eprints_tech messages

From: Tim Brody <tdb01r AT ecs.soton.ac.uk>
Date: Wed, 08 Oct 2008 15:32:05 +0100

*** ↵
http://www.eprints.org/tech.php/id/%3C1223476325.6300.54.camel%40dell-desktop.example.com%3E
*** EPrints community wiki - http://wiki.eprints.org/

On Wed, 2008-10-08 at 13:43 +0100, Ben Wheeler wrote:
> *** ↵
http://www.eprints.org/tech.php/id/%3CEMEW-k97DhS538366f67caa57ec8564b8af51d8d504-20081008124315.GA27188%40brucia.ulcc.ac.uk%3E
> *** EPrints community wiki - http://wiki.eprints.org/
> 
> On Tue, Oct 07, 2008 at 04:08:33PM +0200, Janusz S. Bień wrote:
> > *** ↵
http://www.eprints.org/tech.php/id/%3CEMEW-k96F8c8abc3c25ec02083a97b3e616482abd3a-87iqs41ori.fsf%40mimuw.edu.pl%3E
> > *** EPrints community wiki - http://wiki.eprints.org/
> > 
> > 
> > First a terminological issue: ASCII is not Unicode and vice versa :-)
> > 
> > "ASCII Citation" should be called "Plain text ↵
Citation" or "Unicode
> > Citation" or "UTF-8 Plain text Citation" or something ↵
like that.

Good point.

> > Secondly, there is definitely a bug in some part of the system, as my
> > name is distored in BibTeX, Refer and Record Manager exports. It is
> > rendered correctly in several other formats (I haven't tested all of ↵
them)
> 
> Ok, it seems that the "ASCII Citation" (misnomer as you say) ↵
export 
> has this header:
>   Content-Type: text/plain; charset=utf-8
> while BibTex has this:
>   Content-Type: text/plain; charset=iso-8859-1
> 
> Looking in the plugins, Text.pm does this:
>         $self->{mimetype} = "text/plain; charset=utf-8";
> while BibTex.pm does this:
>         $self->{mimetype} = "text/plain";
> 
> So, this might be fixable by adding the charset to those that use
> text/plain and don't currently specify it. Or making utf-8 the 
> default charset in the Apache configuration might work. In fact
> perhaps if the standard EPrints VHost configuration included
> AddDefaultCharset utf-8
> many of these problems might disappear immediately? I must stress 
> that I haven't tested this...

I think BibTeX output is intended to be ASCII as it has escapes for utf8
to tex, and substitutes '?' for anything it doesn't understand (hence
the output at:
http://demoprints.eprints.org/cgi/export/779/BibTeX/demo-eprint-779.bib

I'm aware of previous discussions here about BibTeX/UTF-8 on import.

Should we output BibTeX in utf-8? Should we include a byte-order-mark?

All the best,
Tim.