SleepyHead Qt v0.8 alpha

archangle wrote:

glenj wrote:Hmmm, tried running the latest version(0.8.2) twice today and both times AVG jumped in claiming malware and identity theft. Now it is blocking me from copying the '.exe' at all.

Spybot Search and Destroy says that libstdc++6.dll has Fraud.IntenetSecurity2011. Also QtOpenGL4.dll.

It's in the Heuristic section, so that's a little bit less certain, but still worrying.

Comodo and MalwareBytes did not have any problem with SleepyHead files.

Edit - Also phonon4.dll.

It did not object to the .exe file.

Avira Antivir is okay with it too, at least with a reasonable choice of options. (With everything turned on, AA becomes paranoid to the point of uselessness.)

One more point of info.

I found an old copy of phonon4.dll in another directory. Dated 11/06/2010. Same message from S&D.

Also ran Windows Defender on the files. No virus messages there.

I'm feeling pretty confident it's a false positive. I trust Spybot S&D a lot more than the other antivirus programs. I'll see if I can figure out how to give them feedback.

The fact that Malwarebytes found nothing gives me a lot of confidence. That's a really good AV scanner.

I don't have much faith in AVG. I dumped it in favor of Comodo a long time ago due to too many false positives from AVG. AVG has also had false positives that would sometimes remove important Windows files and renders your PC unbootable.

I'm terrified of these fake antivirus trojans. They are nasty and being spread by professional criminals, not script kiddies. I've had several friends call me begging for help fixing their PCs after they caught one of these.

I'm pretty confident this is a false positive. i.e. That the problem is in the antivirus software, not in SleepyHead.

Speaking of sleep, CPAP, etc., it's getting so late it's early around here. Sweet dreams and low AHI to all.

archangle wrote:

jedimark wrote:I just discovered Qt SDK comes with a git client on windows.. so it literally has everything needed to build sleepyhead from source code...

I'll see if I can get a chance tomorrow. Downloading now. Should sleep soon.

Do the Qt*.dll files come from the SDK from Nokia? Would I expect to get the same .dll file from there? i.e. should it scan as "bad" as well?

And you're not kidding. It IS freakin huge. I'm downloading the offline just because I prefer everything saved away safe and sound on my own PC.

By the way, one other security comment. I'm running Comodo, and it has a "sandbox" feature. It automagically sandboxed SleepyHead and the program still runs fine.

The DLL's come with the SDK.. You don't even have to worry about them when running via QtCreator..

Qt SDK literally comes with everything needed to write software using Qt.. It's fantastic, but huge..

jedimark wrote:Find folder, as in before or after you click ok? Import is very slow now.. It also does a lot more nifty indexing stuff for later on.. If it's taking a while before the folder select box comes up it may be a missing DLL.. I'll see if I can dig out what it's looking for..

I click file>import data, and the title bar and outline of the dialog box come right up, but the center of the dialog remains transparent for about six seconds. Call the dialog up again, and the same thing happens. 0.8.1 was the same way, but I didn't mention it because I knew you were making major changes.

I forgot to mention Preferences "Sessions" tab now works..

I like the longest gap idea... a little tricky to visualize how to accomplish that though.. (technically I could extend the Combine scale a bit longer and I think it will achieve a similar goal, although I think your referring to not using the split time at all.. hmmm.. I'll have a think on that one)

24 hour days with a single main chunk of sleep are part of sleepyheads design, and it's not easy to have a calendar model and split strange ways.. I myself don't keep a normal 24 hour day myself, and nap alot, but SH has been behaving well for me. Most people sleep somewhere between 12pm and 12am (I know that sounds silly, but I mean the whole sleep "window" falls between these times) My wife works in admin at the local ER and has 3 different shift times, so I'm not sure how well someone like her would be catered for..

Here's a set of rules for the option I'm thinking of.

_Each machine session gets assigned to a calendar day. Sessions are never split (except maybe for extremely long sessions with near 24/7 usage).

_The longest waking (or at least not hosing) period in 24 hours determines the date split.

_For determining the waking period only, optionally ignore naps less than a user-selected length and isolated by a selected number of hours from other machine sessions.

This will need a combine setting too so turning off the machine for a moment to refit the mask doesn't turn your half hour nap into something more than it is. I'm not sure that really needs to be a slider, though. Maybe just for people who are geek enough to select "show advanced options". Perhaps this entire dialog should have a simplified version with defaults that will work for most.

_The waking period attaches to whichever side of the user selected split time most of its hours fall in.

_All sleep sessions between main waking periods are assigned to date of the previous or following waking period, according to user preference.

_User settings determine what to do with naps that were ignored above.

_When new data is loaded, all sessions for 48 hours on either side are re-evalutated to see if they're still on the right day. If something gets moved, expand the area of evaluation recursively until everything is in the right place. It's possible this shift could go back several days, but I doubt that would happen very often.

This plan has a lot of settings to fuss with, but once it's set up, I think it would present data coherently for someone who does not have a regular schedule. Most people will sleep in a particular window most of the time, but for folks with circadian disorders or people like your wife who have bizarre shifts, a fixed window makes it harder to interpret the data.

I think the expanded combine slider would be pretty good. Start taking notes, and the fixed window might get to be more of a hassle. Add a way for users to quantify aspects of their day (sleepiness, alertness, mood, fog, etc), and try to graph the resulting numbers alongside sleep stats, and things start to get very messy for someone who doesn't stay in a fixed window.

In case you haven't run across non-24-hour circadian rhythm disorder before, here's why I would really like to see SleepyHead have something like the above option.
http://en.wikipedia.org/wiki/Non-24-hou ... e_syndrome

Thanks for noticing memory use.. Qt is a big fat hog, but at least I know I'm not being one too. SH no longer has to convert the data to a different data format to display in the graphs.. Plus it doesn't waste an event record for every event.. I grouped each sessions events together by type, and saved a lot of space doing this.

The previous version was Hogzilla, but it was also a pre-alpha. If anything with that designation runs at all, that's good. If it's actually useful, that's outstanding. I was pretty sure you weren't the sort to actually release a monster pig, but I'm impressed at how much leaner 0.8.2 is.

On the version numbers, yeah, 0.9. But you're running out of numbers and still making major changes, so I don't blame you for keeping this 0.8.x. It'd be nice to have all the basics set by 0.9.0.

I just noticed a new bug while playing with the split and slider settings: some days don't have the time below the event flags chart. Two days seem never to display the time, and two more display it sometimes. Restarting doesn't fix this, but it does sometimes change whether the intermittently broken days display the time or not.

Otter wrote:I click file>import data, and the title bar and outline of the dialog box come right up, but the center of the dialog remains transparent for about six seconds. Call the dialog up again, and the same thing happens. 0.8.1 was the same way, but I didn't mention it because I knew you were making major changes.

What version of windows are you running again?

Here's a set of rules for the option I'm thinking of.

_Each machine session gets assigned to a calendar day. Sessions are never split (except maybe for extremely long sessions with near 24/7 usage).

_The longest waking (or at least not hosing) period in 24 hours determines the date split.

_For determining the waking period only, optionally ignore naps less than a user-selected length and isolated by a selected number of hours from other machine sessions.

This will need a combine setting too so turning off the machine for a moment to refit the mask doesn't turn your half hour nap into something more than it is. I'm not sure that really needs to be a slider, though. Maybe just for people who are geek enough to select "show advanced options". Perhaps this entire dialog should have a simplified version with defaults that will work for most.

_The waking period attaches to whichever side of the user selected split time most of its hours fall in.

_All sleep sessions between main waking periods are assigned to date of the previous or following waking period, according to user preference.

_User settings determine what to do with naps that were ignored above.

_When new data is loaded, all sessions for 48 hours on either side are re-evalutated to see if they're still on the right day. If something gets moved, expand the area of evaluation recursively until everything is in the right place. It's possible this shift could go back several days, but I doubt that would happen very often.

This plan has a lot of settings to fuss with, but once it's set up, I think it would present data coherently for someone who does not have a regular schedule. Most people will sleep in a particular window most of the time, but for folks with circadian disorders or people like your wife who have bizarre shifts, a fixed window makes it harder to interpret the data.

I think the expanded combine slider would be pretty good. Start taking notes, and the fixed window might get to be more of a hassle. Add a way for users to quantify aspects of their day (sleepiness, alertness, mood, fog, etc), and try to graph the resulting numbers alongside sleep stats, and things start to get very messy for someone who doesn't stay in a fixed window.

It'll take me a while to parse all that.. I'll get back to you later on it..

In case you haven't run across non-24-hour circadian rhythm disorder before, here's why I would really like to see SleepyHead have something like the above option.
http://en.wikipedia.org/wiki/Non-24-hou ... e_syndrome

I'm a poster child for screwed up circadian rhythms.

The previous version was Hogzilla, but it was also a pre-alpha. If anything with that designation runs at all, that's good. If it's actually useful, that's outstanding. I was pretty sure you weren't the sort to actually release a monster pig, but I'm impressed at how much leaner 0.8.2 is.

Lol. I always came in third place to Firefox & Gnome Shell, so I never felt bad about it.. Try running a 3D game some time and checking memory usage.

On the version numbers, yeah, 0.9. But you're running out of numbers and still making major changes, so I don't blame you for keeping this 0.8.x. It'd be nice to have all the basics set by 0.9.0.

That's kinda why I didn't.. I want 0.9 to be the beta..

I just noticed a new bug while playing with the split and slider settings: some days don't have the time below the event flags chart. Two days seem never to display the time, and two more display it sometimes. Restarting doesn't fix this, but it does sometimes change whether the intermittently broken days display the time or not.

Hmm... That's a new one.. I've just done some more work on the XTickers, so it may be fixed already.. For now, check the debug messages and see if they say anything.. There's a few conditions that will explain it.. (Namely very short sessions)

jedimark wrote:

Otter wrote:I click file>import data, and the title bar and outline of the dialog box come right up, but the center of the dialog remains transparent for about six seconds. Call the dialog up again, and the same thing happens. 0.8.1 was the same way, but I didn't mention it because I knew you were making major changes.

What version of windows are you running again?

XP pro x64. It's basically 64-bit server 2003 wearing an xp pro costume.

Here's a set of rules for the option I'm thinking of. . . .

. . . . and things start to get very messy for someone who doesn't stay in a fixed window.

It'll take me a while to parse all that.. I'll get back to you later on it.

No rush.

I always came in third place to Firefox & Gnome Shell, so I never felt bad about it..

At least you made the podium.

I just noticed a new bug while playing with the split and slider settings: some days don't have the time below the event flags chart. Two days seem never to display the time, and two more display it sometimes. Restarting doesn't fix this, but it does sometimes change whether the intermittently broken days display the time or not.

Hmm... That's a new one.. I've just done some more work on the XTickers, so it may be fixed already.. For now, check the debug messages and see if they say anything.. There's a few conditions that will explain it.. (Namely very short sessions)

I have a bunch of these:
FitTicks==0!

Otter wrote:I have a bunch of these: FitTicks==0!

FitTicks happen when your zoomed in to close, or the sessions are way to short.. I changed the error message to nag, and fixed a few things with this bit.
How long are the sessions on the days your seeing this?

64bit XP probably is a bit dodge as far as running 32bit apps go. I claim to be no expert on windows platform, but I have quite a few painful memories of it.

I've got to wait till my wifes dead ASUS notebook gets back to debug 64bit windows stuff a bit better. (ASUS repaired it, but sent it back with Vista installed on it instead of Win7 it originally came with.. had to go back to the factory to unstuff it up.. I couldn't use my disks or license key.. They claim volume licensing issue, I claim dodge on their part...)

It could be a general 64bit problem. I've been lazy again and using my old 32bit P4 Dell notebook cos it's comfy.. Haven't set foot in my office for weeks.

Here goes Otter, my brains fuzzy

Otter wrote:Here's a set of rules for the option I'm thinking of.
_Each machine session gets assigned to a calendar day. Sessions are never split (except maybe for extremely long sessions with near 24/7 usage).

The machine itself handles splitting the reeeaalllllly long sessions.. I don't physically have to cut any sessions.

_The longest waking (or at least not hosing) period in 24 hours determines the date split.

_For determining the waking period only, optionally ignore naps less than a user-selected length and isolated by a selected number of hours from other machine sessions.

This will need a combine setting too so turning off the machine for a moment to refit the mask doesn't turn your half hour nap into something more than it is. I'm not sure that really needs to be a slider, though. Maybe just for people who are geek enough to select "show advanced options". Perhaps this entire dialog should have a simplified version with defaults that will work for most.

_The waking period attaches to whichever side of the user selected split time most of its hours fall in.

_All sleep sessions between main waking periods are assigned to date of the previous or following waking period, according to user preference.

_User settings determine what to do with naps that were ignored above.

I still like the idea, but it definitely is complicated. My biggest worry is will the user need a maths degree to understand it?

_When new data is loaded, all sessions for 48 hours on either side are re-evalutated to see if they're still on the right day. If something gets moved, expand the area of evaluation recursively until everything is in the right place. It's possible this shift could go back several days, but I doubt that would happen very often.

This would basically be implemented as a resort after Importing data.. The same function would run every time the app loads, or when the preference values are adjusted.

This plan has a lot of settings to fuss with, but once it's set up, I think it would present data coherently for someone who does not have a regular schedule. Most people will sleep in a particular window most of the time, but for folks with circadian disorders or people like your wife who have bizarre shifts, a fixed window makes it harder to interpret the data.

It's definitely something I'm interested in exploring..

I think the expanded combine slider would be pretty good. Start taking notes, and the fixed window might get to be more of a hassle. Add a way for users to quantify aspects of their day (sleepiness, alertness, mood, fog, etc), and try to graph the resulting numbers alongside sleep stats, and things start to get very messy for someone who doesn't stay in a fixed window.

With better splitting this stuff won't really be a problem..

The extra's tab stuff will end up graphed. I don't see a problem with it with a more tailored split algorithm. Most of it is per "day" stuff and would go in the overview (Or perhaps a dedicated page/tab/etc..)

Anyway, this is stuff I'm definitely considering.. It may end up taking a month or two till I can get a clear head and wrap my brain around it.. Some of this I've already attempted before and had to roll back on.. Fog & fuzz is no fun for a programmer..

jedimark wrote:I'm still looking for an semi-advanced windows person or two to help test source code from GIT.. The free Nokia Qt SDK has all of what's needed to build.. (The free Tortise Git client is also needed to download the source code and keep it up to date.) Qt SDK is a quite a huge download.. Everything can be built via graphical interface. I really don't want to release another crashy windows build without some serious testing on this platform first. I've put you guys through enough torture..

My home internet connection is ghastly show and I am presently boycotting anything resembling software development outside work hours but when you have an import for the S9 VPAP Auto, I am happy to do some testing on a build before you release it. I am running 64 bit Windows 7 and can also test in the XP virtual machine I use for ResScan since work's security settings won't let the latest ResScan install on Win7

fiberfan wrote:

jedimark wrote:I'm still looking for an semi-advanced windows person or two to help test source code from GIT.. The free Nokia Qt SDK has all of what's needed to build.. (The free Tortise Git client is also needed to download the source code and keep it up to date.) Qt SDK is a quite a huge download.. Everything can be built via graphical interface. I really don't want to release another crashy windows build without some serious testing on this platform first. I've put you guys through enough torture..

My home internet connection is ghastly show and I am presently boycotting anything resembling software development outside work hours but when you have an import for the S9 VPAP Auto, I am happy to do some testing on a build before you release it. I am running 64 bit Windows 7 and can also test in the XP virtual machine I use for ResScan since work's security settings won't let the latest ResScan install on Win7

I've currently got zero test data sets for the S9 VPAP. It's only edf, so it's a total cinch to add support.

jedimark wrote:How long are the sessions on the days your seeing this?

Looking at one day, the shortest I see is about 20 min. Another has one that's only three minutes long.

64bit XP probably is a bit dodge as far as running 32bit apps go. I claim to be no expert on windows platform, but I have quite a few painful memories of it.

It emulates 32bit very well, provided the 32bit code is actually all 32 bit. 16bit code simply won't run, and in the bad old days, many 32 bit aps had some 16bit legacy code. Also, when the OS came out, many people were still using 16bit installers. But other than that, I've never had any application issues.

If there is any 32/64 trouble, it's always about drivers. When xp x64 was new, the drivers were a nightmare, though, and many people still haven't gotten over that. Windows in general is pretty dodgy. But this particular OS is more solid than anything from MS I've run. Of course, that's really not saying much.

I've got to wait till my wifes dead ASUS notebook gets back to debug 64bit windows stuff a bit better. (ASUS repaired it, but sent it back with Vista installed on it instead of Win7 it originally came with.. had to go back to the factory to unstuff it up.. I couldn't use my disks or license key.. They claim volume licensing issue, I claim dodge on their part...)

I'm not a fan of ASUS. Everyone says their hardware is great, but many years ago, I got a display driver update from them that took out my keyboard until I reinstalled the OS. (I had a backup, but there was a problem with it.) Their front line tech support admitted what had happened, but second tier refused to call me back. Maybe they couldn't figure out how to fix it either. I suspect that at least part of that display driver package was meant for one of their motherboards, which I didn't have, and I've avoided ASUS ever since.

It could be a general 64bit problem. I've been lazy again and using my old 32bit P4 Dell notebook cos it's comfy.. Haven't set foot in my office for weeks.

I tried running SH in XP compatibility mode, but it didn't make any difference

jedimark wrote:I still like the idea, but it definitely is complicated. My biggest worry is will the user need a maths degree to understand it?

I think it could be greatly simplified, though at the expense of configurability that someone is sure to want. The outline I gave you allowed customizing pretty much everything, but most people wouldn't use all those options. With a carefully thought set of defaults, the whole thing could probably be reduced to a "track my sleep pattern" radio button. Just being able to graphically see the length of each sleep cycle in a month would be really cool. I still like the idea of advanced preferences, but there should be a dialog specific way to restore the defaults.

I think the expanded combine slider would be pretty good. Start taking notes, and the fixed window might get to be more of a hassle. Add a way for users to quantify aspects of their day (sleepiness, alertness, mood, fog, etc), and try to graph the resulting numbers alongside sleep stats, and things start to get very messy for someone who doesn't stay in a fixed window.

With better splitting this stuff won't really be a problem..

My concern is that if your day has drifted 12 hours from what it was a couple weeks ago, somewhere you'll have "how was your day" data pairing up with the sleep that came after. Or you'll get caffeine data showing that eight cups of coffee keeps you awake the night before you drink it. I'm fuzzy too, though. Maybe I'm missing how the expanded sliders would solve this.

The extra's tab stuff will end up graphed. I don't see a problem with it with a more tailored split algorithm. Most of it is per "day" stuff and would go in the overview (Or perhaps a dedicated page/tab/etc..)

That's what I"m getting at. If your day sometimes starts with waking up, and sometimes starts with falling asleep, the effect of your sleep, or lack thereof, will sometimes show up on the same day as the sleep that produced it, and sometimes on a different day. To track the effects of things like coffee on sleep, and things like increased pressure on waking performance, we need to know not only what day things happened, but what order they happened on that day. My solution is to shift the concept of "day" to match the user's experience, so that it's always obvious which sleep data points are affecting which waking data points, and vice versa.

Anyway, this is stuff I'm definitely considering.. It may end up taking a month or two till I can get a clear head and wrap my brain around it.. Some of this I've already attempted before and had to roll back on..

Thanks for thinking about my suggestion, but don't use up too many brain cells right now unless it really grabs you. The stuff we're talking about comes under the heading of advanced features, and while thinking about it now might save you another regut later, actual implementation is probably best left until after the first stable release.

What I suggest right now is leaving the door open for "days" that are more or less than 24 hours. That will allow the flexibility for some sort floating split analysis when the time is right.

Fog & fuzz is no fun for a programmer..

I know more about this than I'd like to. I know that you're still trying to dial in your own therapy, and I'm somewhat amazed at what you're able to get done even through the fog.

jedimark wrote:I've currently got zero test data sets for the S9 VPAP. It's only edf, so it's a total cinch to add support.

I will be sending you data on Monday, given the time difference and my delayed sleep cycle, that will be your Tuesday.

As long as I have an option to do a dumb "day starts and ends at noon, dammit" option, if I don't like SH's smart day splitting algorithm, I can muddle through somehow.

1) Do you intend to indicate at some time whether the machine is in Auto/manual/CFLEX/AFLEX mode?

2) I need to do some further confirmation, but it looks like if the PRS1 Auto machine is on manual CPAP, the daily report indicates zero pressure in the graph and in the window on the left.

Both would be useful if you're tinkering with manual/auto/pressure/flex settings to find the right settings.