Date: Mon, 12 May 1997 09:12:06 -0400
From: mueller@inf.ethz.ch (Martin =?iso-8859-1?Q?M=FCller?= )
Subject: Re: linebreaks...

Line 1 of http://www.cis.tu-graz.ac.at/home/hollosi/sgf4_draft/sgf4.html says:
>SGF is a text-only format (not a binary format).

Everyone agrees that different end-of-line conventions are a headache.
However, they will not go away. I think SGF should remain a text format,
not go binary. Being able to edit game records is an advantage, so we
should keep this. All kinds of small transmission errors do happen, and
then editing helps.

This implies adapting to the local text conventions on each platform. There
are tools which can convert between those formats easily. E.g. on the Mac,
I use 'Drop Text' all the time.

I also think about non-roman character sets such as Japanese, which
probably have their own conventions for line breaks etc. Any experts
listening? This is probably a point that needs more discussion. The
standard is not too clear on what should happen if a text is written in a
script that the machine does not understand. Display random characters?
Nothing? A warning? What happens if such a file is edited, adding comments
in another script, and then written out again?

I think we can tolerate each of CR, LF and CRLF as end of line. A program
should preprocess and convert these combinations to the host's native
format when reading in the file. I believe this is only a small concession
to format purity which will turn out very practical.

        Martin