Sleepyhead using Icon auto / Flow chart

[John McGowan]

Hi All,

A seventy year-old here. On therapy now twenty years and wouldn't be without it.......

I'm new to using sleepyhead with a comparatively new F&P Icon auto.  I've read the manuals and have a problem that one of you perhaps can help me with?  The Flow chart is missing from my download.  I have managed to ascertain that the Icon auto does record the data once the card is in the machine.  I also gather that it collects 7 days info at a time.  Maybe I'm being impatient ( I only downloaded the software three days ago) and the chart may suddenly appear.

Any takers?

Thanking you all in advance....

JMcG

PS  All other data is there from initiation (March 2018)

[Crimson Nape]

Hi  John McGowan  -  

What version of #SleepyHead are you using?  You might try an older version, like 0.9.8-Testing.  Here is the link, SleepyHead - Version 0.9.8.1-Testing

Also, here is a thread from "Way back then" discussing the same problem.  I hope it may help.

[John McGowan]

Many thanks Crimson Nape.  Should I delete the present version and data before I reload your suggestion?

TIA
JMcG

[Crimson Nape]

I think I would delete the folders because the latter versions have implemented various "enhancements" since 0.9.8 that may trip it up.  If you're computer savvy, you may wish to delete the registry entry as well.  It is located under, "HKEY_CURRENT_USER\Software\Jedimark".

Good luck!
- Red

[Frodo]

There is a bug in Oscar code that prevents reading icon auto flow data since 1/1/2015.  If someone would fix the following line, it would read Icon Auto again. 

icon_loader.cpp line 846:

    if (ts > QDateTime(QDate(2015,1,1), QTime(0,0,0)).toTime_t()) {
        return false;
    }

It is supposed to eliminate some bad timestamps.
I use the following line instead:

    if (ts < QDateTime(QDate(2013,2,1), QTime(0,0,0)).toTime_t() || ts > QDateTime::currentDateTime().toTime_t()) { 
        return false;
    }

[pholynyk]

A better solution is to make the invalid date something too early - like 1 January 2010 - which happens at line 360, and then reverse the test in line 486 against the earlier date.

Change committed for the next release.

Thank you for bringing this to our attention, and suggesting a fix.

[GuyScharf]

I think you meant line 486 rather than 846?

What can you tell me about "bad timestamps" in the FLW file? Do you know the history, or what it is trying to eliminate? is 2/1/2013 some kind of magic date for the icon FLW file?

Your test may get confused by timezone or DST changes (we had the same test and a DST problem in the ResMed loader). I recommend you allow timestamps sometime in the future. Perhaps,

    QDateTime::currentDateTime().toTime_t().addDays(7)

I'll include your fix, or something similar, when I update icon_loader.cpp. (Icon_loader does try to process sleepstyle records, and fails utterly. I am breaking sleepstyle support into a separate sleepstyle loader. This should be available later this year.)

[Frodo]

Yes, line 486.
Feb 1 2013 is only a magic number for me because that is when I got my machine.  It should be the date when icons were first available.
Line 360 checks if the timestamp is a valid date, not a reasonable date.



The FLW timestamp can occasionally be off by many years.

The DET files are worse and can have several problems in one file:
    Random timestamp
    Index points past end of file
    Data block goes past end of file
    Garbage data (usually random data followed by a large block of FF)

I have also written some code to reject the DET errors

[GuyScharf]

Thank you for your participation! We are always overjoyed when people contribute to the OSCAR project.

We would be more than happy to include your code to reject DET errors. If you have any other improvements to the icon loader, let us know and we can include them if they are of general use.

[Frodo]

What is the best way to send you changes?