Re: Several questions/comments
Hi,
Thank you for your feedback.
Performance:
I'm probably going to disappoint a lot of people, but to be honest, 300k QSO is the limit for which QLog is not designed and in my opinion I am not able to achieve any improvements with the current architecture. From my own testing, I have found that it is acceptable to use qlog up to 20-30k QSOs. More QSOs are noticeably slowed down. I have a test log with 90k QSOs for debugging database functions. More QSO is no longer usable. Therefore, I think that 300k is the limit where practically everything will go wrong.
Therefore, I will not comment on your other questions regarding the performance. There may be some minor improvements, but they could come at the expense of losing certain functions. However, I welcome any code changes that will improve QLog's performance
Cursor RST:? N1MM-style can be implemented. I hope no one sends me a threatening email accusing me of copying their style.
Stats: Closely related to the question about performance. Currently, QLog is not designed for such a number of QSO. Keep in mind that showing 300kQSO to a map takes some time.
Filter:
The GUI cannot contain a filter for everything, because that would be a single button. That is why it is necessary to implement a custom filter, where you can define the yourself.
Regards Ladislav
Hi,
I installed QLog since I was looking for a logger on Linux. I installed the latest flatpack version (v0.38). It's running on an AMD Ryzen 7 4800U with Radeon Graphics and 32GB RAM. Linux flavor is Tuxedo OS 3 v22.04 (jammy).
Installation was straightforward, setup as well. I exported one of my logs from my current logger into an ADIF file which I imported into QLog. Took a while but all data showed up. It now holds some 300k Qs. I expect others to have more.
1. Start up now is slow: Can loading the complete log at start-up be made an option or is it relying on having all data in memory while running?
2. When entering a callsign in the callsign field, it "hangs" for a short while to display all previous contacts and until it moves to the RST field. I have no radio connected yet but would this behavior influence the start of the CW transmission?
3. Wenn hitting Ctrl-N, it takes a while until the reload is done and a new callsign can be entered. That's why I'm asking for the option in 1)? -> It would make working in a pile-up a not very easy task.
4. When getting to the RST field, the cursor sits behing the numbers. As I'm the guy handing out real reports (except during contests), I need to go back, delete and enter the right value and move on to the next field. Any chance to preselect the S value in order to be able to correct the value in one hit and move on?
5. Stats (Logbook/Statistics): Some real nice stats, some are displayed really fast but the "Show on map" Graph Type takes an extremely high amount of time. If you resize the window, it will again take extremely long. That's why I indicated my PC setup. Am I on the low side of resources needed? It's all SSD, BTW.
6. Filter by Band is fine, I'm also looking for a Filter by Mode. Did I miss it or could it by done?
There's usually not running going on on this PC as I'm using it only to surf the Internet and do my emails.
Sri if I missed some answers to my questions in the wiki but I did not see them, That would be my bad.
73 Norby
|
Several questions/comments
Hi,
I installed QLog since I was looking for a logger on Linux. I installed the latest flatpack version (v0.38). It's running on an AMD Ryzen 7 4800U with Radeon Graphics and 32GB RAM. Linux flavor is Tuxedo OS 3 v22.04 (jammy).
Installation was straightforward, setup as well. I exported one of my logs from my current logger into an ADIF file which I imported into QLog. Took a while but all data showed up. It now holds some 300k Qs. I expect others to have more.
1. Start up now is slow: Can loading the complete log at start-up be made an option or is it relying on having all data in memory while running?
2. When entering a callsign in the callsign field, it "hangs" for a short while to display all previous contacts and until it moves to the RST field. I have no radio connected yet but would this behavior influence the start of the CW transmission?
3. Wenn hitting Ctrl-N, it takes a while until the reload is done and a new callsign can be entered. That's why I'm asking for the option in 1) -> It would make working in a pile-up a not very easy task.
4. When getting to the RST field, the cursor sits behing the numbers. As I'm the guy handing out real reports (except during contests), I need to go back, delete and enter the right value and move on to the next field. Any chance to preselect the S value in order to be able to correct the value in one hit and move on?
5. Stats (Logbook/Statistics): Some real nice stats, some are displayed really fast but the "Show on map" Graph Type takes an extremely high amount of time. If you resize the window, it will again take extremely long. That's why I indicated my PC setup. Am I on the low side of resources needed? It's all SSD, BTW.
6. Filter by Band is fine, I'm also looking for a Filter by Mode. Did I miss it or could it by done?
There's usually not running going on on this PC as I'm using it only to surf the Internet and do my emails.
Sri if I missed some answers to my questions in the wiki but I did not see them, That would be my bad.
73 Norby
|
Re: Release 0.38.0 bug????
I'm sorry, but even the upgrade to the latest Hamlib version (static linking) did not solve this problem. Based on the QLog trace file, I can see that QLog is sending requested modes correctly, but Rig reports one later mode. It means QLog sequence is USB LSB, Rig sequence is "no change" , USB, LSB. I have to say that I don't see where the issue is. I don't want to say that it is a Hamlib/Rig issue.
It is necessary to check the rig settings to be sure if there are no automatic events related to changing the mode (any presets or something like this).
I am in contact with Mark. We are testing different variants of QLog. Currently it is clear that it is not the command timing from Qlog.
út 3. 9. 2024 v?13:28 odesílatel Michael Morgan via <cmorgan13= [email protected]> napsal: I am using a Flex radio here and I am not seeing the issue either on Mac or Windows.? My son has a Yaesu 710 so I may can run up to his house this evening and test/play with it.
Michael,?
exactly this piece of code is where the potential problem is (maybe).?The bad thing about it is that I'm not able to simulate it.?I tried different CPU underclocking to see if the call order changes a behaviour (Qlog is a heavily multithreaded application).?Tested Windows/Linux...But found no issue. Qlog?works without any issue? with my IC7300.That's the bad thing, I have no idea if it's in this signal/slot order issue or if it's in Hamlib call, because Qlog sends commands very fast.?It seems that FTx users are impacted, but with different Hamlib versions (4.3, 4.5.x and even with 4.6).
I'll try to go through all the commits in release 0.38 today.?Maybe something will come to mind.?Unfortunately, there are many commits ;-(
út 3. 9. 2024 v?1:42 odesílatel Michael Morgan via ??<cmorgan13= [email protected]> napsal: Ladislav, I have just been looking at the code a little on this.? I think I may have traced how this is working (with the tunedx signal), I'm still trying to get my head wrapped around QT.? But it looks like QLog is setting the Mode first then the Frequency.? Wonder if it would make a difference if the frequency is changed then the mode?? Or I may not be following the code.
Michael, AA5SH
Hallo to all, I can reproduce the behavior as described in the messages before with my rig, FT-991, under openSUSE LINUX Leap 15.6 with QLog 0.38 and 0.37.2 with Hamlib 4.6.? Hamlib 4.5.5 does not have it. So it seems to be a problem between QLog and Hamlib 4.6. However this is not a showstopper for me. I only have to click twice if the band and mode changes. ? 73, Walter DL8FCL
|
Re: Release 0.38.0 bug????
I am in contact with Mark. We are testing different variants of QLog. Currently it is clear that it is not the command timing from Qlog. út 3. 9. 2024 v?13:28 odesílatel Michael Morgan via <cmorgan13= [email protected]> napsal:
I am using a Flex radio here and I am not seeing the issue either on Mac or Windows.? My son has a Yaesu 710 so I may can run up to his house this evening and test/play with it.
Michael,?
exactly this piece of code is where the potential problem is (maybe).?The bad thing about it is that I'm not able to simulate it.?I tried different CPU underclocking to see if the call order changes a behaviour (Qlog is a heavily multithreaded application).?Tested Windows/Linux...But found no issue. Qlog?works without any issue? with my IC7300.That's the bad thing, I have no idea if it's in this signal/slot order issue or if it's in Hamlib call, because Qlog sends commands very fast.?It seems that FTx users are impacted, but with different Hamlib versions (4.3, 4.5.x and even with 4.6).
I'll try to go through all the commits in release 0.38 today.?Maybe something will come to mind.?Unfortunately, there are many commits ;-(
út 3. 9. 2024 v?1:42 odesílatel Michael Morgan via ??<cmorgan13= [email protected]> napsal: Ladislav, I have just been looking at the code a little on this.? I think I may have traced how this is working (with the tunedx signal), I'm still trying to get my head wrapped around QT.? But it looks like QLog is setting the Mode first then the Frequency.? Wonder if it would make a difference if the frequency is changed then the mode?? Or I may not be following the code.
Michael, AA5SH
Hallo to all, I can reproduce the behavior as described in the messages before with my rig, FT-991, under openSUSE LINUX Leap 15.6 with QLog 0.38 and 0.37.2 with Hamlib 4.6.? Hamlib 4.5.5 does not have it. So it seems to be a problem between QLog and Hamlib 4.6. However this is not a showstopper for me. I only have to click twice if the band and mode changes. ? 73, Walter DL8FCL
|
Re: Release 0.38.0 bug????
I am using a Flex radio here and I am not seeing the issue either on Mac or Windows. ?My son has a Yaesu 710 so I may can run up to his house this evening and test/play with it.
toggle quoted message
Show quoted text
On Sep 3, 2024, at 2:12?AM, ok1mlg via groups.io <ok1mlg@...> wrote:
Michael,?
exactly this piece of code is where the potential problem is (maybe).?The bad thing about it is that I'm not able to simulate it.?I tried different CPU underclocking to see if the call order changes a behaviour (Qlog is a heavily multithreaded application).?Tested Windows/Linux...But found no issue. Qlog?works without any issue? with my IC7300.That's the bad thing, I have no idea if it's in this signal/slot order issue or if it's in Hamlib call, because Qlog sends commands very fast.?It seems that FTx users are impacted, but with different Hamlib versions (4.3, 4.5.x and even with 4.6).
I'll try to go through all the commits in release 0.38 today.?Maybe something will come to mind.?Unfortunately, there are many commits ;-(
út 3. 9. 2024 v?1:42 odesílatel Michael Morgan via ??<cmorgan13= [email protected]> napsal: Ladislav, I have just been looking at the code a little on this.? I think I may have traced how this is working (with the tunedx signal), I'm still trying to get my head wrapped around QT.? But it looks like QLog is setting the Mode first then the Frequency.? Wonder if it would make a difference if the frequency is changed then the mode?? Or I may not be following the code.
Michael, AA5SH
Hallo to all, I can reproduce the behavior as described in the messages before with my rig, FT-991, under openSUSE LINUX Leap 15.6 with QLog 0.38 and 0.37.2 with Hamlib 4.6.? Hamlib 4.5.5 does not have it. So it seems to be a problem between QLog and Hamlib 4.6. However this is not a showstopper for me. I only have to click twice if the band and mode changes. ? 73, Walter DL8FCL
|
Re: Release 0.38.0 bug????
Michael,
exactly this piece of code is where the potential problem is (maybe). The bad thing about it is that I'm not able to simulate it. I tried different CPU underclocking to see if the call order changes a behaviour (Qlog is a heavily multithreaded application). Tested Windows/Linux...But found no issue. Qlog works without any issue? with my IC7300. That's the bad thing, I have no idea if it's in this signal/slot order issue or if it's in Hamlib call, because Qlog sends commands very fast. It seems that FTx users are impacted, but with different Hamlib versions (4.3, 4.5.x and even with 4.6).
I'll try to go through all the commits in release 0.38 today. Maybe something will come to mind. Unfortunately, there are many commits ;-(
út 3. 9. 2024 v?1:42 odesílatel Michael Morgan via <cmorgan13= [email protected]> napsal:
Ladislav, I have just been looking at the code a little on this.? I think I may have traced how this is working (with the tunedx signal), I'm still trying to get my head wrapped around QT.? But it looks like QLog is setting the Mode first then the Frequency.? Wonder if it would make a difference if the frequency is changed then the mode?? Or I may not be following the code.
Michael, AA5SH
Hallo to all,
I can reproduce the behavior as described in the messages before with my rig, FT-991, under openSUSE LINUX Leap 15.6 with QLog 0.38 and 0.37.2 with Hamlib 4.6.? Hamlib 4.5.5 does not have it. So it seems to be a problem between QLog and Hamlib 4.6.
However this is not a showstopper for me. I only have to click twice if the band and mode changes.
?
73, Walter DL8FCL
|
Re: Release 0.38.0 bug????
Trace file emailed to you.
?
Thanks
|
Re: Release 0.38.0 bug????
Ladislav, I have just been looking at the code a little on this.? I think I may have traced how this is working (with the tunedx signal), I'm still trying to get my head wrapped around QT.? But it looks like QLog is setting the Mode first then the Frequency.? Wonder if it would make a difference if the frequency is changed then the mode?? Or I may not be following the code.
Michael, AA5SH
Hallo to all,
I can reproduce the behavior as described in the messages before with my rig, FT-991, under openSUSE LINUX Leap 15.6 with QLog 0.38 and 0.37.2 with Hamlib 4.6.? Hamlib 4.5.5 does not have it. So it seems to be a problem between QLog and Hamlib 4.6.
However this is not a showstopper for me. I only have to click twice if the band and mode changes.
?
73, Walter DL8FCL
|
Re: Release 0.38.0 bug????
Hallo to all,
I can reproduce the behavior as described in the messages before with my rig, FT-991, under openSUSE LINUX Leap 15.6 with QLog 0.38 and 0.37.2 with Hamlib 4.6.? Hamlib 4.5.5 does not have it. So it seems to be a problem between QLog and Hamlib 4.6.
However this is not a showstopper for me. I only have to click twice if the band and mode changes.
?
73, Walter DL8FCL
|
Re: Release 0.38.0 bug????
Hi,
enclosed is qlog.trace.?
?
Qlog 0.38 on Ubuntu 24.04, with FTDX10.?
?
Clicked on several USB spots where the mode changed from CW to SSB on the second click.?
?
?
73 de LB2EG.?
|
Re: Release 0.38.0 bug????
It will work under Linux
In the meantime, I tested all possible variants of QLog 0.38 under Windows and Linux, with the CPU in PowerSave Mode/Performance setting, and I couldn’t detect any of the behavior you described - my rig is IC7300. I also reviewed the code for detecting and setting the mode several times. The result is that I can’t see any issue. Therefore, the trace file will be very important in this case. It would be better to send it to my private email because the file will likely be large.
Regards Ladislav
I will but will probably be tomorrow. ?
Im using Linux Mint. ? ?Will that command work in Linux or do I need to run it in Windows?
|
Re: Release 0.38.0 bug????
I will but will probably be tomorrow. ?
Im using Linux Mint. ? ?Will that command work in Linux or do I need to run it in Windows?
|
Re: Release 0.38.0 bug????
|
Re: Release 0.38.0 bug????
That's strange. Therefore, I need the trace file. Please, could someone with FTDX10 run QLog as shown below and send me the trace file (~/qlog.trace)? Please, use my private email address to avoid spamming of the non-FTDX users
Test it on the 20m band first, then on the 40m band, or use a similar testing scenario with a description of what you did.
The command is:
QT_LOGGING_RULES="qlog.rig.driver.hamlibdrv.*=true" qlog 1>&2 2>~/qlog.trace
po 2. 9. 2024 v?16:51 odesílatel Richard, LB2EG via <la4sga= [email protected]> napsal:
Hi,?
Using QLog 0.38 with a FTDX10 with Hamlib, I can confirm the? WW4DX findings that it sometimes takes more clicks on a sport to move the mode to the spot mode.??
?
73 de LB2EG Richard
?
?
|
Re: Release 0.38.0 bug????
Hi,?
Using QLog 0.38 with a FTDX10 with Hamlib, I can confirm the? WW4DX findings that it sometimes takes more clicks on a sport to move the mode to the spot mode.??
?
73 de LB2EG Richard
?
?
|
Re: Release 0.38.0 bug????
100ms is also unusual for me. Although Hamlib doesn't handle it this quickly, the setting hasn't changed, as Mark mentioned.
Mark, are you using flatpak or deb package? We could try to debug directly via the trace file, but it requires your cooperation and patience.Is it possible to send you a private email with instructions on how to generate a trace file?
I sure did raise it to 500ms yesterday. ? Same problem.? Changed it back for to 100ms.?
Thanks!
|
Re: Release 0.38.0 bug????
I sure did raise it to 500ms yesterday. ? Same problem. ?Changed it back for to 100ms.?
Thanks!
|
Re: Release 0.38.0 bug????
I wonder if this is a Hamlib bug - but doesn't make sense if it worked previously.? I see where the latest Hamlib shows the updates for a few months ago for the 101.? But one thing maybe to try is to raise your update frequency from 100ms to like 500ms to just see if something is colliding or something.
|
Re: Release 0.38.0 bug????
I can just multiple click the spot until it changes to the correct mode.??
?
I'll keep up with the updates, and maybe something will fix it.
?
Thanks for all you do....
?
WW4DX
|
Re: Release 0.38.0 bug????
Changed to "DIRECT FREQ".?
?
Power cycled rig.? Rebooted PC.?
?
Same result.
?
Thanks
?
|
WW4DX
Walter, DL8FCL