com.sun.httpserver

[Ashton Hogan]

Hi Michael
Thank you for the below
1. I would be happy to but I would need some direction on some of the more low-level stuff. I've tried going through RFC's but didn't really get anything constructive out of it. I've also read several books on the topic but can't seem to find one that gives information on the binary payload structures and flows. If you could let me know of a good place to get the information I need then I'd be happy to contribute this. I have already studied the existing source quite a bit so I am fairly familiar with it.
2. I don't fully understand the purpose of the timer and looking at the code it looks as if you could prevent it by changing MAX_REQ and MAX_RES to not be hard-coded to -1 but I'm not entirely sure what the implications of changing that would be. Again, I would be happy to contribute this but I would need access to more background information if it's possible?

Thanks & RegardsAshton
      From: Michael McMahon <michael.x.mcmahon at oracle.com>
 To: "ashtonhogan at ymail.com" <ashtonhogan at ymail.com> 
Cc: net-dev at openjdk.java.net; Pavel Rappo <pavel.rappo at oracle.com>
 Sent: Tuesday, 28 March 2017, 18:09
 Subject: Re: com.sun.httpserver
   
  Hi Ashton,
 
 On 28/03/2017, 17:40, ashtonhogan at ymail.com wrote: 
Hi Michael
 
 Thanks for the below.
 
 1. Would it be possible to contribute this feature or is there a reason to not have it going forward?
 
 
 I think it would be useful. Maybe Pavel here in Oracle might comment on it since he has
 done work on a client implementation. It probably depends on how much additional API
 footprint it would create. Are you offering to do it?
 
 
2. Is it possible to disable the other threads and just have the listener? Smaller devices that don't have as many cores really struggle with multithreaded applications and if there are already threads in an application it may add additional overhead and complexity
 
 
 
 Yes, I think it would be possible to combine the timer functionality with the listener, and build the timing capability
 into the nio selector. Again that is more work, but if it is something you are offering to do..?
 
 Thanks,
 
 Michael.
 
 
Thanks & Regards
 Ashton 
 
 -------- Original Message --------
 Subject: Re: com.sun.httpserver
 From: Michael McMahon 
 To: ashtonhogan at ymail.com
 CC: net-dev at openjdk.java.net
 
 
 
Hi Ashton
 
 On 28/03/2017, 12:26, ashtonhogan at ymail.com wrote:
 > Hi all
 >
 > I was looking through the archives but did not find any answers to these questions so I figured I'd try here, hopefully this helps someone else as well.
 >
 > I've done some testing on the httpserver and come across some things:
 >
 > 1. No support for websockets (ws) or secure websockets (wss) - will this be in a future release?
 There are no plans at present to add support for websockets to this 
 implementation.
 > 2. When starting the server a number of threads are spawned regardless of whether you specify an executor or not. Was this the intention? If so then why does it spawn multiple threads when it could just spawn one for listening that places all HttpExchange instances into a BlockingQueue or something for the developer to use in a single or multithreaded environment?
 Yes, just looking at the implementation, there are 2-3 threads created. 
 One thread is the listener for incoming connections,
 plus one or two more for operating timer functionality. The executor is 
 only used for handling incoming requests.
 
 Regards,
 Michael.
 
 > Thanks& Regards
 > Ashton
 
  
 

   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/net-dev/attachments/20170328/0f40bb95/attachment.html>