Re: Newsgroups files

Hi Nigel,

Yes, we've sort of had this discussion before about encoding.  This one
is more about the inconsistency of the labeling of the groups.

The inconsistency of the labeling can come from several causes.  One of them is the encoding, depending on how the news server interprets the control articles (if it follows the declared encoding or, in the lack of declared encoding in the control message, which local encoding it will use).
Another cause can be that the news server no longer process control articles (change of PGP key not updated, problem in GnuPG or like) so it will not see possible description changes in checkgroups.  It probably explains some variants you saw on fr.bienvenue.
Or the news server does not have the feature to automatically update the descriptions, or it has been disabled by the newsmaster.  Not all news servers do that; as far as INN is concerned, I added that feature in its 2.4.6 version in 2009; I reworked the docheckgroups program at that time, with a new -u flag that does the right magic of updating the descriptions in the newsgroups file with a proper number of tabulations and an alphabetical sort, removing obsolete descriptions and adding new ones.  Descriptions from newgroup and checkgroups control articles have been properly reflected since then.  Nonetheless, not all news software does that.
It would be interesting to know whether fr.bienvenue is still declared moderated in the active file of the news server which have "(Moderated)" at the end of its description.  It may just happen that they processed the newgroup control article once sent to unmoderate it, but dit not update the description.

In the newsgroups list above, pretty much every group that contains
non-standard A-Z letters is garbled.
 Probably because it's ISO-8859 when I'm using UTF-8. The cn.* groups
are definitely garbled.
 I'll just do my best to make a valid UTF-8 file for my server.

In fact, the newsgroup list from GitHub was properly encoded in UTF-8 but your navigator did not use UTF-8 to render it for a reason I do not know.  Might you have to force the charset in your navigator?
The HTTP headers correctly have:
   Content-Type: text/plain; charset=utf-8
Does it appear better with this version?
   http://usenet.trigofacile.com/hierarchies/data/newsgroups.utf8
Or maybe you donwloaded the file and then opened it with an editor in another charset?
% file newsgroups.utf8
newsgroups.utf8: UTF-8 Unicode text
--
Julien ÉLIE
« Love is blind but marriage is an eye-opener. »