Arrays are [x,y] or [y,x] coordinates? C

[pieguy]

My initial thinking was the conventional x,y coordinates, but I'm starting to think its y,x. Just need conformation.

The first pic is y,x and the second is x,y.
http://ubuntuforums.org/attachment.php?attachmentid=157918&stc=1&d=1274563992
http://ubuntuforums.org/attachment.php?attachmentid=157921&stc=1&d=1274564368

[Bachstelze]

It's whatever you want it to be. In memory, there are no rows or columns.

[StephenF]

For dumping to a console the y,x scheme holds true. There is no correspondence with any of the quadrants of the Cartesian plane. In GUI toolkits the top left coordinate of a drawable surface is also 0,0.

In general it is quantity y of x of type. The spacial mapping is coincidental to the left to right, top to bottom writing system so others may think differently.

[MadCow108]

C uses row major storage for its arrays, which means [y,x]

if you need column major you can always linearize your data yourself:
array[col + NCOLS*row] // row major equal to array[NROWS][NCOLS]
array[row + NROWS*col] // column major (fortrans default)

[Queue29]

It's whatever you want it to be. In memory, there are no rows or columns.
Not true not true not true.


C uses row major storage for its arrays, which means [y,x]

if you need column major you can always linearize your data yourself:
array[col + NCOLS*row] // row major equal to array[NROWS][NCOLS]
array[row + NROWS*col] // column major (fortrans default)



And this is sometimes a lot more important sometimes than you might think. In anything that is to be high-performance, it is necessary to take into consideration the way data from RAM is loaded into the cache, which means explicitly organizing your data row major or column major, depending on how the language you are working with works behind the scenes. To the best of my knowledge, gcc cannot optimize things like this, however the intel compiler can.


Here is a 'fun' walk-through a former professor of mine made which demonstrates just how important organizing data and operations can be, by implementing a matrix-matrix multiplication function in C.

http://z.cs.utexas.edu/wiki/LA.wiki/OptimizingGemm

[wmcbrine]

C doesn't actually have multi-dimensional arrays as a language construct. If you have something like "int foo[10][10]", that's an array of arrays of integers.

[croto]

[QUOTE=Queue29;9344481]Not true not true not true.

I agree that they can be whatever you want them to be, and I also agree with the performance considerations. For instance: a[.][.]...[x] is next in memory to a[.][.]...[x+1] so it is faster to traverse the matrix last index first, first index last. However, if you want to call the last index row or column is up to you.