[cvsnt] Re: Problem importing files including 0x1A character

[Tony Hoyle]

Gill Ernst wrote:
> Originally it was from the brief editor, but later one we use
> CodeWrite. CodeWrite opens all files in binary mode and appends
> for sure all new text after the EOF. It seems all other tools are
> also working with binary mode, so no one was detecting this problem
> until now.

Sounds like a bug in CodeWrite.

The ^Z convention goes all the way back to CP/M...

Some history..

http://blogs.msdn.com/oldnewthing/archive/2004/03/16/90448.aspx

> But why the hell is this working when I add or modify
> such a file and later on commit it to the CVSNT server.
> It looks like you are using binary mode in this case.

The RCS files are accessed as binary, as they're always Unix standard, 
but the client always accesses text files as text (so that the line 
endings work, amongst other things).

> 
>>The alternative if you really must keep the ^Z is treat them as binary 
>>files.
> 
> 
> No, it must be a text file.
> 
You have a choice - remove the ^Z to turn them into valid text files, or 
add as binary... no third option, unfortunately.

Tony