Searching for the bug ... but there isn't any

[jkon]

There are some times in my life that I am sure there is a bug in my code but it turns up that the bug was in the language or in the system. This thread is talking about those times. Please share your own relevant experience.

I ll start sharing my today adventure that started at 2:00 am and ended at 11:00 am .If the problem is in the language there are some techniques to realize it after a while , I am creating a WebSocket application (with our own framework that has a WS layer) with OOP PHP server side and JS client side ( debugging is a bit strange in those cases) . The app is almost finished , but in 2:00 am I realized that even the basics stopped to work , client side the error messages (that didn't helped) seemed to have something to do when there where content of external iframe (like youtube) or even inline onmouseover events generated through WYSIWYG. Sever side there where no error at all and it was like the client closed the connection normally.

I thought that I have made some changes the last days but not checking those extended (luckily I kept backup both sides). So I started to move each layer out (one by one, and the PHP framework after the app... ) … till there was only the core in both sides. Client side I was using Firefox with Firebug (nothing new to that since till now I first develop in that environment and then check the others , the other environments had also been checked). Just a while before going to sleep feeling really awful I just tested in Chrome developer tools to see if there were any other messages , and … it was working, I deactivated Firebug in Firefox and it was working in Firefox as well.

I had a colleague some time ago that he always said PHP must have an error to that , or JS is not working or Firebug has a problem , Chrome developer tools don't work. All that for not recognizing that it is more probable to have errors in his code. But today I learned that after X hours , don't trust no one ;) ;)

[jwenting]

it happens. Just as your code isn't immune from bugs on release, neither is any other code.
Last month I ran into a major problem writing some JPA code, whatever we tried we couldn't get it to work properly.
Some research into the extremely obscure error messages we got (when catching runtime exceptions that would otherwise be hidden by the application server) pointed me towards a bug in Eclipselink that had been known about and reported 7 years ago and never yet fixed.

Years ago I ran into a bug in a Borland Pascal compiler, which caused a system call to reproducibly fail.
Even weirder, this bug was present in one specific version, but neither in the previous nor in the next version.

[Dani]

Every. Single. Time that someone installs a buggy browser plugin/toolbar that screws up DaniWeb, complains about it, and then I spend hours trying to isolate which users have a problem that I can't reproduce.

[Reverend Jim]

I found a bug in the PL/1 compiler (verified by my professor) that caused me to had in an incomplete project. I spent two days trying to figure out what I was doing wrong until I finally approached my professor.

Anyone remember PL/1?

[Agilemind]

I use so many libraries that fail completely as soon as you try to use them on non-standard dataset and I end up spending hours trying to find/fix where they have made an assumption that isn't mentioned in the manual/documentation or what the unmentioned limitations are.

Heh heh. I produce so many of my own bugs, everybody else's just fade into background noise. I remember the one that got me was certain DateTime (PHP) function parameter values, which failed to run on Windows (think it was a VC6 vs VC9 issue), but did just dandy on *nix. I went out of my head for days until I found some obscure, 20-deep reply forum posts about it.

[jwenting]

Anyone remember PL/1?

never used it, was using Cobol and C on Tandem at the time, but do remember the frustrated PL/1 zombies prowling the halls at night :)

I found a bug in the PL/1 compiler (verified by my professor) that caused me to had in an incomplete project.

Once found a bug in an exam that would have caused half the students to fail. Prof acknowledged it, question wasn't counted if people got the answer wrong.

[Slavi]

Speaking of this (from this week)!!
I made a python script to start multiple excel macros (located in different excel files), manipulate some of the data and finally upload it to an elasticsearch database running on one of the servers at work. All good and great .. made a batch file to start the script and scheduled it to run every morning. At that point thing's just got really really .. i mean REALLY nasty .. because when the batch file was ran as a scheduled task it wouldn't start the excel macros. After spending insanely lots of time on this i found the following:
Please make this folder. C:\Windows\SysWOW64\config\systemprofile\Desktop
On microsoft forums. Like are you kidding me???? By default the folder 'Desktop' doesn't exist in that location since server2k8 ..
Yep, windows.

[jwenting]

sounds like a very old Excel file with macros written to expect a hardwired directory?

[Slavi]

Excel 2013, apprently microsoft just decided to get rid of that folder
Found the solution here ...