Problems with stupid answers

[Jarhyn]

So, just more of a complaint/bitch/rant thread for when we inevitably as software and hardware engineers run into insanity-inducong errors that have ridiculously stupid causes.

Anything from "I forgot a semicolon" to "something it was apparent I should fill in didn't need to be filled in, and in fact should not have been."

Example:

Recently in QT, there has been an issue where I am the only one on my team who has ever been able to debug into a shared object that is dynamically loaded at runtime.

Over the last several months, it has been pointed out that this is really fucking annoying. It reached a head this week when the engineer who will be taking over on firmware when I leave the company was unable to debug the software they are getting ownership of.

We spent all this week looking at everything, and it turned out to be that the compiler couldn't find "ld". Eventually I had them copy their debug logs and did a full diff on the log and the only difference was buried in a setting several layers in that isn't even a part of the project config: the sysroot was entered with a sane seeming value in theirs and mine was blank.

 

[rousseau]

I've been working with an EHR vendor (with a poorly developed toolset) for just over six years now. I'd bitch about it, but the absurd debugging scenarios I get into with it are the most fun I have at my job these days.

It's the problems that don't hurt my head enough that are the pain.

 

[Jarhyn]

I've been working with an EHR vendor (with a poorly developed toolset) for just over six years now. I'd bitch about it, but the absurd debugging scenarios I get into with it are the most fun I have at my job these days.

It's the problems that don't hurt my head enough that are the pain.

I know what you mean. Most days it's someone asking me to walk up to the surface of the void and pluck out for them stupid baubles like "ask it for a number and make a gated function call on over/under."

I live for the days it's something like "a machine that does phased linear regression and interpolates between regressions based on real-time measurements, and produces a number that is sane". I've been doing that for a year now, and it's finally done, and honestly, I feel almost listless.

 

[Jarhyn]

I swear, the number of times my own software engineer employees have to be asked whether it's plugged in...

 

[Swammerdami]

I've been working with an EHR vendor (with a poorly developed toolset) for just over six years now. I'd bitch about it, but the absurd debugging scenarios I get into with it are the most fun I have at my job these days.

It's the problems that don't hurt my head enough that are the pain.

I agree that the hardest-to-diagnose bugs can be interesting and exciting!

Is the following an example of what OP is looking for? I consider it to be a stupidity in the Windows interface. Perhaps it belongs in the "Why I detest Windows" thread.

There is a File pop-up that is ubiquitous in Windows: You find a file in a list and click on it. Easy-peasy. Or is it? If your fingers twitch a bit while clicking on a folder, that folder may be MOVED to become a subfolder of the folder adjacent in the list. If you don't notice this happening, that folder will seem to have disappeared, which can cause much grief.

WHY, in heaven's name, do they "need" to make it so easy to move a folder? This is not done deliberately very often; when it is done the user will be careful and methodical. Making it the default action for a finger twitch in the ubiquitous File pop-up is absurd.

Misclicking in that pop-up can also cause you to inadvertently edit the name of a file or folder but at least in that case you're likely to see what's happening.

 

[Jarhyn]

I've been working with an EHR vendor (with a poorly developed toolset) for just over six years now. I'd bitch about it, but the absurd debugging scenarios I get into with it are the most fun I have at my job these days.

It's the problems that don't hurt my head enough that are the pain.

I agree that the hardest-to-diagnose bugs can be interesting and exciting!

Is the following an example of what OP is looking for? I consider it to be a stupidity in the Windows interface. Perhaps it belongs in the "Why I detest Windows" thread.

There is a File pop-up that is ubiquitous in Windows: You find a file in a list and click on it. Easy-peasy. Or is it? If your fingers twitch a bit while clicking on a folder, that folder may be MOVED to become a subfolder of the folder adjacent in the list. If you don't notice this happening, that folder will seem to have disappeared, which can cause much grief.

WHY, in heaven's name, do they "need" to make it so easy to move a folder? This is not done deliberately very often; when it is done the user will be careful and methodical. Making it the default action for a finger twitch in the ubiquitous File pop-up is absurd.

Misclicking in that pop-up can also cause you to inadvertently edit the name of a file or folder but at least in that case you're likely to see what's happening.

I don't find it bad making it easy to move a folder. But it would be useful that any folder move would seek confirmation, or maybe have a screen where it takes a moment unless you're holding a mod key of some kind (shift/Ctrl/alt/combo thereof). Even just making differential fanfare when it happens, such as lighting up the folder that it got dropped in with a temporary glow, or even an accessible log.

The worst is when I do this, and don't even know which folder ate my other folder...

(The current answer to this is, in fact, to drop the mouse entirely and hit CTRL-Z when you make a mistake, or be a nerd and move stuff with CTRL-X/CTRL-V, assuming emacs hasn't ruined your hotkeying).

 

[steve_bank]

warning: sarcasm

As someone who has never known what it is like to make stupid mistakes I can understand how frustrating it is to have to deal with sup id mistakes.

Like somebody trying to apply a PID control algorithm without understanding the basic feedback control theory. You try and help when they obviously don't know what they are doing and they insist they know it all alrea puting out stupid resonses.

And of course I know for certain if everybody did things the way I do there would be no mistakes.

 

[Jarhyn]

warning: sarcasm

As someone who has never known what it is like to make stupid mistakes I can understand how frustrating it is to have to deal with sup id mistakes.

Like somebody trying to apply a PID control algorithm without understanding the basic feedback control theory. You try and help when they obviously don't know what they are doing and they insist they know it all alrea puting out stupid resonses.

And of course I know for certain if everybody did things the way I do there would be no mistakes.

You say this, but your lack of context is showing.

The problem turned out to be entirely related to effect scaling in the system which PID control just isn't capable of addressing.

Ultimately the problem was, in fact, solved by wrapping the PID output with a scalar modifier and move that along with the request.

As it is, the PID controller itself was already written. But sure, pretend to crow about the youngin's not listening to the oldies.

 

[Loren Pechtel]

I agree that the hardest-to-diagnose bugs can be interesting and exciting!

Well, I've got one that was just baffling:

Borland Pascal, in the later part of the era where you could write programs to run in protected mode and thus break the 640k wall. A protected mode program can't make any sort of driver or operating system call, they all must be thunked through real mode handlers. All the usual stuff was already supported but if you needed to call anything directly you had to write your own handler. They provided a library to assist such translations.

Ok, I have a piece of code that makes one of those unsupported calls. If it runs it runs correctly, but it usually throws an access violation. Single step through the code and it works 100% of the time. After much head-scratching I figure out there are two bugs involved. First, the debugger isn't actually single-stepping, but rather simulating everything. There's a slight flaw in the simulator: It's not verifying stores to segment registers, "executing" them even if they are invalid. Second, the library has a repeated flaw: The moron who wrote it used pointer types for pointers to real mode addresses. The pointers were never being dereferenced (they would only be used by the real mode code) but when faced with copying a 32 bit pointer the compiler emitted code that loaded it to a segmentffset pair and then stored it. If the segment part of it happened to be valid (it's a real mode pointer, no guarantee it's valid in protected mode) this worked, if it wasn't it crashed. Because of the debugger flaw it always worked if stepped through. I'm sure they never actually tried to use it, just tested that everything worked right when stepped through.

 

[Loren Pechtel]

I don't find it bad making it easy to move a folder. But it would be useful that any folder move would seek confirmation, or maybe have a screen where it takes a moment unless you're holding a mod key of some kind (shift/Ctrl/alt/combo thereof). Even just making differential fanfare when it happens, such as lighting up the folder that it got dropped in with a temporary glow, or even an accessible log.

The worst is when I do this, and don't even know which folder ate my other folder...

(The current answer to this is, in fact, to drop the mouse entirely and hit CTRL-Z when you make a mistake, or be a nerd and move stuff with CTRL-X/CTRL-V, assuming emacs hasn't ruined your hotkeying).

That's assuming you even realize you dragged something by mistake.

I've got one that's been annoying me--VMware supports tab tearing. I find it very easy to accidentally tear one off when I meant to switch machines.

 

[Jarhyn]

I don't find it bad making it easy to move a folder. But it would be useful that any folder move would seek confirmation, or maybe have a screen where it takes a moment unless you're holding a mod key of some kind (shift/Ctrl/alt/combo thereof). Even just making differential fanfare when it happens, such as lighting up the folder that it got dropped in with a temporary glow, or even an accessible log.

The worst is when I do this, and don't even know which folder ate my other folder...

(The current answer to this is, in fact, to drop the mouse entirely and hit CTRL-Z when you make a mistake, or be a nerd and move stuff with CTRL-X/CTRL-V, assuming emacs hasn't ruined your hotkeying).

That's assuming you even realize you dragged something by mistake.

I've got one that's been annoying me--VMware supports tab tearing. I find it very easy to accidentally tear one off when I meant to switch machines.

The tearaway click time and tear distance threshold probably needs changing.

Maybe it's user configurable? Good luck!

 

[Loren Pechtel]

I don't find it bad making it easy to move a folder. But it would be useful that any folder move would seek confirmation, or maybe have a screen where it takes a moment unless you're holding a mod key of some kind (shift/Ctrl/alt/combo thereof). Even just making differential fanfare when it happens, such as lighting up the folder that it got dropped in with a temporary glow, or even an accessible log.

The worst is when I do this, and don't even know which folder ate my other folder...

(The current answer to this is, in fact, to drop the mouse entirely and hit CTRL-Z when you make a mistake, or be a nerd and move stuff with CTRL-X/CTRL-V, assuming emacs hasn't ruined your hotkeying).

That's assuming you even realize you dragged something by mistake.

I've got one that's been annoying me--VMware supports tab tearing. I find it very easy to accidentally tear one off when I meant to switch machines.

The tearaway click time and tear distance threshold probably needs changing.

Maybe it's user configurable? Good luck!

I've never found any configuration for it. I don't want it at all.

 

[steve_bank]

warning: sarcasm

As someone who has never known what it is like to make stupid mistakes I can understand how frustrating it is to have to deal with sup id mistakes.

Like somebody trying to apply a PID control algorithm without understanding the basic feedback control theory. You try and help when they obviously don't know what they are doing and they insist they know it all alrea puting out stupid resonses.

And of course I know for certain if everybody did things the way I do there would be no mistakes.

You say this, but your lack of context is showing.

The problem turned out to be entirely related to effect scaling in the system which PID control just isn't capable of addressing.

Ultimately the problem was, in fact, solved by wrapping the PID output with a scalar modifier and move that along with the request.

As it is, the PID controller itself was already written. But sure, pretend to crow about the youngin's not listening to the oldies.

I am not crowing. Electronics and feedback control do not appear to be your strong suit. Same with cell phones and radio theory. The way you described thecontrol problem in terms of old style metaphysics is a dead give away.

When you said the temperature was 2x what was expected it pointed to to scaling the feedback signal, which is why I pointed you to the general feedback equation. In a negative feedback system divide the feedback signal by 2 and the output is 2x.

The 'kludge and tweek' approach. Throw something together and fiddle with it until it works. Empirically adjust parameters until it works without knowing why it works. Technician level. Been there. As a tech I realized there was a limit to what I could do without more theory. So I went back to school.

A rhetorical question. Do you feel under appreciated and under valued by peers and mnagers?

 

[Jarhyn]

When you said the temperature was 2x what was expected it pointed to to scaling the feedback signal

Except it wasn't. And you weren't reading the post. Or I wasn't explaining it clearly.

There was an effect that the sensor could not possibly see, that was nonetheless something needing accounting-for. PID control can't target with that which cannot be seen.

As it is, I feel exactly as appreciated as I spend the time to earn, which is commensurate to what they are paying me. There's exactly one "peer" I have if you could call them that that has built their career on cribbing the work of others, but I regularly have the opportunity to serve them crow, so it's something I just kind of deal with.