I just installed Arch Linux on a Raspberry Pi and wrote a little perl script. But whenever I try to run the script, I get the following error:

Yes, I know that usually happens when you transfer a file from a Windows system to a Linux system (since Windows uses CRLF pairs, and Linux only LF as line endings). But the problem here is, that there is no CR at the end of the lines.

I checked the file (on Linux) in a Hex Editor and it confirmed there are only LFs (0x0A).

So what/who is adding the CRs? Is it bash? How do I get rid of them? I know about dos2unix, but that doesn't help since there aren't any CRs in the file in the first place...

To answer my own question: it was caused by a number of reasons:

First of all, I use vsftpd to transfer the file from a Windows machine to the Arch Linux machine. I specifically used ASCII mode for the transfer. vsftp by default ignores ASCII file transfer requests and does a binary transfer instead. I did not know that.

The reason I didn't see those CRs was because I use JOE as my editor (it has a HEX edit mode). And JOE by default automatically translates CRLF to LF when reading, and translates it back to CRLF when writing. I did not know that either (and is new for me since I've installed JOE on many, many machines before).

So in short, I had to tell vsftpd to accept ASCII transfers as well, and tell JOE not to automatically translate CRLF to LF and vice versa.

off: here is an adcr/rmcr program I wrote some years ago

thanks for the update. but it doesnt explain why dos2unix didnt work ?

I always simply do removes that silly newline thing as well... and by preference I never use windows, which helps reduce the amount of times I need this :-)

pfe32 -a free single executable for Windows does Linux Files very well.

Use "file" to check if it's a problem, and "dos2unix" to fix it. Easy peasy.