On Sun, 27 Nov 2011, nierveze wrote: I used to give that and the inverse as early exercises in file I/O in my [beginning] Assembly class, and as simple examples in my Data Structures and Algorithms class for unions, shifts, ANDs, file I/O Some students had difficulty grasping the concept of one input, two outputs (and vice versa, where I showed it right before merging files) Another difficulty that many students always had was processing data, such as sorting or searching, WITHOUT reading it all into memory. "Bigger than 64K? ADD RAM!; Bigger than 1M? ADD RAM!", so I made them write a sort program with NO file size limits, and one whose only size limits were the file size limits of the OS (MS-DOS FAT12 permitted files from -2147483648 to 2147483647 bytes), which was even harder. Amongst the obvious possibilities for splitting the file are: read byte write byte to file1 read byte write byte to file2 read word write byte to file1 write byte to file2 WATCH OUT FOR: some compilers want to [by default] fopen() files in "TEXT" mode, and stop when they hit char26, or other none end-of file "EOF" characters. Most such compilers have switches to over-ride without necessarily changing from fopen() to open() Surely you can reduce it to much less than 12 lines in "high level" languages! -- Grumpy Ol' Fred cisin at xenosoft.com

On Nov 28, 2011, at 2:37 PM, Chuck Guzis wrote: In Python, you could do the following interactively (it reads the whole thing into memory first, but any modern machine shouldn't have trouble with that, and it's really short): ---- in_data = open("infile.bin", "rb").read() open("even.bin", "wb").write(in_data[0::2]) open("odd.bin", "wb").write(in_data[1::2]) ---- If you wanted to be more memory efficient, you could do it the old-fashioned way, though Python actually gets in your way about it a little bit: ---- infile = open("infile.bin", "rb") outfiles = [open(x, "wb") for x in ["even.bin", "odd.bin"]] while infile.peek(): for f in outfiles: f.write(infile.read(1)) ---- That also has the benefit of fairly easily allowing arbitrary interleaving (if you had 4 ROMS for a 32-bit system, say). There's probably a nicer way of doing it than using file.peek(), but since there's no feof() equivalent for the file object, it'll do. - Dave

In article <4ED3939B.20362.F4C5DD at cclist.sydex.com>, "Chuck Guzis" <cclist at sydex.com> writes: TECO ftw -- "The Direct3D Graphics Pipeline" -- DirectX 9 version available for download <http://legalizeadulthood.wordpress.com/the-direct3d-graphics-pipeline/> Legalize Adulthood! <http://legalizeadulthood.wordpress.com>

On Mon, 28 Nov 2011 05:47:05 -0500 Sean Conner <spc at conman.org> wrote: Yes, and does cruel things like for () { write( fd, buf, 1); } OK, you may do things like this on DOS, but on anything else system call overhead for every single byte will bite you. Portability is allways a trae off. On Unix mmap(2) is the most efficient way of IO. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}