48Re: 2009/08/11 Version Released!
- Aug 11, 2009--- In email@example.com, "Peter Loveall" <pete@...> wrote:
>I (hopefully) never said it was a server bust. I'm dealing with a bunch of unknowns here, and I've ONLY seen the problem while driving down the road going in and out of cellular coverage. It's not the world's most reproducable environment.
> I would say that since there have been no other reports of javAPRSSrvr just quitting sending packets but maintaining the keep-alives, the server "going dumb" is actually the result of an improper filter setting. The advantage of setting the filter at login is the logresp comment you get back from the server confirming your new filter setting. At any time later, you can still use a comment line to set the filter to something else.
Since I'm already fielding the #logresp and you've confirmed what I suspected that I'm able to update the filter later no matter which way I specify it, I'll see about embedding the first filter in the logon command.
>I'm not sure I understand the "reconnect as soon as possible" bit. APRSISCE does try to immediately reconnect if a connection is lost UNLESS the Windows Mobile connection manager status says that it won't succeed anyway. The disconnect time is not delayed, by not counting the keep-alives as traffic, I'm actually ALLOWING the disconnect to occur after the quiet time has expired. Before I started ignoring the keep-alives, the client would simply go (effectively) dead receiving ONLY keepalives until the user got truly frustrated and closed the client.
> I am unable to test the software at this point because I have no need for highly active data flows to my cell phone and I want a reconnect attempt as soon as possible when I do use my cell phone for APRS. Personnaly, I find masking a client issue by extending the disconnect time only delays determination of the real cause of a problem and usually leads me to frustration with any software that I work with.
I'm not trying to "mask a client issue", I'm trying to resolve the issue while still keeping the Beta functioning at the same time. That's the whole point of BETA software, to work around unknown issues to allow other testing proceed while still attempting to find a permanent resolution to the workarounds. Frustration comes with the Beta landscape and testing partially baked software isn't for everyone.
I have no doubt that the issue is in my client code, it's certainly the new kid on the well established block. As soon as I understand all of the vagueness of "always on" cellular data connections and the behavior of continuously open sockets across cellular switches, I have every confidence that I'll eliminate the problem. In the meantime, at least the client recovers the connection and gets the feed going again as soon as it notices the lack of non-keep-alive traffic.
Lynn (D) - KJ4ERJ - Adding my contribution to the APRS community
- << Previous post in topic