What tips do people have for a player to develop a sound opening repetoire?

I think it's very important to keep a record of all the games you have played, and organize them using an opening key, so you can see which lines are working for you, and where you need to find improvements.  You can try to do this by hand, or use repertoire software like Chessbase Light, ChessAssistant, Chess Position Trainer or SCID.

I bought lots of opening books took what I owned and discarded that which wasn't mine. 

The first thing you have to do when learning openings is to learn the principles (develop pieces, control the centre), which are much more important than concrete openings anyway

he is about 1600 rated.  I would assume he understands the basic principles of openings. 

To go into a little more detail.  I bought some repoirtoire books.  Used the lines I liked. The ones which I didn't like I used other opening books to fill in the holes. 

I started off by buying chess openings for white (and black).  I liked the hyper accellerated dragons and the nimzo indian but didn't like the bogo too much or the positions from the english that they gave.  So I filled in the gap by playing the kings english (learned it from mastering the chess openings)and benko/benoni's(abc's of the benko by martin) 

When working on a repatoire, I try to find openings with similar structures and themes.  It helps with understanding plans, opportunies, advantages and disadvantages after leaving the opening.  Examples would be Slav & Caro Kann, King Indian Def and Sicilian Dragon, French Winawer, Nimzo-Indian, Ruy Lopez Exchange.  Think about things like IQP positions, B+N vs 2 B with compromised P structure, etc.

Do not, I repeat DO NOT buy a bunch of books.  I did and now I have an extensive opening library of chess books of which I use about 3.  Get a primer book.  Mastering the Opening, by Byron Jacobs is pretty good as is World Champion Openings, By Eric Schiller, although I wouldn't reccomend many of his other books.

See what you like and try it out then if you really like it buy a book about the opening in question.

You can look at my blog page here where I talk about learning a new opening.  It's kind of long so I'll give you the abbreviated version. 

1. Choose something.

2. Use it in a bunch of 5 minute blitz games.

3. Analyze these games, really simple stuff.  See where you deviated from book how you came out of the opening etc...

4. Like it?

5. Understand it?

6. If yes then play some G15's with it.

7. Analyze G15's.

8. Still like it?

9. If yes, then buy a book on the opening.  If no, then find something else and start back at #1 until you do.

In chess every move has advantages and disadvantages.  Discovering what positions just naturally make sense to your mind is very important in selecting an opening repertoire.

All openings have characteristic pawn formations that result after just a few moves.  Learning what those characterisitic pawn formations are and becoming expert in handling those pawn formations (specifically much more expert than your opponent in handling those pawn formations) is key. 

A good book that covers in detail several characteristic pawn formations is "Pawn Power in Chess" by Hans Kmoch  

That's an interesting comment.  Are you saying that I would be able to input my own OTB games?  Can I easily backsolve each line in the archive of my games by result to see if with best play, white wins, black wins or it's a draw?  That would be an amazing feature if chess.com actually has it.

Incidentally, Chess Position Trainer and SCID are free.

 This.

I used to base my repertoire on finding out which moves were considered best. With horrible results. For instance, as black after 1.e4 c5 2.f4 I would play the Tal Gambit 2...d5 3.exd5 Nf6 (after all, that is why 2.f4 was replaced by 2.Nc3 by GPA afficionados, right? my memory ended there). Well, you get clobbered if you sac a pawn as black on move 3 against strong players, especially if you don't know what you're sacrificing it for. It is the strongest move, but in my games it scored 0%.

Similarly as white against the Benoni, where I "knew" that theoretically the f4/Bb5+ line is strongest (since, as many articles will tell you, that's the reason many people only play the Benoni after Nf3 instead Nc3 -- again, that was more or less all I knew). And I would suffer when theory stopped for me after move 9 or so, and I would try to prevent black's breaks on the queenside instead of furiously implementing my own e4-e5 plan (an IM was kibitzing the post mortem of a recent loss... I was explaining that in the Benoni I was always trying to prevent black's plans, always failed, and usually collapsed -- his dry answer "why do you play a f4 line then" -- uhm...).

So, yes, this is paramount. Play a line that you understand. What does that mean? Try to find lines that you would play yourself if you had no idea of the theory but 30 minutes of concentrated thinking time (unless there turns out to be a clear refutation afterwards; but just =+ for him is no problem).

I have a monster score (+13 =5 -3) against equal-rated opposition since september, by playing openings that I don't really know at all. The moves I figure out over the board work better for me since I at least know _why_ I play them (and I have better concentration this way). And then I'll look up the theory afterwards and decide if they're worth a repeat. And who cares if they're =+, I'm a 1900 player, not 2400...

I enjoy Tal's style of play, and certainly mean no disrespect.

1.d4.

1.d4 is probably better. You have the Queens gambit and the Trompowsky to go at.

King's Indian attack isn't exactly that good in my opinion. How can it be a forced win for white ? But that would be a good thing even more so to play it.

You think 1.g3 is the best move. I think 1.d4 is the best move. But hey you defeated me.

Familiarity with the middlegame positions and pawn structures that tend to result from a particular opening is paramount. This only happens when you play the same opening again and again.

Going over your games, it's true that you will soon discover what works for you and what doesn't. Against 1. e4, I always chose Alekhine (1...Nf6) or Caro-Kann (1...c6). Only after going over my games did I find that my winning percenttage with the Caro-Kann was much higher!

Okay.

If you want a SOUND repertoire, then you might consider the Colle-Zukertort on White, the QGD on Black plus any number of safe e-pawn Black defenses.

I wrote a book on the Zukertort [www.zukertort.com] and Matthew Sadler has written a good book for understanding the QGD.

This is similar to the point I was trying to make, although it is important to realize that it only takes one move to bust a line.  I scored a high percentage of wins using the Caro-Kann as well, but it was only after I backsolved my games by result that I realized that there were a whole bunch of holes in my repertoire that were leading to losses with best play.  At the moment, my Caro-Kann is in the repair shop, but I'm hoping to bring it back out, if I can find improvements in the lines where I'm losing as black.  In any case, analyzing your own games, and in particular backsolving to see what happens when both sides play the best moves are two of the best ways of developing a sound opening repertoire.

 How many games did you do that with? Isn't it likely that in many lines you'll end up with 1 game quickly, and then the result of that depends on the entire rest of the game? That one of you made a middlegame or endgame mistake says nothing about the opening.

That's in general a problem with backsolving just games and their results.