zero or one based collection?

re: zero or one based collection?

Curt Hagenlocher — Tue, 16 Mar 2004 22:17:00 GMT

How many of you - and be honest here - have ever heard of machines where the word size wasn't a multiple of 8?

Many of the machines in DECs PDP family had 36 bit words. Is there a prize for this? ;)

re: zero or one based collection?

Mike Kozlowski — Tue, 16 Mar 2004 22:19:00 GMT

It should come as no surprise to anyone that Perl lets you define your own array base on a dynamic basis by setting the special $[ variable. It's, um, not recommended, though.

RE: zero or one based collection?

nospamplease75@yahoo.com (Haacked) — Tue, 16 Mar 2004 22:20:00 GMT

An obvious reason to keep things as they are is when dealing with coordinate plains. If you use a 2-D array to plot points for example, it would suck if you used a 1 based array. The distance to the point 1,1 would be counterintuitively 0, not sqrt(2).

I think it helps to think of arrays as a distance. So instead of thinking about the 1st element of an array, think of an element at distance 0.

re: zero or one based collection?

Larry Osterman — Tue, 16 Mar 2004 22:31:00 GMT

The first machine I ever did significant programming on was a Decsystem-20 - 36 bit words, segmented into 2 18 bit halfwords.

Stop by my office sometime and I'll show you the reference manual :)

re: zero or one based collection?

pb — Tue, 16 Mar 2004 23:11:00 GMT

The processor used in the Chromatic Mpact of a few years ago used 9-bit bytes. It was great for the days when others had 16-bit color (565) and we had 18 bit color (666). Writing to it from the PC host was painful but hey...

re: zero or one based collection?

Ferris Beuller — Tue, 16 Mar 2004 23:16:00 GMT

Yes the typical argument of zero based arrays is the underlaying processor, well hey this isnt assembly, get over it, how you implement this in the runtime or compiler I couldnt give a rats butt about, I say, walk over and get me the first item from that collection on the bench, you dont pick the zeroth item, you pick the first. Then you go on to mumble about legacy minesets, well hey, time for a change. I guess people are afraid of change, well I for one arn't and welcome such changes. Why bother creating a new language if you are still hugging your legacy pillow at night. Waste of time.

re: zero or one based collection?

Ferris Beuller — Tue, 16 Mar 2004 23:16:00 GMT

Maybe we need a new designer on the C# team with NEW ideas and concepts and one that isnt afraid of change.

re: zero or one based collection?

Sean Chase MCSD.NET — Tue, 16 Mar 2004 23:49:00 GMT

travesty is a good word Eric. I always hated that about VBA. Some collections were base 1 others zero. It was nuts. Forcing to base zero was a great move.

re: zero or one based collection?

Ferris Beuller — Wed, 17 Mar 2004 00:45:00 GMT

Forcing base 1 is a good move.

re: zero or one based collection?

Ferris Beuller — Wed, 17 Mar 2004 00:46:00 GMT

*takes notes* "Steve Chase is an arse licker"

re: zero or one based collection?

Dan Crevier — Wed, 17 Mar 2004 01:41:00 GMT

I have a feeling lots of code is going to be ported from C/C++ to C#. If arrays were 1 based in C#, it would lead to tons of errors.

Also, I think the 0 based array thing made a lot of sense with the tight relationship between pointers and arrays in C. It would be really weird if p[n] = *(p+n-1).

re: zero or one based collection?

Beam me up! — Wed, 17 Mar 2004 01:57:00 GMT

That should read:
"Damn it Jim I'm a programmer, not a compiler."

re: zero or one based collection?

Anand — Wed, 17 Mar 2004 03:42:00 GMT

VB used to support the dynamic lower bounds. But they removed it in VB.NET. There was also support for a 0 based or 1 based array using the Option Base statement. That did give you a lot of flexibility...

re: zero or one based collection?

Jens Samson — Wed, 17 Mar 2004 06:10:00 GMT

'Why cant we either have 1 based or user definable array bounds?

The CLR does support this kind of construct (I had a hard time not using the term “travesty“ here...), but there aren't, to my knowledge, any languages that have built-in syntax to do that.'

Even Pascal lets you determine your own Array bounds. It goes even further, you could construct a set [Monday, Tuesday, ..., Friday] (Have forgotten exact syntax) and use that as Array indexes. Some great functionality that other languages sadly never bothered to implement.

Eric Gunnerson just keep pumping out the cool posts...

Code/Tea/Etc... — Wed, 17 Mar 2004 09:53:00 GMT

Eric Gunnerson just keep pumping out the cool posts...

Code/Tea/Etc... — Wed, 17 Mar 2004 09:53:00 GMT

re: zero or one based collection?

Stuart Dootson — Wed, 17 Mar 2004 07:50:00 GMT

Eric: The CLR does support this kind of construct (I had a hard time not using the term “travesty“ here...), but there aren't, to my knowledge, any languages that have built-in syntax to do that.

Ada has that built-in..and more...

type SomeTypeName is array(4..10) of Integer;

or, for truth tables

type TruthTable is array(Boolean, Boolean, Boolean) of SomeResultType;

(I have a love-hate relationship with Ada - it does have some nice features, like named associations, but I'd rather use C++ :-)

Also on starting numbering at zero - Dijkstra's argumnet for numbering from zero (see http://www.cs.utexas.edu/users/EWD/ewd08xx/EWD831.PDF) is of some interest.

re: zero or one based collection?

Shane King — Wed, 17 Mar 2004 08:50:00 GMT

I'm not sure the argument is necessarily efficiency vs logic.

Mathematically, it does make good sense to start at 0. Like it was mentioned, when you're representing geometry, it's good to have a 0 co-ordinate.

Additionally, I think 0 addressing means that arrays and lists have more in common conceptually. It makes sense that 0 = head, and length = end.

Finally, it makes no sense to change. I've never met anyone who's got any natural aptitude at programming what so ever, and also can't grasp 0 based collections. You say it's a secret handshake, I say it's just the way life is. If you can't grasp it, chances are you lack the aptitude at programming anyway, so why make programs less efficient for the sake of people who aren't going to be programmers?

re: zero or one based collection?

JasonM — Wed, 17 Mar 2004 08:57:00 GMT

I've never used them, but wouldn't 1 based arrays screw up "looping around" arrays using modulo arithmetic?

Maybe it's just me, but I do that _all_ the time.

re: zero or one based collection?

Ferris Beuller — Wed, 17 Mar 2004 09:57:00 GMT

re: zero or one based collection?

Philip Panyukov — Wed, 17 Mar 2004 10:04:00 GMT

There already was a heated discussion on the subj here: http://discuss.fogcreek.com/newyork/default.asp?cmd=show&ixPost=1822&ixReplies=8

I personally find it extremely painful to use APIs that mix 1 and 0-based collections/arrays in VB.

re: zero or one based collection?

Nicholas Paldino [.NET/C# MVP] — Wed, 17 Mar 2004 12:47:00 GMT

That code for looping through an array looks like all of the VB code I use for arrays that I don't have control over. I never know what the bounds are, and have to code around it.

Travesty is right.

re: zero or one based collection?

Dave Mackersie — Wed, 17 Mar 2004 13:10:00 GMT

Your argument doesn't hold water !

The expression:
address = base_address + (x - 1) * sizeof(x)

can be re-written as:
address = (base_address - sizeof(x)) + x * sizeof(x)

The expression (base_address - sizeof(x)) is a constant that can be determined at compile time, so there is no need for a run time decrement !

The real reason for zero based collections goes back to the original design of the "C" programming language, in which the designers wanted array operations and pointer arithmetic to be equivalent, so that a[x] would be the same as writing *(a + x). Since then, all languages that followed in the tradition of "C" have used zero based indicies: C++, Java, and now C#.

Non-zero lower bounded arrays (the other side of the coin)

Panopticon Central — Wed, 17 Mar 2004 22:00:00 GMT

One we lost.

re: zero or one based collection?

Don Smolen — Wed, 17 Mar 2004 23:37:00 GMT

I was a systems programmer on the Control Data 6000 series. 60 bit words (12 bits in Peripheral Processors)

re: zero or one based collection?

Doug — Thu, 18 Mar 2004 00:45:00 GMT

<RE>Your argument doesn't hold water !

The expression:
address = base_address + (x - 1) * sizeof(x)

can be re-written as:
address = (base_address - sizeof(x)) + x * sizeof(x)

The expression (base_address - sizeof(x)) is a constant that can be determined at compile time, so there is no need for a run time decrement !</RE>

Eric was talking about a time before C, a time before optimizing compilers. If you re-read what he wrote, he was actually talking about a time before COMPILERS! Gee I feel old!

Sorry. Zero rules.

System.Environment.NewLine, etc...

TraviBlog — Thu, 18 Mar 2004 22:14:00 GMT

re: zero or one based collection?

Alfred Thompson — Thu, 18 Mar 2004 22:46:00 GMT

No one else is claiming programming on the 12 bit word PDP-8. Now there was a system for real progammers. :-) None of those frills like multiply, divide and subtract.
I have to say though that I miss the ability to set different lower bounds for arrays in VB. It let you write things that were intuitive like having arrays that went from 1900 to 2000 for a set of years. Sure you can get around it but it adds complexity that many would just as soon let the compiler worry about.

Paul Vick, Language Designer on the VB team, discusses arrays with non-zero lower bounds...

Code/Tea/Etc... — Fri, 19 Mar 2004 10:04:00 GMT

re: zero or one based collection?

HateToAdmitIt — Fri, 19 Mar 2004 16:17:00 GMT

Fortran.NET has built in non-zero based arrays...

re: zero or one based collection?

Luke Stevens — Sat, 20 Mar 2004 01:08:00 GMT

Back in the old old days, everything was 1-based, because zero was not yet considered a number. So you had AD 1 immediately following 1 BC, and most years in the second century began with a 1 rather than a 2. To the Romans, December was the third month after October, and a newborn was in his first year; so we can consider ourselves more sophisticated now that we say, more sensibly, that December is two months later than October and a newborn is of age zero. Unlike centuries, we now start decades (like "the 60's") on years that end with 0 instead of 1, though we still have no good name for our current decade. The day starts not at 1 AM but at 12! It even sounds downright strange to say that we can't vote until our nineteenth year. So now that mankind has finally gotten used to the idea that zero is a number in its own right, numbering things from zero is seeming ever more sensible.

re: zero or one based collection?

Luke Stevens — Sat, 20 Mar 2004 01:10:00 GMT

Oops, no HTML in comments, huh?

Why doesn't VB support non-zero lower bounds for arrays?

Panopticon Central — Fri, 16 Apr 2004 20:50:00 GMT

Why doesn't VB support non-zero lower bounds for arrays?

Panopticon Central — Sat, 17 Apr 2004 01:45:00 GMT

Why can't you say

Why doesn't VB support non-zero lower bounds for arrays?

Visual Basic Frequently Asked Questions — Tue, 20 Apr 2004 13:01:00 GMT

re: zero or one based collection?

Ben Hutchings — Tue, 20 Apr 2004 18:17:00 GMT

Eric, you don't know your history. FORTRAN, which is one of the oldest high-level languages, uses 1-based arrays.

Eric Gunnerson just keep pumping out the cool posts...

Code/Tea/Etc. — Wed, 25 Aug 2004 13:17:00 GMT

Paul Vick, Language Designer on the VB team, discusses arrays with non-zero lower bounds...

Code/Tea/Etc. — Wed, 25 Aug 2004 13:18:00 GMT

Date

Hackward and Foreword — Sun, 03 Oct 2004 21:18:00 GMT

Date

Hackward and Foreword — Sun, 31 Oct 2004 22:29:00 GMT

MBA

MBA — Sat, 18 Dec 2004 18:55:00 GMT

Helpful For MBA Fans.

Paul Vick, Language Designer on the VB team, discusses arrays with non-zero lower bounds...

Duncan Mackenzie .Net — Thu, 05 Oct 2006 19:50:09 GMT

One of the changes from VB6 to VB.NET was the removal of non-zero lower bounded arrays... a concept discussed by Eric Gunnerson