[cvsnt] Jar files becoming corrupt

[Gerhard Fiedler]

Bo Berglund wrote:

> Each file is added to the repository with a marker that tells CVS if
> it will be treated as a text file, a binary file or a unicode file.

This "Unicode" is written in the docs, and is being repeated in
discussions. AFAIK, it is not completely correct, and causes
misunderstandings. What cvsnt means with "Unicode" seems to be UTF-16. At
least this is how files get checked out that got added as "Unicode" files
-- independently of the encoding they were in when they got added.
Encodings different from UTF-16 are not preserved by this setting.

Gerhard