> On Feb. 14, 2014, 12:07 a.m., Jie Yu wrote:
> > src/java/jni/org_apache_mesos_state_AbstractState.cpp, line 99
> > <https://reviews.apache.org/r/17686/diff/3/?file=470405#file470405line99>
> >
> >     What's the reason of returning false all the time? Is that weird if 
> > isCancelled() returns false and a CancellationException being thrown?
> 
> Benjamin Hindman wrote:
>     Java's cancellation semantics are weird IMHO, and I choose to return 
> false here to match them. Here's what we have to make sure:
>     
>     (1) If 'cancel' is called we should always return false since we've 
> started the computation and we don't know if we'll actually be able to cancel 
> it and we don't want too block to figure out if we should cancel it (we 
> should really only block in 'get').
>     
>     (2) After 'cancel' is called 'isCancelled' only has to return true if 
> 'cancel' returned true. Since we returned false in 'cancel', we can return 
> false here. Like (1), we don't want to block to figure out if it was actually 
> cancelled.
>     
>     (3) After 'cancel' is called 'isDone' always has to return true ... which 
> is totally crazy because it seems very realistic that 'cancel' returned false 
> because the computation is still running! Again, we're choosing not to block 
> in 'cancel' or anywhere else but only block in 'get' since in practice this 
> is probably where most Java people will expect a blocking operation.
>     
>     There aren't anything in the semantics that say getting true for 
> 'isCancelled' implies you can't get a CancellationException, but we could 
> imagine returning an ExecutionException instead of a CancellationException if 
> we really thought this was going to throw people off. Likewise, we could 
> imagine making 'isCancelled' check 'isDiscarded' but there is a race here so 
> I just decided to be deterministic.
>     
>     Please let me know if something doesn't make sense or you have a 
> different interpretation of the Java semantics.

> (2) ... Like (1), we don't want to block to figure out if it was actually 
> cancelled.

We don't need to block to figure out if it was actually cancelled. If 
future->isDiscarded(), we can return true without blocking. In the java spec, 
if cancel() returns false, isCancelled() is allowed to return true.

> Likewise, we could imagine making 'isCancelled' check 'isDiscarded' but there 
> is a race here so I just decided to be deterministic.

I don't understand why bother the race here. Seems that isCancelled() is 
totally useless in your code. What we are suggesting is: make isCancelled() a 
best effort check (without blocking).


- Jie


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/17686/#review34443
-----------------------------------------------------------


On Feb. 18, 2014, 8:46 a.m., Benjamin Hindman wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/17686/
> -----------------------------------------------------------
> 
> (Updated Feb. 18, 2014, 8:46 a.m.)
> 
> 
> Review request for mesos, Adam B, Ben Mahler, Ian Downes, Jie Yu, Niklas 
> Nielsen, TILL TOENSHOFF, Vinod Kone, and Jiang Yan Xu.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> See summary.
> 
> 
> Diffs
> -----
> 
>   src/java/jni/org_apache_mesos_state_AbstractState.cpp 
> 2ee0b1b631b80ec783e6bce683cdeaa77e56b2aa 
>   src/linux/cgroups.cpp 8ac25993886f5092fe6a58abdddbbb71e02911af 
>   src/log/catchup.cpp 59facbf9b5d065ba5066b6b443db4ee8e05ee33b 
>   src/log/consensus.cpp b89673a3b8f233e901eaf9ae69a9979099f4eb73 
>   src/log/log.cpp e83f822af86a2389e2b1abab9489713cb59838c2 
>   src/log/recover.cpp d06e5ada714ba0e5359ff1d3381edb6d526c6ad3 
>   src/master/detector.cpp 7e10433013b9415ec73c388e8dc69ab0989cdbc2 
>   src/master/registrar.cpp 915885a160f790399e8185c28c6e6555af1ee76e 
>   src/sasl/authenticatee.hpp f1a677f8aed0979f958e51f85e0a8210a03bd343 
>   src/sasl/authenticator.hpp 1478f6771b424555c34586a0d61f208dc15b0e7d 
>   src/slave/gc.hpp 328aa315d8ebd7ea5d05b57626ea4dbfc2206270 
>   src/slave/gc.cpp 405350bf8f498d2e59e9e6b4c4c19b7bdaa974de 
>   src/zookeeper/contender.cpp 6710da4e64fc0a43c1eabfc0f39fb0133c13df14 
>   src/zookeeper/group.cpp 793763e8312676f3b7cceb54bbad0337c8445ea7 
> 
> Diff: https://reviews.apache.org/r/17686/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Benjamin Hindman
> 
>

Reply via email to