GNU Emacs: nXML mode and code pages

I have noticed some time ago, that nXML mode has problems with encoding names specified in the mandatory XML prolog, namely this one:

<?xml version = ‘1.0‘ encoding = ‘utf-8‘?>

And so, when one types windows-1251 (or opens XML file whose prolog is such), nXML puts this message in the status bar:

Unsupported encoding

Even worse, it refuses to save the XML, since it persists in its intention to save the file in the proper encoding, which is good. But this “unsupported encoding” thing is not quite true. It’s just that GNU Emacs does not “know” of such name in its codepage names list. You need to explicitly tell it what exactly do you mean by windows-1251 and everything is good. Put this in your “.emacs” file:

(codepage-setup 1251)
(coding-system-put ‘cp1251 ‘mime-charset ‘windows-1251)

So, what’s this? Well, GNU Emacs knows about “cpNNNN” code pages and uses them; note that it also needs to explicitly load them, too (not very smart, I think). Why it uses such charset names, I am not sure, since IANA charset registry says that windows-1251 is the only name of this encoding and no official aliases exist.

Of course, it is possible that GNU Emacs was the first one to handle multiple charsets than the IANA agency, and in absence of official and standard names, it used its own. I do not know.

These “125x” encodings are between 1250 and 1258. You may alias them all, if your underlying OS has support for them (and your fonts, too).