CR/LF hell

From: kevin_vanhorn_at_[hidden]
Date: 2001-03-14 11:00:33

I'm running into problems with the boost libraries because the sources
are all in the absurd MSDOS text format (two characters for
end-of-line instead of one).  This means that every time the code has
a line of the form

  #define some_macro(some_args) ... \

I run into trouble, because the "\" is not followed by a newline
character -- it's followed by a CR, and so the compiler has a fit.

I tried re-extracting the files using unzip -a to get proper
conversion of text files, but all of the files are marked as BINARY in
the archives, even when they are clearly text.

So there are several problems here:

1. The text files should either be marked as text and not binary, or
there should be a UNIX-friendly package I can download (e.g., gzipped
tarball.)

2. Even if the text files in the zip archive were properly marked,
many (most?) people aren't going to know that they need to use the -a
flag to properly extract the text files.  As a result, they're going
to get compiler errors, and if they haven't run into this problem
before (luckily, I had) they are going to be quite baffled -- for
example, emacs doesn't show the extra CR character, so the file looks
just fine, and it's easy to overlook the little "(DOS)" indicator in
the bottom-left corner of the window.