beepcore-java-users Mailing List for Java BEEP Core (Page 14)
Status: Beta
Brought to you by:
huston
Menu
▾
▴
beepcore-java-users — Discussion list for developers using beepcore-java library
You can subscribe to this list here.
| 2001 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
(3) |
Dec
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2002 |
Jan
(24) |
Feb
(3) |
Mar
(18) |
Apr
(2) |
May
(11) |
Jun
(6) |
Jul
(11) |
Aug
(37) |
Sep
(22) |
Oct
(11) |
Nov
(11) |
Dec
(29) |
| 2003 |
Jan
(8) |
Feb
(4) |
Mar
(19) |
Apr
(13) |
May
(16) |
Jun
(15) |
Jul
(2) |
Aug
(1) |
Sep
(1) |
Oct
(7) |
Nov
(13) |
Dec
|
| 2004 |
Jan
(1) |
Feb
(4) |
Mar
(2) |
Apr
|
May
(3) |
Jun
(2) |
Jul
(8) |
Aug
|
Sep
(7) |
Oct
(15) |
Nov
(8) |
Dec
|
| 2005 |
Jan
|
Feb
|
Mar
|
Apr
(1) |
May
(1) |
Jun
|
Jul
(6) |
Aug
|
Sep
(3) |
Oct
|
Nov
|
Dec
(5) |
| 2006 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
(1) |
Jul
(4) |
Aug
|
Sep
|
Oct
(4) |
Nov
|
Dec
|
| 2007 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
(1) |
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|
From: Huston <hu...@us...> - 2002-03-08 16:32:34
|
> I'd be interested in seeing how it was > conducted. There is some BEEP overhead but depending on the message sizes > and what not I would think it varies significantly. True. If you are sending small frames the overhead of sending the beep header and trailer (at least 19 bytes) can be significant. > It should be a priority > to categorize the performance of BEEP (is that done or being done already?). Not that I'm aware of. --Huston |
|
From: Paul H. <Pau...@ya...> - 2002-03-03 22:00:43
|
Iain, Folks, Many have asked for how to repeat our tests. Here goes... See the README at http://cvs.apache.org/viewcvs/jakarta-commons-sandbox/altrmi/ for general info on the other benchmarks. The ant targets that do the business are "socketa-serve" & "socketa-client". The addition (two dos shells) instrctions to invoke BEEP are : In DOS shell #1 - "ant -buildfile tests2.xml beep-serve" In DOS shell #2 - "ant -buildfile tests2.xml beep-client" (note the different build file). It is more complex that you folks might hope for, but it is not really about testing beep, more of using it in a mature project. AltRMI has no downloadable listing, it is best got from CVS (not hard for persons assembed here?). See instructions : http://jakarta.apache.org/site/cvsindex.html - Paul >>I have run a test of BEEP usage. Basically the speed is about 10% of >>that of the same test without BEEP as a layer. I was wondering if that >>was expected. >> >>The test is a client and server in different VMs, with BEEP handling >>streaming over sockets. The test uses no multiplexing features. >> > >Hi, > >I'm new to BEEP but thouht I'd throw in my 2 cents... > >Could you post the test somewhere? I'd be interested in seeing how it was >conducted. There is some BEEP overhead but depending on the message sizes >and what not I would think it varies significantly. It should be a priority >to categorize the performance of BEEP (is that done or being done already?). > >-iain > > >_________________________________________________________ >Do You Yahoo!? >Get your free @yahoo.com address at http://mail.yahoo.com > > >_______________________________________________ >Beepcore-java-users mailing list >Bee...@li... >https://lists.sourceforge.net/lists/listinfo/beepcore-java-users > |
|
From: Iain S. <iai...@ya...> - 2002-03-01 21:30:35
|
On 2/28/02 5:21 AM, "Paul Hammant" <Pau...@ya...> wrote: > I have run a test of BEEP usage. Basically the speed is about 10% of > that of the same test without BEEP as a layer. I was wondering if that > was expected. > > The test is a client and server in different VMs, with BEEP handling > streaming over sockets. The test uses no multiplexing features. Hi, I'm new to BEEP but thouht I'd throw in my 2 cents... Could you post the test somewhere? I'd be interested in seeing how it was conducted. There is some BEEP overhead but depending on the message sizes and what not I would think it varies significantly. It should be a priority to categorize the performance of BEEP (is that done or being done already?). -iain _________________________________________________________ Do You Yahoo!? Get your free @yahoo.com address at http://mail.yahoo.com |
|
From: Paul H. <Pau...@ya...> - 2002-02-28 13:20:42
|
Hi I have run a test of BEEP usage. Basically the speed is about 10% of that of the same test without BEEP as a layer. I was wondering if that was expected. The test is a client and server in different VMs, with BEEP handling streaming over sockets. The test uses no multiplexing features. Regards, - Paul |
|
From: Huston <hu...@us...> - 2002-02-12 18:08:44
|
> My question is how do I go about creating a proper message that mimics the > example payload above? You should be able to create the appropriate OutputDataStream (probably StringOutputDataStream), set the Content-Type header, and add the rest as the data. > I know I can use InputStreamDataStream to add custom > headers to the message, I would use the latest code in cvs for this type of work. The DataStreams have been split into Input and Output DataStreams which cleared up a lot of confusion in the library. > but the fact that there are multiple Content-Type and > Content-Transfer-Encoding headers and boundaries is confusing me quite a bit. The thing to remember is that the java library only appends the mime headers to or removes them from the rest of the data. It doesn't actually change behavior based on any of the mime headers, that is left to the application. --Huston |
|
From: Michael J. R. <mjr...@ea...> - 2002-02-12 17:41:08
|
Hi,
I have a question or two about formatting BEEP messages. I'm doing an
undergraduate project at Temple University implementing IMPP through APEX and
the APEX Internet Drafts describe a bunch of different ways to map BEEP
payloads, including:
C: MSG 1 1 . 42 1234
C: Content-Type: multipart/related; boundary="boundary";
C: start="<1...@ex...>";
C: type="application/beep+xml"
C:
C: --boundary
C: Content-Type: application/beep+xml
C: Content-ID: <1...@ex...>
C:
C: <data content='cid:2...@ex...'>
C: <originator identity='fr...@ex...' />
C: <recipient identity='ba...@ex...' />
C: </data>
C: --boundary
C: Content-Type: image/gif
C: Content-Transfer-Encoding: binary
C: Content-ID: <2...@ex...>
C:
C: ...
C: --boundary--
C: END
My question is how do I go about creating a proper message that mimics the
example payload above? I know I can use InputStreamDataStream to add custom
headers to the message, but the fact that there are multiple Content-Type and
Content-Transfer-Encoding headers and boundaries is confusing me quite a bit.
I understand the purpose of this type of BEEP message, I'm just a bit
confused on how to go about creating the message itself. Any suggestions or
examples on how to go about this would be greatly appreciated. Thanks for
your time.
-Mike Riggio
|
|
From: Huston <hu...@us...> - 2002-01-25 12:28:34
|
> SOmeone else mentioned the problem with beecore-java stalling > on large RPY's. My hunch, unconfirmed, is that the threading > model is more in flux than first suspoected, as evidenced by the > Echo profile needing to be reverted. There haven't been any changes to the threading model added to cvs since the last release, I only started the threading changes in late December so from the dates in cvs you can verify this. I found a bug in the InputDataStream while working on the threading changes. The bug was significant enough I wanted to add it before the new threading model was finished. I just made a mistake with EchoProfile.java and checked in the wrong version. > With the Echo profile handling it's own thread management and the > main thread free to read on teh socket and process SEQ's it seems > to be working fine. I'd suggest having the profile start create > it's own worker thread to process the receiveMSG calls (hand them > off with a queue or something) and I bet the stalling bug goes away. This has been the requirement since the first release of beepcore-java. The new threading model I am working on will remove this requirement. --Huston |
|
From: William J. M. <wm...@ol...> - 2002-01-25 05:43:41
|
SOmeone else mentioned the problem with beecore-java stalling on large RPY's. My hunch, unconfirmed, is that the threading model is more in flux than first suspoected, as evidenced by the Echo profile needing to be reverted. With the Echo profile handling it's own thread management and the main thread free to read on teh socket and process SEQ's it seems to be working fine. I'd suggest having the profile start create it's own worker thread to process the receiveMSG calls (hand them off with a queue or something) and I bet the stalling bug goes away. Don't have time at the moment to go find the name of the person who was also having this problem. -bill |
|
From: William J. M. <wm...@ol...> - 2002-01-24 00:42:01
|
Was there a label set prior to the start of the threading changes? Is there going to be one? -bill On Wed, Jan 23, 2002 at 11:46:48AM -0700, Huston wrote: > Bill, > > Sorry, my mistake. I checked in the wrong version of the Echo profile (the > one in cvs is for the new threading model). Try the new version I just > checked in. > > > I thought the socket IO had it's own thread rather than running > > in the main app thread for Beepd? > > The current version in cvs has an thread for each socket but the problem is > that the profile callbacks happen on that thread too so the profile can > stall the library. In the new threading model this is fixed. I haven't > checked the new threading code it yet because I haven't had time to do the > final cleanup, if you would like a copy I can send you the code (as I recall > only Channel.java changed). > > --Huston > > > _______________________________________________ > Beepcore-java-users mailing list > Bee...@li... > https://lists.sourceforge.net/lists/listinfo/beepcore-java-users |
|
From: Huston <hu...@us...> - 2002-01-23 18:47:34
|
Bill, Sorry, my mistake. I checked in the wrong version of the Echo profile (the one in cvs is for the new threading model). Try the new version I just checked in. > I thought the socket IO had it's own thread rather than running > in the main app thread for Beepd? The current version in cvs has an thread for each socket but the problem is that the profile callbacks happen on that thread too so the profile can stall the library. In the new threading model this is fixed. I haven't checked the new threading code it yet because I haven't had time to do the final cleanup, if you would like a copy I can send you the code (as I recall only Channel.java changed). --Huston |
|
From: Gabe W. <gw...@wa...> - 2002-01-23 18:02:17
|
+1 that it should be exposed. On Wed, 23 Jan 2002, William J. Mills wrote: > Would there be any interest in, or excellent reason not to, expose > an interface to allow the frame size of the session to be controlled by the > application? > > -bill > > _______________________________________________ > Beepcore-java-users mailing list > Bee...@li... > https://lists.sourceforge.net/lists/listinfo/beepcore-java-users > -- Gabe Wachob gw...@wa... Personal http://www.wachob.com CTO, WiredObjects http://www.wiredobjects.com |
|
From: William J. M. <wm...@ol...> - 2002-01-23 07:54:59
|
Would there be any interest in, or excellent reason not to, expose an interface to allow the frame size of the session to be controlled by the application? -bill |
|
From: William J. M. <wm...@ol...> - 2002-01-23 07:20:44
|
Ah, this is an empty Mime header... On Wed, Jan 23, 2002 at 01:05:05AM -0600, William J. Mills wrote: > In the following why is ther an open line added to the > Bing payload? > > In continuation frames the open line is not present... > > -bill > > ---------------------------------------------------- > > Interesting case. -size optin was 26. > ---------------------------------------------------- > RPY 0 0 . 0 59 > Content-Type: application/beep+xml > > <greeting></greeting>END > SEQ 0 119 4096 > MSG 0 1 . 59 125 > Content-Type: application/beep+xml > > <start number='1'><profile uri='http://xml.resource.org/profiles/NULL/ECHO' /></start>END > SEQ 0 217 4096 > MSG 1 1 . 0 28 > > abcdefghijklmnopqrstuvwxyzEND > > > ---------------------------------------------------- > > > Case with continuation frames... > ---------------------------------------------------- > > > RPY 0 0 . 0 59 > Content-Type: application/beep+xml > > <greeting></greeting>END > SEQ 0 119 4096 > MSG 0 1 . 59 125 > Content-Type: application/beep+xml > > <start number='1'><profile uri='http://xml.resource.org/profiles/NULL/ECHO' /></start>END > SEQ 0 217 4096 > MSG 1 1 * 0 256 > > abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstEND > MSG 1 1 * 256 256 > uvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopEND > MSG 1 1 . 512 2 > qrEND > > > > _______________________________________________ > Beepcore-java-users mailing list > Bee...@li... > https://lists.sourceforge.net/lists/listinfo/beepcore-java-users |
|
From: William J. M. <wm...@ol...> - 2002-01-23 07:05:10
|
In the following why is ther an open line added to the Bing payload? In continuation frames the open line is not present... -bill ---------------------------------------------------- Interesting case. -size optin was 26. ---------------------------------------------------- RPY 0 0 . 0 59 Content-Type: application/beep+xml <greeting></greeting>END SEQ 0 119 4096 MSG 0 1 . 59 125 Content-Type: application/beep+xml <start number='1'><profile uri='http://xml.resource.org/profiles/NULL/ECHO' /></start>END SEQ 0 217 4096 MSG 1 1 . 0 28 abcdefghijklmnopqrstuvwxyzEND ---------------------------------------------------- Case with continuation frames... ---------------------------------------------------- RPY 0 0 . 0 59 Content-Type: application/beep+xml <greeting></greeting>END SEQ 0 119 4096 MSG 0 1 . 59 125 Content-Type: application/beep+xml <start number='1'><profile uri='http://xml.resource.org/profiles/NULL/ECHO' /></start>END SEQ 0 217 4096 MSG 1 1 * 0 256 abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstEND MSG 1 1 * 256 256 uvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopEND MSG 1 1 . 512 2 qrEND |
|
From: William J. M. <wm...@ol...> - 2002-01-23 01:58:23
|
Nevermind, answered my own question.. The behavior is gotten when you reply to teh same message again.... This might actually want to cause an exception? The message alreayd has a reply sent and is not the next one to be processessed anymmore? -bill On Tue, Jan 22, 2002 at 07:20:55PM -0600, William J. Mills wrote: > I am seeinf the following when running a slightly modified > Echo profile on Beepd. > > SEQ 1 201 4096 > RPY 1 1 . 0 201 > > abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqEND > RPY 1 1 . 201 0 > END > > > Note the empty extra reply frame for the same message. This > is being generated, as far as I can tell, from a single call to > sendRPY(). > > Any hints? > > -bill > > > _______________________________________________ > Beepcore-java-users mailing list > Bee...@li... > https://lists.sourceforge.net/lists/listinfo/beepcore-java-users |
|
From: Michael J. R. <mjr...@ea...> - 2002-01-23 01:31:08
|
Hi Bill,
Thanks for the quick response, I really appreciate it. I'm not sure if
you completely understood what I meant about the seperate profiles. If both
a client app and a server app created channels based on profile X, would it
be bad if each app had a serarate version of profile X that had specific code
in it to deal with either a client app or a server app? Perhaps that was
your original interpretation of my last e-mail?
Also, concerning the message notification. I've been going through the
API documenation for a few days now and couldn't find anything that would
allow the profile to notify the main application when a message is received.
Is there something in the API that will allow me to do this that I have
overlooked?
Thanks again for your time and consideration.
-Mike Riggio
On Tuesday 22 January 2002 08:13 pm, you wrote:
> Mike,
>
> Beepd as written already p[rocessed the message to send it to
> the channel to be processed with receiveMSG, it's very generic
> and won't likely want to deal with the message further. The
> receiveMSG method may well notify some other part of the app
> about a message being recieved, especially in an app where
> you are using the BEEP library that is not beepd. A profile
> may well interact with lots of other stuff within your program,
> and several profiles could for example all update the same
> user interface display window.
>
> An example of sep[arate implementations of the ECHO profile can
> be seen in the Echorofile inplemenation and the BING program.
> Bing actually talks to the ECHO profile without using any of the
> code in the EchoProfile itself. EchoProfile in this case is
> really a server profile, and the Bing client is a specific
> client.
>
> Hope this helps,
>
> -bill mills
>
> On Tue, Jan 22, 2002 at 08:07:13PM -0500, Michael J. Riggio wrote:
> > Hi,
> > I'm doing undergraduate research at Temple University and have a
> > simple question or two regarding message data from the receiveMSG method.
> > In the example EchoProfile, when the channel receives a message the Echo
> > Profile simply sends the same message back through the channel. However,
> > what if, for example, the Beepd program wanted to know when the channel
> > receives a message? How can the echo profile notify the Beepd program
> > that a message was received AND, if needed, allow the Beepd program to
> > see the new message? A second question: should a particular profile be
> > implementation specific? In other words, as long as the profile acts
> > according to the Internet Drafts, etc, does it really matter if there is
> > code that pertains to a client app, or a server app, or MY application
> > that uses the profile? Can there be separate version of a profile, maybe
> > one for clients to use and one for servers to use, but both version act
> > according to the internet drafts?
> > Thanks for your time and consideration, I really appreciate any
> > help or suggestions you may have. Sorry in advance if I misunderstood
> > any "key" concepts or API functionality.
> >
> > -Mike Riggio
> > Temple University Netlab
> >
> > _______________________________________________
> > Beepcore-java-users mailing list
> > Bee...@li...
> > https://lists.sourceforge.net/lists/listinfo/beepcore-java-users
-------------------------------------------------------
|
|
From: William J. M. <wm...@ol...> - 2002-01-23 01:29:54
|
Huston, Rather than have the conversation via sourceforge bug comments... I am using the current tip of the CVS tree and java 1.4beta3. You have never seen this starvation problem sending a large message? -bill |
|
From: William J. M. <wm...@ol...> - 2002-01-23 01:20:58
|
I am seeinf the following when running a slightly modified
Echo profile on Beepd.
SEQ 1 201 4096
RPY 1 1 . 0 201
abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqEND
RPY 1 1 . 201 0
END
Note the empty extra reply frame for the same message. This
is being generated, as far as I can tell, from a single call to
sendRPY().
Any hints?
-bill
|
|
From: William J. M. <wm...@ol...> - 2002-01-23 01:17:07
|
Heya, debugging something in beepcore-java and find that the current ti[ pf the tree will deadlock on the Echo profile if the payload is larger than one frame, since the receiveMSG method wait()'s on the message to complete, but it has grabbed the thread so no more data is read in core. Has this been seen before? I thought the socket IO had it's own thread rather than running in the main app thread for Beepd? -bill |
|
From: William J. M. <wm...@ol...> - 2002-01-23 01:13:16
|
Mike, Beepd as written already p[rocessed the message to send it to the channel to be processed with receiveMSG, it's very generic and won't likely want to deal with the message further. The receiveMSG method may well notify some other part of the app about a message being recieved, especially in an app where you are using the BEEP library that is not beepd. A profile may well interact with lots of other stuff within your program, and several profiles could for example all update the same user interface display window. An example of sep[arate implementations of the ECHO profile can be seen in the Echorofile inplemenation and the BING program. Bing actually talks to the ECHO profile without using any of the code in the EchoProfile itself. EchoProfile in this case is really a server profile, and the Bing client is a specific client. Hope this helps, -bill mills On Tue, Jan 22, 2002 at 08:07:13PM -0500, Michael J. Riggio wrote: > Hi, > I'm doing undergraduate research at Temple University and have a simple > question or two regarding message data from the receiveMSG method. In the > example EchoProfile, when the channel receives a message the Echo Profile > simply sends the same message back through the channel. However, what if, > for example, the Beepd program wanted to know when the channel receives a > message? How can the echo profile notify the Beepd program that a message > was received AND, if needed, allow the Beepd program to see the new message? > A second question: should a particular profile be implementation specific? > In other words, as long as the profile acts according to the Internet Drafts, > etc, does it really matter if there is code that pertains to a client app, or > a server app, or MY application that uses the profile? Can there be separate > version of a profile, maybe one for clients to use and one for servers to > use, but both version act according to the internet drafts? > Thanks for your time and consideration, I really appreciate any help or > suggestions you may have. Sorry in advance if I misunderstood any "key" > concepts or API functionality. > > -Mike Riggio > Temple University Netlab > > _______________________________________________ > Beepcore-java-users mailing list > Bee...@li... > https://lists.sourceforge.net/lists/listinfo/beepcore-java-users |
|
From: Michael J. R. <mjr...@ea...> - 2002-01-23 01:03:48
|
Hi,
I'm doing undergraduate research at Temple University and have a simple
question or two regarding message data from the receiveMSG method. In the
example EchoProfile, when the channel receives a message the Echo Profile
simply sends the same message back through the channel. However, what if,
for example, the Beepd program wanted to know when the channel receives a
message? How can the echo profile notify the Beepd program that a message
was received AND, if needed, allow the Beepd program to see the new message?
A second question: should a particular profile be implementation specific?
In other words, as long as the profile acts according to the Internet Drafts,
etc, does it really matter if there is code that pertains to a client app, or
a server app, or MY application that uses the profile? Can there be separate
version of a profile, maybe one for clients to use and one for servers to
use, but both version act according to the internet drafts?
Thanks for your time and consideration, I really appreciate any help or
suggestions you may have. Sorry in advance if I misunderstood any "key"
concepts or API functionality.
-Mike Riggio
Temple University Netlab
|
|
From: Huston <hu...@us...> - 2002-01-22 03:53:07
|
Bill, > I f you run Bing with a -size of 1500 you get normal operation? Yes, I have tried sizes from 10 bytes to 1,000,000 bytes (including 1500 bytes)with no problems. Check to make sure to are working with the head of the cvs tree because I did check in a fix the middle of last week. > Looks like the second frame comes in and the getNextReply is > hanging up in the wait, not sure why, bit h frames appear to > have been recieved with recieveRPY ... this is the big one at the > moment. This sounds like you are not using the API correctly. Could you send me an example of your code that I could try? > Do not know whether the fact that we are using java 1.4 is > significant. I use both 1.4 and 1.2.2 in my testing so that should be causing a problem. --Huston |
|
From: William J. M. <wm...@ol...> - 2002-01-22 01:51:36
|
Huston, I f you run Bing with a -size of 1500 you get normal operation? This is what I am using for testing on right now on the bug I am trying to find. Different one form the stallign bug on send. Looks like the second frame comes in and the getNextReply is hanging up in the wait, not sure why, bit h frames appear to have been recieved with recieveRPY ... this is the big one at the moment. Do not know whether the fact that we are using java 1.4 is significant. java version "1.4.0-beta3" Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.0-beta3-b84) Java HotSpot(TM) Client VM (build 1.4.0-beta3-b84, mixed mode) On Mon, Jan 21, 2002 at 06:26:51AM -0700, Huston wrote: > Bill, > > If you are sending the message and then adding data to the DataStream then > yes, I haven't added the call to continue sending as more data is added to > the DataStream. That is next on my list of tasks. > > If this is not what you are doing, I haven't seen the problem so please send > some sample code to reproduce the problem and I will look into it. > > --Huston > > > I appear to be hitting a condition where I stall in sendMSG when > > I have put more than 4K on the output datastream to send. > > Looking at the conversation as it goes by, the RPY starts > > being sent, and I get the first 3 frames, the recieving end > > is sending SEQ frames and opening the window properly, but > > the sendMSG function aparently stalls after sending 4K of > > data. Should'nt the SEQ's be getting processed? > > > > Is this a known problem? > |
|
From: William J. M. <wm...@ol...> - 2002-01-22 01:14:09
|
Well, we were just stuffing a really large (10K+) thing in a single sendMSG and seeing it stall. No additional adds. -bill On Mon, Jan 21, 2002 at 06:26:51AM -0700, Huston wrote: > Bill, > > If you are sending the message and then adding data to the DataStream then > yes, I haven't added the call to continue sending as more data is added to > the DataStream. That is next on my list of tasks. > > If this is not what you are doing, I haven't seen the problem so please send > some sample code to reproduce the problem and I will look into it. > > --Huston > > > I appear to be hitting a condition where I stall in sendMSG when > > I have put more than 4K on the output datastream to send. > > Looking at the conversation as it goes by, the RPY starts > > being sent, and I get the first 3 frames, the recieving end > > is sending SEQ frames and opening the window properly, but > > the sendMSG function aparently stalls after sending 4K of > > data. Should'nt the SEQ's be getting processed? > > > > Is this a known problem? > |
|
From: Huston <hu...@us...> - 2002-01-21 13:29:07
|
Bill, If you are sending the message and then adding data to the DataStream then yes, I haven't added the call to continue sending as more data is added to the DataStream. That is next on my list of tasks. If this is not what you are doing, I haven't seen the problem so please send some sample code to reproduce the problem and I will look into it. --Huston > I appear to be hitting a condition where I stall in sendMSG when > I have put more than 4K on the output datastream to send. > Looking at the conversation as it goes by, the RPY starts > being sent, and I get the first 3 frames, the recieving end > is sending SEQ frames and opening the window properly, but > the sendMSG function aparently stalls after sending 4K of > data. Should'nt the SEQ's be getting processed? > > Is this a known problem? |
6 messages has been excluded from this view by a project administrator.
