[news.eclipse.platform] Re: How do I disable my own plugin at start up t

[Stefan Langer <mailtolanger@xxxxxxxxxxxxxx>]

Throwing an exception during startup is a definite way of stopping a 
bundle. But this won't work once your bundle has been started.
Calling stop of your bundle will shut your bundle down but I'm not sure 
it will work if you have a perspective or view or some resource from 
your plugin currently open/inuse. (You'll have to try to see what 
happend :) )

Some things to consider:
If checking for proper operation takes long then start is not the right 
point of doing it. Start should simply startup the bundle and not take 
too long or else the framework might think your bundle crashed and 
simply skip it. A better way is to use a proxy that will check for 
proper operation when the plugins functionality is first accessed.

If this error state you are talking about can be resolved while the 
plugin is running I would suggest to use a custom error flag and report 
upon that as long as things are not in proper state. Once they reach 
proper state simply reset the error flag and the user can continue 
working. IMHO This is a much more user friendly approach.
This also encompasses that your views, perspectives and editors know how 
to handle this error state.

Regards
Stefan

Farokh Morshed schrieb:
> There is another similar case to this:
>
> While our plugin is being used, it may discover that something 
> catastrophic has happened.  For example a server that it constantly 
> talks to has closed its socket  connection (more than likely user shut 
> down the server).
>
> What's the best way to "disable" our plugin now?  Is there an API to 
> call that effectively calls the plugin's stop method?  Should I just 
> call the stop method?  Of course, I can  pop up a message and say that 
> something has gone wrong.  The question is, what should I do after 
> that.  I suppose I can set a flag that every point of use (of the 
> plugin) can check and pop up an error message, etc.
>
> I can try all this, but if someone has solved this problem before, I 
> would appreciate a heads up.