Date: Mon, 12 May 1997 00:20:13 +0200 (MET DST)
From: Arno Hollosi <hollosi@sbox.tu-graz.ac.at>
Subject: Re: linebreaks...


>  I read in the description of the diffrent propertyvalues that in 'Text'
> type linebreaks should be LF only. Does this apply to the whole SGF file? I

Yes.

> think that we should not specify this. I think the programs should be able
> to handle LF or LFCR or CR. That way users can still read the SGF file in

The spec states:
> Attention: a single linebreak is represented differently on different 
> systems, e.g. 'LFCR' for DOS, 'LF' on Unix. But when writing a SGF file, 
> a linebreak has to be written as a single 'LF' (ASCII code: 10 
> (C-Syntax: '\n')) in all cases. 

conclusion:
- applications must be able to deal with different linebreaks
- when writing use a single char (LF)

> their favorite text editor. One more thing. In the 'Text' type text it says

SGF is NOT intended to be human readable!

No fiddling with text editors should be necessary - if you do use a text 
editor then use one able to understand single LF linebreaks.

> that LF equals \n in C. That is NOT true. On UNIX systems yes. But on

Ok. So what?
I think the spec is clear at that point.

Note: the spec doesn't say use >>fputc('n', file)<< it says:
> written as a single 'LF' (ASCII code: 10 (C-Syntax: '\n'))
LF is '\n' in C syntax. (char c = '\n')

> MS-DOS/MS-Windows \n equals LFCR. So you can not realy use \n in cases like

Wrong - this transformation is only done when you write text files.
Solution: write binary files.

SGFC reads/writes binary files (I suggest you all use binary instead of 
text files - think of other charsets - or multibyte characters)


/Arno