[quote="SANEAlex"]You should only get worried when its longer than the novel[/quote]
I had to check: how do you judge, though? 
Code lines are often just empty, or '}', so possibly lines are not a fair comparison at all. Writing a line of code can take days of research or not... well I guess the same is true of writing a line in a book so fair enough on that one And what defined a "word" in code? I've no clue, but I suspect it's "anything that's not whitespace, surrounded by whitespace" - so the bracket is a "word". Still, that's arguably true, for programmning.
As well as clientside code, I also have serverside stuff (not including the 4,000 lines of "shared libraries" I use, since I originally wrote them for other projects).
8,302 lines, 33,382, words, 285,640 characters clientside
3,950 lines, 16,526 words, 146,184 characters serverside
12,252 lines, 49,908 words, 431,824 characters total.
A pre-pub .doc of the book is reported by OpenOffice as having:
9,434 lines, 110,012 words, 605,745 characters.
So, both projects are roughly comparable, but with the lower word and character count, I can't convincingly argue that the code is fersure the longer of the two!
... yet! 
I would go on number of words as i think of pTerry as one of the great wordsmiths so:-
A pre-pub .doc of Thud book 110,012 words
clientside and serverside Thud game 49,908 words
so only 60104 words to go i think if you properly comment the code and bolt on all the extras all us players clamour for (thank you btw for what we have already) then you might just make it
but if the code is better optermised i for one wont complain after all pTerry has to list all his preveous works as well and that is long list now. 
SA
Rats. Looks like it's counting only alphanumeric strings as words, too. So, functions, variables, classes and constants.
So, assuming I keep to about the same level (I was surprised by how similar the proportions were between php and java: I guess my coding style is fairly constant), at four words per line, I've another 15klocs before I'm level. Fair enough, plenty of time before I need to worry
I might well get there by the time Thud reaches the dream that I'd like to see.
Particularly if I include regex matching for userdefined swearword filters. Which would be nice, but might be hard to duplicate: other than richtext, transparent components, I want the protocol NOT to require things that are very hard to duplicate in any language, and a standardised regex engine is one of those things.