Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[roseway]

Tweaks and improvements mainly arising from comments on the previous version:

- added option to autosnapshot the event log
- added option to save the event log on exit
- corrected error in recording bitswaps per tone with some routers
- corrected issue of spike in swaps per minute graph after pause or missing sample
- pauses and missing samples are plotted as empty spaces instead of zero values as before
- fixed bug whereby the QLN and HLog graphs couldn't be seen in some cases
- added option to hide the legend on the time-related graphs (if they obscure the graph)
- QLN, HLog and SNR per tone graphs now respond to Expanded/Compressed option with ADSL connections

http://rsux.plainroad.me.uk

[broadstairs]

Up and running Eric.... now hidden legend on snrm graph and save log on exit set!

Stuart

[roseway]

 

[burakkucat]

Downloaded and started executing at 2208 hours UTC. 

[roseway]

Would that be 2208 exactly? Every second might be important.

[burakkucat]

Would that be 2208 exactly? Every second might be important.

No,    it was closer to 2208.30 hours and rs-ux recorded it (correctly) as 2209 hours. 

Evidence attached, below.

I'm happy to report that the contracted/expanded feature of all the graphs now works as expected for ADSL(2(+)) services.

Just one oddity, which the below graph shows. At present, I seem to be subjected to some form of RFI at the US frequencies and so my US SNRM takes a battering. Normally it is 6 dB.    I configured rs-ux to clip the SNRM at 10 dB and that was the original height of the Y-axis of the graph. Having seen the graph with a number of minutes showing a green line at the Y-axis maximum, I decided to adjust the clipping to 15 dB. That setting was accepted and as further data is plotted, I see that the Y-axis maximum is now set at 20 dB.

Confusing. 

[burakkucat]

Hmm . . . Now I am having a problem in getting the SNRM graph to re-size back to a Y-axis of 10 dB maximum, following a DSL down/up re-sync and the re-setting the clipping limit to 10 dB. 

Observant viewers will note that I paused recording whilst adjusting the configuration.

[roseway]

What you're seeing is intended behaviour, although I accept that it can be improved. The rescaling function doesn't know anything about the clipping value - it simply responds to the range of values which it has to plot. Any clipping is performed before the data is sent to the graph.

The reason that in the case you quote it rescaled to a maximum of 20 dB and not 15 dB is that the rescaling function tries to maintain nice round numbers for the scale intervals. 20 divides nicely into 10 x 2, but 15 doesn't divide so conveniently. I chose to keep the number of scale intervals to 10 for simplicity - I considered making it more flexible, but it began to look horribly complicated.

Regarding your last point, as it works at present the graph remembers the highest value currently stored and scales accordingly. It doesn't rescale just for the range of values on view, it scales according to the total range of values stored. In your case there was still a value of just over 14 dB stored, so it maintained the scaling to accommodate this. If you had stopped and restarted when you forced a re-sync, the graph scaling would have been reset to its default range.

There are numerous ways in which I can improve this. What I'm thinking of is to store the unclipped values and use the clipping value to control the scaling directly. I think I would have done it this way from the start if I'd thought of it.

[broadstairs]

Eric I was looking at the graph options page and wondered what the third value (currently 20) was because it has no description on the LH side at least on my screen in Linux 32bit. I only realised what it was because the popup appeared when I hovered the mouse over it.

By the way I still have a problem with these popup message not displaying all the text, the popup bar appears the correct length but the text is cut off. Some of the boxes are complete if the text is short enough so the 'Upstream SNR Margin' box shows the complete text but the one about the clipping only shows 'The clipping level to avoid exce'.

One last point the Bits swaps per minute graph does not obey the graph line thickness option.

Stuart

[roseway]

Eric I was looking at the graph options page and wondered what the third value (currently 20) was because it has no description on the LH side at least on my screen in Linux 32bit. I only realised what it was because the popup appeared when I hovered the mouse over it.

I'll check to see if some oddity has crept into the 32-bit build of the program, but there certainly is a caption there.

By the way I still have a problem with these popup message not displaying all the text, the popup bar appears the correct length but the text is cut off. Some of the boxes are complete if the text is short enough so the 'Upstream SNR Margin' box shows the complete text but the one about the clipping only shows 'The clipping level to avoid exce'.

I can't reproduce this unfortunately. I might be able to work round it by making the hints occupy several short lines instead of one long line. I'll put this on the todo list.
Just to add to this, it appears that, on your system, hints are limited to a line length of 32 characters. I wonder if this is a limitation of some desktop systems or OS versions?

One last point the Bits swaps per minute graph does not obey the graph line thickness option.

An omission on my part. I'll correct it in the next version.

[roseway]

I've found the reason for the missing captions. It relates to font heights and the height of the group boxes that these configuration items are contained in.

The question of the truncated hints is a bit more complicated than I thought, but I think making the hints multi-line will probably deal with it.

[broadstairs]

Eric on the problem of partial hints, do you use a default font in Linux or a specific one? My thought could be that either I may have a missing font or if it is a default one I can identify then I could change the font or size of the default to see if it changes what is displayed. At least that way we might find out how/why it's happening.

Stuart

Edit: I've just been playing with the default font sizes in KDE and it seems like they are not being picked up correctly in rs-ux. I can alter one and apply the change and the hint will appear OK even though that is not the font I cahnged as the font size in the hint does not change it just starts displaying. If I change some of the fonts the rs-ux display does change. Dont know if I explained that well but anyway it does seem to be something to do with the font or font size.

[roseway]

The program uses the system default font, but I don't know how that is determined. It's easy to define on a Windows system, but on a Linux system there are many different ways in which the default might be defined. Different desktop systems have their own ways of handling fonts, and there is also the underlying OS default. In rs-ux I've specified the actual font size, because not to do so results in anarchy, but I've left the font type as the default. I don't think I can specify a particular font, because there's no guarantee that it will exist on any particular system.

[broadstairs]

Eric how about allowing the user to override the default font option(s) by providing a set up for the font(s) and size(s) used for the gui, that way if folks have a problem (or simply dont like the default) then they can change it. That way if it works fine no change but in other cases the font can be changed if desired.

Its strange that if I make a change using the System Settings/Application Appearance/Fonts option from the start menu in KDE it seems to fix the problems I have with rs-ux. However that only sticks for that run of rs-ux, as soon as I close and open the program again the problem is back until I make the change.

Could you perhaps let me know exactly what system call/process/setting you use do select/define the font and size used and then I will see if there is a problem more generally with KDE.

Stuart

[roseway]

Eric how about allowing the user to override the default font option(s) by providing a set up for the font(s) and size(s) used for the gui, that way if folks have a problem (or simply dont like the default) then they can change it. That way if it works fine no change but in other cases the font can be changed if desired.

That's certainly possible, and I'll look into it, but I'd have to change the definition of rather a large number of text items, so I think I prefer to avoid doing this if I can.

Could you perhaps let me know exactly what system call/process/setting you use do select/define the font and size used and then I will see if there is a problem more generally with KDE.

I appreciate the offer, but unfortunately it's not that simple. The development system I use (Lazarus) is a high level RAD system, and the programmer doesn't get involved with low level system calls. The GUI side of the program is built in a WYSIWYG fashion, and most of the visual elements are selected and dimensioned by pointing and clicking. The system calls which support this are buried in the Lazarus code; Lazarus is a cross-platform system, so the program which I develop on Linux can be built with little or no modification on other platforms.

The supporting low-level code is of course open to inspection (and modification if necessary) but it would be a very large job to seek out all the relevant system calls.

The bottom line is that I'm sure there are simpler solutions to these display issues, which don't involve diving down into the underlying code.