Largest Single Code Module

I work in the videogames industry, and I once came across a single code module in a user interface system, which, at ship date, was over 100,000 lines long. (And it was only a sub-sub-system.)

Now, I know there has got to be larger ones out there. I'm in a SizeIsEverything? mindset right now. I want to know the absolute BIGGEST.

Thanks.

If it was a video game, was that maybe 100K "lines" of assembly? How many lines of CeePlusPlus or JavaLanguage would that be?

In my cobol days the main account update program was about 8" thick on 132 col music rule. It didn't really use subroutines, mostly goto's, and of course one global variable space. so can it be counted as one Code Module? I don't know how many pages per in you get with those big old folders.

This won't win, but when I was in college, I wrote an assembler in Pascal with a procedure so large it made the compiler run out of memory. One big select-case statement with all possible op-codes and most of the processing for each.

Did I refactor it? No. I just made the system give the compiler more memory. ;->

Were you using a Mac?

I had a similar experience with my compiler class in college. We were supposed to write it on the school's Vax in C. I had Turbo C++ which had far superior editting and debugging capabilities, so I first wrote it on my PC. At the time, Turbo C++ didn't export make files and I didn't want to learn how to use them, so I structured the source as all include files.

When it came time compile and run the code on the Vax, the compiler reported an obscure error message about a stack overflow error. It seems that one of the default optimizations had a bug in it which was triggered when the source was too large. Turning off the optimization did the trick, but debugging the problem was a pain!

-- MarkAddleman

I used to work for Univac, and I heard of a 65K-line COBOL program which, when converting from IBM, exceeded our compiler's line count maximum.

Although not huge, I also worked on an 12K-line AS/400 RPG program. To debug it, we used to spread the listing all the way down the hall, so we could highlight all occurrences of certain variables. One of the programmers got so confused he hard-coded a GOTO out of a subroutine back into the mainline. We fired him. For being confused, or for coding instead of highlighting? It must have been the second, since it was operative and represented being the long pole in the tent.

I know of a 4GL that did not support separate compilation (no subroutines). A 10K-line program got so deep into the indentation that we had to reset the indentation, and add a comment above the code that said ("this code is really indented to 132 columns").

Pierre Cloutier

We fired the programmer because he got lazy and hard-coded the GOTO as an expedient hack because he couldn't be bothered to figure out how to do it right. Plus he obvioulsy had no idea what he was doing.