Re: [wsjt-devel] crash in jtmsk
Hi Steve, Thanks for the bug report. Were you running with T/R sequence lebngth 30 s? I think there are problems there. Try setting T/R = 15 s, which is what everyone is using. I will trace the problem. -- Joe On 8/29/2015 8:59 AM, Steven Franke wrote: Attemptimg to play with jtmsk here using r5824. I am seeing the following: $ ./wsjtx At line 65 of file /home/radio/Builds/wsjtx_exp/lib/jtmsk.f90 Fortran runtime error: Index '-8543' of dimension 1 of array 'c' outside of expected range (1:524288) Segmentation fault (core dumped) Also, the frequency selection dropdown does not work. Should it? Steve k9an -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] crash in jtmsk
Steve -- No apparent problem here when monitoring a birdie. Does it die when attempting to decode, at the end of a sequence? If you hace checked Save all and re-open the saved file, does that crash the program? If so, could you send me the file? -- Joe On 8/29/2015 10:29 AM, Steven Franke wrote: OK, I spoke too soon. Running with 15s T/R sequence. Seems to run OK when the input is “pure” noise - but I can make it crash within seconds if I tune to or tune slowly through a birdie. Steve k9an On Aug 29, 2015, at 2:22 PM, Steven Frankes.j.fra...@icloud.com wrote: T/R sequence length was set to 5s. I am not accustomed to these modes, so I am essentially a random-input-vector tester… I switched to T/R length 15s and it seems to be running OK now. Steve k9an On Aug 29, 2015, at 2:01 PM, Joe Taylorj...@princeton.edu wrote: Hi Steve, Thanks for the bug report. Were you running with T/R sequence lebngth 30 s? I think there are problems there. Try setting T/R = 15 s, which is what everyone is using. I will trace the problem. -- Joe On 8/29/2015 8:59 AM, Steven Franke wrote: Attemptimg to play with jtmsk here using r5824. I am seeing the following: $ ./wsjtx At line 65 of file /home/radio/Builds/wsjtx_exp/lib/jtmsk.f90 Fortran runtime error: Index '-8543' of dimension 1 of array 'c' outside of expected range (1:524288) Segmentation fault (core dumped) Also, the frequency selection dropdown does not work. Should it? Steve k9an -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] crash in jtmsk
Hi Bill, Steve, and all, On 8/29/2015 9:44 AM, Bill Somerville wrote: On 29/08/2015 13:59, Steven Franke wrote: Hi Steve, Attemptimg to play with jtmsk here using r5824. I am seeing the following: $ ./wsjtx At line 65 of file /home/radio/Builds/wsjtx_exp/lib/jtmsk.f90 Fortran runtime error: Index '-8543' of dimension 1 of array 'c' outside of expected range (1:524288) Segmentation fault (core dumped) The bounds error is probably best addressed by Joe, I can't see an obvious issue but I suspect an uninitialized variable may be the culprit. r5826 has a (temporary?) fix that will avoid this crash. Also, the frequency selection dropdown does not work. Should it? This is an area that needs addressing with respect to fast modes. Due to the wider bandwidths spot working frequencies are not necessarily appropriate. The drop down list can be added to for each mode in Settings-Frequencies by right clicking the working frequencies table and selecting Insert Whether we supply default frequencies for fast modes is debatable and other than for logging purposes the whole issue of rig control may not be relevant to most VHF up users except for the Doppler correction facility for slow modes off the Moon. Actually I think mode-specific spot frequencies for the fast modes probably do make sense. In NA, for example, we have been using 50.280 for JT9H and now also JTMSK. You can leave the program running all day, and PSKreporter shows your spots in the usual way. With meteor scatter propagation, a single frequency is effectively shared with time division multiple access rather than frequency division multiple access. On another matter: as you have probably noticed, all these modes have produced a lot of special cases (mostly in mainwindow.cpp) of the form if(m_mode == JTMSK) do this... To keep the main window GUI from getting too complicated, I've been making irrelevant controls invisible. Most likely there are better ways to do some of these necessary things. Ways that a real programmer would know about. We talked once about using a QStateMachine for handling mode-specific behavior. And maybe tabs or layers of controls? Anyway, I'm sure these things could be much improved, and the GUI made cleaner. I've been concentrating mostly on making the new modes work. -- Joe, K1JT -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] crash in jtmsk
Steve -- On 8/29/2015 10:29 AM, Steven Franke wrote: OK, I spoke too soon. Running with 15s T/R sequence. Seems to run OK when the input is “pure” noise - but I can make it crash within seconds if I tune to or tune slowly through a birdie. That may be a different problem. I will investigate with some handy birdies (got plenty of them...) -- Joe Steve k9an On Aug 29, 2015, at 2:22 PM, Steven Frankes.j.fra...@icloud.com wrote: T/R sequence length was set to 5s. I am not accustomed to these modes, so I am essentially a random-input-vector tester… I switched to T/R length 15s and it seems to be running OK now. Steve k9an On Aug 29, 2015, at 2:01 PM, Joe Taylorj...@princeton.edu wrote: Hi Steve, Thanks for the bug report. Were you running with T/R sequence lebngth 30 s? I think there are problems there. Try setting T/R = 15 s, which is what everyone is using. I will trace the problem. -- Joe On 8/29/2015 8:59 AM, Steven Franke wrote: Attemptimg to play with jtmsk here using r5824. I am seeing the following: $ ./wsjtx At line 65 of file /home/radio/Builds/wsjtx_exp/lib/jtmsk.f90 Fortran runtime error: Index '-8543' of dimension 1 of array 'c' outside of expected range (1:524288) Segmentation fault (core dumped) Also, the frequency selection dropdown does not work. Should it? Steve k9an -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] crash in jtmsk
OK, I spoke too soon. Running with 15s T/R sequence. Seems to run OK when the input is “pure” noise - but I can make it crash within seconds if I tune to or tune slowly through a birdie. Steve k9an On Aug 29, 2015, at 2:22 PM, Steven Franke s.j.fra...@icloud.com wrote: T/R sequence length was set to 5s. I am not accustomed to these modes, so I am essentially a random-input-vector tester… I switched to T/R length 15s and it seems to be running OK now. Steve k9an On Aug 29, 2015, at 2:01 PM, Joe Taylor j...@princeton.edu wrote: Hi Steve, Thanks for the bug report. Were you running with T/R sequence lebngth 30 s? I think there are problems there. Try setting T/R = 15 s, which is what everyone is using. I will trace the problem. -- Joe On 8/29/2015 8:59 AM, Steven Franke wrote: Attemptimg to play with jtmsk here using r5824. I am seeing the following: $ ./wsjtx At line 65 of file /home/radio/Builds/wsjtx_exp/lib/jtmsk.f90 Fortran runtime error: Index '-8543' of dimension 1 of array 'c' outside of expected range (1:524288) Segmentation fault (core dumped) Also, the frequency selection dropdown does not work. Should it? Steve k9an -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] crash in jtmsk
Steve -- I think the problem is related to T/R =30 s sequences. For frequency selection: you need to explicitly enter one or more frequencies for JTMSK mode. -- Joe On 8/29/2015 9:52 AM, Steven Franke wrote: Thanks Bill. I hadn’t thought through the frequency selection issue. A bit more information for Joe regarding the crashes. I am running on linux and simply starting 1.6.1 and letting it read the initialization file that was written by 1.6.0. The crashes occur reliably and happen as soon as a few seconds after starting and as long as a couple of minutes after starting the program. In most cases, I just get a coredump without the “Fortran runtime error” text. Steve k9an On Aug 29, 2015, at 1:44 PM, Bill Somervilleg4...@classdesign.com wrote: On 29/08/2015 13:59, Steven Franke wrote: Hi Steve, Attemptimg to play with jtmsk here using r5824. I am seeing the following: $ ./wsjtx At line 65 of file /home/radio/Builds/wsjtx_exp/lib/jtmsk.f90 Fortran runtime error: Index '-8543' of dimension 1 of array 'c' outside of expected range (1:524288) Segmentation fault (core dumped) The bounds error is probably best addressed by Joe, I can't see an obvious issue but I suspect an uninitialized variable may be the culprit. Also, the frequency selection dropdown does not work. Should it? This is an area that needs addressing with respect to fast modes. Due to the wider bandwidths spot working frequencies are not necessarily appropriate. The drop down list can be added to for each mode in Settings-Frequencies by right clicking the working frequencies table and selecting Insert Whether we supply default frequencies for fast modes is debatable and other than for logging purposes the whole issue of rig control may not be relevant to most VHF up users except for the Doppler correction facility for slow modes off the Moon. Steve k9an 73 Bill G4WJS. -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] crash in jtmsk
It still crashes on the file that I sent to you Joe… Steve k9an On Aug 29, 2015, at 3:35 PM, Joe Taylor j...@princeton.edu wrote: Hi Bill, Steve, and all, On 8/29/2015 9:44 AM, Bill Somerville wrote: On 29/08/2015 13:59, Steven Franke wrote: Hi Steve, Attemptimg to play with jtmsk here using r5824. I am seeing the following: $ ./wsjtx At line 65 of file /home/radio/Builds/wsjtx_exp/lib/jtmsk.f90 Fortran runtime error: Index '-8543' of dimension 1 of array 'c' outside of expected range (1:524288) Segmentation fault (core dumped) The bounds error is probably best addressed by Joe, I can't see an obvious issue but I suspect an uninitialized variable may be the culprit. r5826 has a (temporary?) fix that will avoid this crash. Also, the frequency selection dropdown does not work. Should it? This is an area that needs addressing with respect to fast modes. Due to the wider bandwidths spot working frequencies are not necessarily appropriate. The drop down list can be added to for each mode in Settings-Frequencies by right clicking the working frequencies table and selecting Insert Whether we supply default frequencies for fast modes is debatable and other than for logging purposes the whole issue of rig control may not be relevant to most VHF up users except for the Doppler correction facility for slow modes off the Moon. Actually I think mode-specific spot frequencies for the fast modes probably do make sense. In NA, for example, we have been using 50.280 for JT9H and now also JTMSK. You can leave the program running all day, and PSKreporter shows your spots in the usual way. With meteor scatter propagation, a single frequency is effectively shared with time division multiple access rather than frequency division multiple access. On another matter: as you have probably noticed, all these modes have produced a lot of special cases (mostly in mainwindow.cpp) of the form if(m_mode == JTMSK) do this... To keep the main window GUI from getting too complicated, I've been making irrelevant controls invisible. Most likely there are better ways to do some of these necessary things. Ways that a real programmer would know about. We talked once about using a QStateMachine for handling mode-specific behavior. And maybe tabs or layers of controls? Anyway, I'm sure these things could be much improved, and the GUI made cleaner. I've been concentrating mostly on making the new modes work. -- Joe, K1JT -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] crash in jtmsk
Steve -- I didn't get a file from you, today??? -- Joe On 8/29/2015 1:02 PM, Steven Franke wrote: It still crashes on the file that I sent to you Joe… Steve k9an On Aug 29, 2015, at 3:35 PM, Joe Taylorj...@princeton.edu wrote: Hi Bill, Steve, and all, On 8/29/2015 9:44 AM, Bill Somerville wrote: On 29/08/2015 13:59, Steven Franke wrote: Hi Steve, Attemptimg to play with jtmsk here using r5824. I am seeing the following: $ ./wsjtx At line 65 of file /home/radio/Builds/wsjtx_exp/lib/jtmsk.f90 Fortran runtime error: Index '-8543' of dimension 1 of array 'c' outside of expected range (1:524288) Segmentation fault (core dumped) The bounds error is probably best addressed by Joe, I can't see an obvious issue but I suspect an uninitialized variable may be the culprit. r5826 has a (temporary?) fix that will avoid this crash. Also, the frequency selection dropdown does not work. Should it? This is an area that needs addressing with respect to fast modes. Due to the wider bandwidths spot working frequencies are not necessarily appropriate. The drop down list can be added to for each mode in Settings-Frequencies by right clicking the working frequencies table and selecting Insert Whether we supply default frequencies for fast modes is debatable and other than for logging purposes the whole issue of rig control may not be relevant to most VHF up users except for the Doppler correction facility for slow modes off the Moon. Actually I think mode-specific spot frequencies for the fast modes probably do make sense. In NA, for example, we have been using 50.280 for JT9H and now also JTMSK. You can leave the program running all day, and PSKreporter shows your spots in the usual way. With meteor scatter propagation, a single frequency is effectively shared with time division multiple access rather than frequency division multiple access. On another matter: as you have probably noticed, all these modes have produced a lot of special cases (mostly in mainwindow.cpp) of the form if(m_mode == JTMSK) do this... To keep the main window GUI from getting too complicated, I've been making irrelevant controls invisible. Most likely there are better ways to do some of these necessary things. Ways that a real programmer would know about. We talked once about using a QStateMachine for handling mode-specific behavior. And maybe tabs or layers of controls? Anyway, I'm sure these things could be much improved, and the GUI made cleaner. I've been concentrating mostly on making the new modes work. -- Joe, K1JT -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel