* SDK Name
* SDK Path
* class that implements ISDK interface (similar to what you described
below)
Then mtj will automatically import all SDKs that are registered via
this
EP.
But to be honest, I only see this working if the whole SDK is
delivered
via p2 and placed in some "pre-defined" folder. Then it makes sense to
have to have a plugin that automatically says where the sdk is
installed
Motorola does something like that, but I think that it is the only sdk
provider that does it.
More thoughts???
:)
gep