jndn

06/09/2015

03:39 PM Bug #2869: NFD Restart Breaks Face Socket

Hi Andrew. Your option number 2 looks like feature request #2075 for the library to inform the application when the c... Anonymous

03:06 PM Bug #2869 (Closed): NFD Restart Breaks Face Socket

If a Face is connected to a remote NFD using TcpTransport, an NFD restart will break the TCP socket. If the default c... Andrew Brown

06/08/2015

03:34 PM Task #2860 (Closed): expressInterest should queue interests while async I/O is connecting

If an application calls expressInterest several times, the library needs to connect on the first call. With blocking ... Anonymous

09:30 AM Feature #2565: Implement async operations for Face.processEvents

Yes, makes sense; like we talked about on the phone, some CPU usage benchmarks between the AsyncSocketChannel and the... Andrew Brown

06/05/2015

12:31 PM Feature #2565: Implement async operations for Face.processEvents

I want to keep the base class agnostic about which (if any) thread pool implementation is being used. So I don't want... Anonymous

10:03 AM Feature #2565: Implement async operations for Face.processEvents

From slide 6 and slide 27 at http://openjdk.java.net/projects/nio/presentations/TS-4222.pdf: I read that to mean that... Andrew Brown

06/04/2015

11:39 AM Feature #2565: Implement async operations for Face.processEvents

- you have the setIsRemoved() on the PIT entries; have you thought of just setting that and not removing the entry im... Anonymous

06/03/2015

04:35 PM Feature #2565: Implement async operations for Face.processEvents

Yes, I believe you are right: if I understand the commit correctly, you use Collections.synchronizedList so any add()... Andrew Brown

12:39 PM Feature #2565: Implement async operations for Face.processEvents

Hi Andrew. A sanity check: In Python we have a Face subclass to dispatch pretty much every method call to the single ... Anonymous

12:28 PM Feature #2565: Implement async operations for Face.processEvents

Here's the commit to use synchronizedList for pendingInterestTable, etc. I tried to be careful not to call the callba... Anonymous

12:34 PM Bug #2846 (Closed): selfSign certificate name timestamp should be a version component

Anonymous

06/02/2015

03:56 PM Bug #2847 (Closed): Created cert name should use lower-case ksk

Merged changes to master in all libraries. Anonymous

03:40 PM Bug #2847 (Closed): Created cert name should use lower-case ksk

IdentityStorage.getNewKeyName adds a name component starting with "KSK-". But it should be "ksk-" as required by the ... Anonymous

09:28 AM Feature #2565: Implement async operations for Face.processEvents

Yes, still in progress. Branch issue/2565-async-IO<br>
https://github.com/named-data/jndn/tree/issue/2565-async-IO
Anonymous

09:16 AM Feature #2565: Implement async operations for Face.processEvents

Sounds good; are you working in a specific branch? Andrew Brown

06/01/2015

01:58 PM Bug #2846 (Resolved): selfSign certificate name timestamp should be a version component

Merged changes to master. Anonymous

01:05 PM Bug #2846 (Closed): selfSign certificate name timestamp should be a version component

When IdentityManager.selfSign creates the certificate name from the keyName, it just appends a name component with th... Anonymous

11:13 AM Feature #2565: Implement async operations for Face.processEvents

Hi Andrew. For starters, I'll switch the data structures like pendingInterestTable_ and interestFilterTable_ to use s... Anonymous

05/27/2015

11:05 AM Feature #2565: Implement async operations for Face.processEvents

Ha, I hear you; the benefits of a single managed event loop are awesome, even if it does make JavaScript single-threa... Andrew Brown

10:30 AM Feature #2565: Implement async operations for Face.processEvents

.. or is the idea that code should use both: dispatch to "some thread" in a thread pool where that thread does a bloc... Anonymous

10:28 AM Feature #2565: Implement async operations for Face.processEvents

The problem with synchronized is that a call to a synchronized method is effectively a blocking call (until the other... Anonymous

10:14 AM Feature #2565: Implement async operations for Face.processEvents

I mean speed like how many more bytes could the async version process if the network channel is always saturated with... Andrew Brown

09:57 AM Feature #2565: Implement async operations for Face.processEvents

Hi Andrew. Thanks for the mention of [ScheduledExecutorService](https://docs.oracle.com/javase/7/docs/api/java/util/c... Anonymous

09:55 AM Feature #2565: Implement async operations for Face.processEvents

My idea for real-world CPU benchmarks of the event loop is to run on a Raspberry Pi and look at uptime. Hopefully, t... Anonymous

05/26/2015

04:20 PM Feature #2565: Implement async operations for Face.processEvents

Makes sense; I wonder if there is a good way to benchmark the before and after for both speed and CPU time. Andrew Brown

04:09 PM Feature #2565: Implement async operations for Face.processEvents

Yes, the application will provide the thread pool to the library.
About the need for NIO, the concern is not for s... Anonymous

03:52 PM Feature #2565: Implement async operations for Face.processEvents

Yeah, I implemented my event loop with https://docs.oracle.com/javase/7/docs/api/java/util/concurrent/Executors.html#... Andrew Brown

01:28 PM Feature #2565: Implement async operations for Face.processEvents

Yes, Java nio seems to have a different approach from Python, JavaScript and Boost. In those the library provides an ... Anonymous

01:11 PM Feature #2565: Implement async operations for Face.processEvents

I hadn't seen that one; I guess I thought the point of using NIO for jndn was to avoid the `while(true){ face.process... Andrew Brown

11:38 AM Feature #2565: Implement async operations for Face.processEvents

Hi Andrew. You say "the two NIO classes would be ...." but don't we need to use the async versions such as https://do... Anonymous

05/19/2015

11:53 AM Feature #2565: Implement async operations for Face.processEvents

OK, thanks. I'll try to use a scheduled event for the interest timeout. Anonymous

11:50 AM Feature #2565: Implement async operations for Face.processEvents

This "loop" (actually it is event rescheduled every 100ms) is just a legacy part that is in pipeline to be replaced t... Alex Afanasyev

10:26 AM Feature #2565: Implement async operations for Face.processEvents

Hi Alex. For the interest timeout it is possible to schedule an event to fire at the time of expiration. But ndn-cxx ... Anonymous

05/13/2015

10:50 AM Feature #2565: Implement async operations for Face.processEvents

Exceptions that are not explicitly handled within callback are in most cases critical. NFD/NLSR and some other apps ... Alex Afanasyev

09:46 AM Feature #2565: Implement async operations for Face.processEvents

Thanks for the reference in the NIO classes.
At least in Java, the application is already forced to handle I/O exc... Anonymous

09:33 AM Feature #2565: Implement async operations for Face.processEvents

And just for reference, the two NIO classes would be https://docs.oracle.com/javase/7/docs/api/java/nio/channels/Data... Andrew Brown

09:31 AM Feature #2565: Implement async operations for Face.processEvents

Jeff, what's your opinion on having some type of callback to handle those exceptions? That way I can avoid having a t... Andrew Brown

05/12/2015

11:56 AM Feature #2565: Implement async operations for Face.processEvents

For example, it doesn't look like ndn-cxx catches exceptions in the async I/O callback. If onInterest throws an excep... Anonymous

09:31 AM Feature #2565: Implement async operations for Face.processEvents

If the application's onInterest or onData callback throws an exception (in the nio thread), where would you expect th... Anonymous