Possible bug: Can't close Untitled.exe sometimes

[STxAxTIC]

Been running the same Win10 machine a few years, the only variable is the new QB64... twice today I haven't been able to close an exe window after trying to close it. Anyone else?

[SMcNeill]

Do you have an _EXIT in your program anywhere?  Or $CHECKING:OFF?

[STxAxTIC]

I dont use underscore commands except for _Width typically. No fancy meta stuff. It could still be my OS causing the issue, just fishy how sudden....

[FellippeHeitor]

From which qb64 version did you switch to the latest? We have replaced the whole c++ compiler with 1.3, it may be the case of whitelisting the whole folder again (antivírus? Windows defender?).

And since it's Windows 10, didn't it update itself sometime around the time you updated QB64? Then it wouldn't be the only variable.

[STxAxTIC]

All things considered, the new QB64 is the only variable. The facts:

i) I've been using this same computer the same way since the day I had it. Antivirus off, firewall off, training wheels off.

ii) Since I can only imagine other people are also using Windows 10, and are subject to the same updates as I am, perhaps on the same brand laptop for some people. So that makes it not my personal variable, statistically, in a weird way. Something about the rising tide raising all ships... The only thing I did all by myself was install the new QB64, from the previous version. Maybe 2 months old.

iii) You all know I don't do any exotic QB64 stuff - all QB45 commands whenever possible, rite?

This "crash" happened while making my last two programs:

https://www.qb64.org/forum/index.php?topic=2143.0

https://www.qb64.org/forum/index.php?topic=2164.msg114120#msg114120

[FellippeHeitor]

Do you still have the older version there? Does untitled compiled with it still close immediately?

The bulk of 1.4 additions and fixes happened in the past two months and it's been a lot of changes. We'd have to pinpoint which older version didn't cause the issue on your machine. Steve is on W10 and it seems ok there.

Would you be able to keep going back in versions and test them locally?

[STxAxTIC]

Oh jeez - thats a needle in a haystack, would probably need to automate that test. In the past 5 days, I've probably attempted to close a program 500 times and it's worked 498, I may never see a pattern. All I know is I downloaded the latest about a week ago, and the time before that was a few months. Deleted old version on download.

[FellippeHeitor]

Cloning the repository locally and checking out each automated commit labeled "Update internal/source" by Luke and then running setup_win.bat would take you step by step without having to download each version in full. https://github.com/QB64Team/qb64/commits/development

[STxAxTIC]

All true - but actually *doing* the test will need better tools. For me at least... While we have many retirees at this site I only have time to write a program to open and close windows if I were to tackle this.

Pretend it has a 1% chance of happening. How many times should each version be checked before it passes? Just 100? Okay, suppose it happens. We're still far from blaming either the OS or QB64. I would only be satisfied by a Bayesian analysis. If it exists, bugs like this can't be hunted like looking for Sasquatch, it should be done very systematically.

EDIT

Bayesian analysis, a method of statistical inference (named for English mathematician Thomas Bayes) that allows one to combine prior information about a population parameter with evidence from information contained in a sample to guide the statistical inference process.

[FellippeHeitor]

You lost me there.

[STxAxTIC]

Oh, where are my manners? The quote is from Wikipedia.

[bplus]

I think STxAxTIC is saying he can't replicate the problem 100% of the time, no, more like a fraction of a percent 2/500 = .4%  << Edit: from .04%

Yeah, that is about the number of times a program fails to close leaving a Screen hanging and I have to go into Windows Task Manager and close it myself. It is far too flukey to figure out what happened because I am always modifying this and that.

[Cobalt]

I'm going to call OS BS here, I've been using 1.3DB and 1.4RC for a while now and have never had an issue with a program not wanting to go away, go away too soon maybe, but never being stubborn about closing.

I'll pull those programs from the posts you listed and see If I can ever get a hangup on close up.

[Cobalt]

Yep just compiled and ran it 250 times. And I know it was 250 times cause I hit space on the same line then F5 until I was 250 spaces over. not once did it give any resistance to closing. used your "Ellipse Intersect Line" code for the test. now that I've wasted a quarter hour back to Dragon Warrior!which has been compiled and ran countless times with no closing issues... premature closing issues but thats my fault.

so in recap, with 90% certainly, this is most likely OS BS or hardware+OS BS.

if you want to blame QB64 for something blame it for your code not working how you want, cause you know as programmers our code is 100% flawless!  :D

[bplus]

Yep definitely could be operator, might more likely happen when run a mod, forget to close it, make a mod, run that, close that, make a mod run that close that... the old run still up keeps getting it's exe name closed and modified.