Debug logs time format

[simonc]

Probably better to have another field, I reckon the date isn't terribly important.

<4739496.666 09:58.03.324> [eDVBSectionReader] DMX_SET_FILTER pid=18

or

<4739496.666> 09:58.03.324 [eDVBSectionReader] DMX_SET_FILTER pid=18

The inclusion of the human readable time could be optional - the timestamp is on a flag

My only question is whether the < > surrounding the timestamp is significant for any existing tools. Can someone with a bit more Linux background comment?

[smipx]

Agree about the date. I was just thinking that doing it every 10 or so entries would stop the actual debug info being shoved too far over to the right
Any option would be fine by me :-)

[ccs]

........ deleted.

[birdman]

What prompted this question is that I was getting a spinner for 2 seconds on the hour every hour and I was intrigued to try and discover what it was that was doing it. No a crash so no crash log (and thus no time stamp at the beginning of the log file). I thought it would be a simple case of looking at the debug log and looking at the correct times each hour but it turned out to not be so simple so I kind of gave up.l

The log contains an entry for showing the spinner when it does so. Just look for "spinner".
Or look at the log when (just after) the spinner has been showing.

[birdman]

My only question is whether the < > surrounding the timestamp is significant for any existing tools. Can someone with a bit more Linux background comment?

It make parsing out timestamps simpler if you wish to do so (there may be debug lines without them - the current debug line contain some file-system device entry info near the start that someone has left in place for some reason - or just forgotten about....

[Huevos]

How about this...

<03:08:18.3133> [ABM-Scheduler][Scheduleautostart] reason(0), session <__main__.Session instance at 0xb20ef490>
<03:08:18.3135> [ABM-Scheduler][Scheduleautostart] AutoStart Enabled
<03:08:18.3137> [ABM-Scheduler][AutoScheduleTimer] Schedule Disabled at Sun 12 Apr 2020 03:08:18 CEST
<03:08:18.3139> [XMLTVImport] autostart (0) occured at 1586653698.31
<03:08:18.3147> [XMLTVImport] WakeUpTime now set to -1 (now=1586653698)
<03:08:18.3152> [JoyneScan-Scheduler][Scheduleautostart] reason(0), session <__main__.Session instance at 0xb20ef490>
<03:08:18.3153> [JoyneScan-Scheduler][Scheduleautostart] AutoStart Enabled
<03:08:18.3155> [JoyneScan-Scheduler][AutoScheduleTimer] Schedule Disabled at Sun 12 Apr 2020 03:08:18 CEST
<03:08:18.3156> [ImageManager] AutoStart Enabled
<03:08:18.3157> [ImageManager] Backup Schedule Disabled at (now=Sun 12 Apr 2020 03:08:18 CEST)
<03:08:18.3158> [BackupManager] AutoStart Enabled
<03:08:18.3159> [BackupManager] Backup Schedule Disabled at (now=Sun 12 Apr 2020 03:08:18 CEST)
<03:08:18.3411> [Skin] Processing screen 'Screensaver', position=(0, 0), size=(1920 x 1080) for module 'Screensaver'.

https://github.com/Huevos/enigma2/co...66b9c300e78dcf

[abu baniaz]

Looks good. Can we try adding a date? No need for year, IMO.

[Huevos]

Not sure of the value of the date.

Options are here:
http://www.cplusplus.com/reference/ctime/tm/

[Huevos]

Date without year is a problem as it is not clear if it is mm/dd or dd/mm.

Also, adding pointless chars to the left of the log means the log is right shifted maybe making it necessary at times to scroll horizontally to read it.

[Huevos]

Also this has a big negative point which is, if you are trying to determine the time difference in seconds between one event and another having the log time in seconds is much easier to cope with.

e.g. if I want to know how long a blindscan took to complete I just subtract one stamp from another. In clock format that is not possible.

Probably better to have another field, I reckon the date isn't terribly important.

<4739496.666 09:58.03.324> [eDVBSectionReader] DMX_SET_FILTER pid=18

or

<4739496.666> 09:58.03.324 [eDVBSectionReader] DMX_SET_FILTER pid=18

The inclusion of the human readable time could be optional - the timestamp is on a flag

If we go with stamp and date we only need the u_sec on one field.

[simonc]

I got as far as this yesterday https://github.com/SimonCapewell/eni...-debug-logging, but as ever, getting the C++ build working again after an absence seems to be a black art. Main thing here is that you can chose to have either format or both.

[Huevos]

Simon can you make this configurable from the UI? If not how do we change from one to the other?

What box are you using? ARM or MIPS?

[simonc]

It'd be an environment variable, so perhaps the UI would be able to manipulate the current environment value. If not, you should be able to set it in the shell before starting the e2 executable.

I think mine is a MIPS box - it's an ET8500

[Huevos]

That variable is static ( static int debugTime = 3; ). Does that mean it is immutable? I have no idea of the rules of cpp.

Could be set in the shell script? The shell script can read the settings file to decide what value to send to enigma. That is more or less how debug logs are enabled/disabled at the current time.

[simonc]

Static just means it's shared across all instances, like Python's

class MyClass:
    myStatic = 5

    def __init__(self):
        print MyClass.myStatic
        myInstance = 23

Yes, you should be able to test the flag from the shell using

export ENIGMA_DEBUG_TIME=0
enigma2