Re: [geclipse-dev] Dissapearing workspace in grid project

[Ariel Garcia <garcia@xxxxxxxxxx>]

Hi,

> i got to reproduce the problem in which the grid project view stays
> completely empty when loading geclipse.
>
> Probably not the only way to trigger it, but bug
>     https://bugs.eclipse.org/bugs/show_bug.cgi?id=226712
> is triggering it. Namely, an empty BDII URI for one of the defined VOs
> is causing a NPE. It is not clear from the logs what happens afterwards
> but the grid project view stays empty.

i now debugged it, the problem is that "thanks" to the NPE the execution is 
jumping "up" in the stack to line 153  (catch (Throwable) {} block) of 
   ResourceGridContainer.fetchChildren(...)

(in the stacktrace of bug #226712 that would be: 
eu.geclipse.core.model.impl.ResourceGridContainer.fetchChildren(ResourceGridContainer.java:**153**)
at
eu.geclipse.core.model.impl.ResourceGridContainer.<init>(ResourceGridContainer.java:62)
at eu.geclipse.core.internal.model.GridRoot.<init>(GridRoot.java:56)
at eu.geclipse.core.internal.model.GridRoot.getInstance(GridRoot.java:71)
at eu.geclipse.core.model.GridModel.getRoot(GridModel.java:93)
)

when the VoManager is loading/creating the VOs, so the GridProject object 
doesn't get created at all. And actually we land back in 
fetchChildren(...)  but the loading was interrupted and the Grid Project 
view stays empty. I do not know what was exactly happening some time ago 
when the same results were observed, but a similar NPE might have been 
involved.

Of course  bug #226712 is easy to solve, but the question would be how can 
we make that core code more resilient to such problems in the middleware 
plugins, such that this doesn't happen again? Any thoughts?

Of course the recommendation for everybody is: always assume your input 
data (both from user, from disk, from network, from everywhere) is 
_wrong_  :-) and _check_ it.

Have a nice WE, Ariel