Thread: I never want to see another undefined symbol error in my life, EVER

I've "solved" the problem now, but I really feel like a massive rant!!! Sorry guys. I've spent the last few hours fighting with an unresolved symbol error. I'm much more a C programmer than C++, so I was staring at the mangled name (and unmangling it to see if that helped) for AGES.

Then I tried scrubbing my whole project and pasting the files back in. I pored over the highest verbosity of linker logs looking for a clue.

Finally, I became convinced that it must be Visual Studio at fault -- ridiculous right, but I used to be a software tools engineer so I know very vididly that compilers aren't ironglad.

So I decided to try MingW, which nearly had me throwing my computer across the room (I'm really not an angry person.... just been a rough morning).

So then I remembered that of course, my computer dual boots so I have a perfectly happy version of g++ just a few seconds away.

Same error!! Except I'm slightly more au fait with GNU tools, so I was able to have a dig around the symboll table/disassembly and make sure that the function really was there.

Anyway, want to know the conclusion to this epic saga? Of course you do, you're on the edge of your seats, right??

I'll give you a clue. It starts with "mis" and ends with "spelling". Apparently I think "column" is spelled "columnn".

Would have been simple enough to figure out if I hadn't spelled it correctly in one source file, and incorrectly in both the header and the other file. Hence linker error rather than warm fuzzy compiler error.

Anyway, sorry for the rant. I just feel pretty thick!!! I do use version control even for casual stuff, so I could have rolled myself back out of this mess. But Git on Windows..... not so much fun.

Anyway, now I have totally selfishly whinged at you all, I'm off to the C forum to try to make up for it a bit. Hope everyone is well!

Really? I'd never have thought to try integrating git with Visual Studio.

I love it as a debugger - I really do, but I've never delved into the features. I'll have to give the git integration a shot.

I have the free version at home, but I did have the opporunity to use a shiny swanky version at an interview last year. They'd really seriously bought into it -- did all their testing and everything with it. Must make a nice change from the usual mishmash of tools, processes, approaches.

Thanks for that. I may give it a try. I got a bit freaked when I read that there's no interactive rebase support -- but I guess at home that doesn't matter. At work I need to seriously sanitise my commits before they see the light of.... server.

Haha - so very very true, Bonus semicolons, stupid operator precedence mistakes, "oops I forgot to include the file in the build" (spent quite some time chasing my taill around that one this morning.

I love it really -- feeling quite sunny about it all. Expect the MingW bit, that sent me a bit screwy.

Originally Posted by Rodaxoleaux

Your rant is basically programming in a nutshell. Spend days trying to find a "complex" error. The problem was you forgot a semicolon. gg no re.

Not a problem with programming per se, but a problem with all mainstream compilers.

Aye - I once spent a whole two days tracking down some memory corruption at work. Couldn't use valgrind either since it wasn't heap memory. Compared to that this was like fricking Mardi Gras. Not that I've ever been, but y'know, good.

I'd still take a nice pleasant line numbered compiler error though.

In fact, what I would like is to never make a mistake again

I went back to the same code again and this time managed to typo it into "coliuimn". What, is that half way between a colosseum (I did have to look that one up....) and a column.

So yeah, another linker error but no stress this time. Until VS decided to go haywire. It kept refusing to compile on the basis that one of its internal headers couldn't be written to. THen after a reboot, it came bak up in super paranoid mode -- wouldn't let me compile anything unless I got rid of all the scanf()s in my code. Managed to get that to stop (thanks, internet). Then it gave me a few pop up dialog lectures, some of which were slighty absurd. "Your array is too big!! It should not be more than 81 elements".

What an utterly random day. HAve a lovely evening everyone.

Given my work with the WinAPI, I expect a zero for success or a non-zero as a code if theres an error.

Edit: DaveSimmons covered it already. Never "expect" anything, refer to the documentation.