Avoid analysis server crash when reanalyze invoked during analysis.

Avoid analysis server crash when reanalyze invoked during analysis.

BUG=dartbug.com/21448
R=brianwilkerson@google.com

Committed: https://code.google.com/p/dart/source/detail?r=41372

[Paul Berry, 2014-10-28 20:13:39]

paulberry@google.com changed reviewers:
+ brianwilkerson@google.com

[Brian Wilkerson, 2014-10-28 20:37:14]

DBC

https://codereview.chromium.org/678413003/diff/20001/pkg/analysis_server/lib/...
File pkg/analysis_server/lib/src/analysis_server.dart (right):

https://codereview.chromium.org/678413003/diff/20001/pkg/analysis_server/lib/...
pkg/analysis_server/lib/src/analysis_server.dart:527: operationLoopRunning =
false;
If an exception occurs in sendStatusNotification, won't this make it impossible
for the server to ever perform another operation?

I think it's safer to change the flag to mean that "there is a pending future
which will execute [performOperation]".

[Paul Berry, 2014-10-28 21:03:49]

On 2014/10/28 20:37:14, Brian Wilkerson wrote:
> DBC
> 
>
https://codereview.chromium.org/678413003/diff/20001/pkg/analysis_server/lib/...
> File pkg/analysis_server/lib/src/analysis_server.dart (right):
> 
>
https://codereview.chromium.org/678413003/diff/20001/pkg/analysis_server/lib/...
> pkg/analysis_server/lib/src/analysis_server.dart:527: operationLoopRunning =
> false;
> If an exception occurs in sendStatusNotification, won't this make it
impossible
> for the server to ever perform another operation?
> 
> I think it's safer to change the flag to mean that "there is a pending future
> which will execute [performOperation]".

It was already impossible for the server to ever perform another operation in
that circumstance, since the call to sendStatusNotification is outside the body
of a try block, and therefore an exception occurring at this point will crash
the server.

However, on further reflection, I think you're right that the code is clearer if
we change the flag to mean that there is a pending call to performOperation. 
PTAL.

[Brian Wilkerson, 2014-10-28 21:08:38]

LGTM

[Paul Berry, 2014-10-28 21:16:05]

Committed patchset #3 (id:40001) manually as 41372 (presubmit successful).