[not strictly linux] why is file 2 bytes not 1 byte?

geekpie · 07-26-2007, 01:11 PM

If I open a new file in vi, type 1 character (no new-line), and save, the filesize is 2 bytes. Why isn't it 1 byte?

Is it even possible to create a file with size 1 byte?

Brydon · 07-26-2007, 01:45 PM

End of file character which is hidden, control-z if I remember it correctly.

Or it's the way the file is stored in the filesystem, more likely the answer. Let's find out what everyone else thinks.

Nylex · 07-26-2007, 01:50 PM

FWIW, I can create a file that's 1 byte in size using vi (Elvis).

pixellany · 07-26-2007, 01:55 PM

My editor (nano) also throws in a "newline" (0x0a) for free.

If you really, really want a 1-byte file, you can delete the newline char. thusly:
cat filename|tr -d '\n' > newfilename

I could have sworn there was also a way to do this with sed, but I can't find it.

To see what actually gets into a file:
cat filename| hexdump -C|more

Brydon · 07-26-2007, 02:15 PM

That's it a 'new-line' character not 'end of file'...

colucix · 07-26-2007, 02:16 PM

Another way to create a 1-byte file could be

Code:

echo -n A > onebytefile

A quick way to verify (dump) the content of a file is

Code:

od -c onebytefile
0000000   A
0000001

For a 2-byte file, you will see

Code:

0000000   A  \n
0000002

which reveals the unwanted presence of the newline character. Nothing hides from the good ol' od command!

ilikejam · 07-26-2007, 02:27 PM

Geekpie - What do you get from 'echo $LANG'? Are you using UTF16?

Edit: Never mind. No difference.

geekpie · 07-26-2007, 02:42 PM

yes od and hexdump reveal what is going on.

I hadn't realised you get a \n at the end of the final line of your file in vi, but I suppose it makes sense.