Re: [PATCH v2 0/4] UTF8 BOM follow-up

[Junio C Hamano <gitster@pobox.com>]

Karsten Blees <karsten.blees@gmail.com> writes:

> Wouldn't it be better to just strip the BOM on commit, e.g. via a
> clean filter or pre-commit hook (as suggested in [1])?

The users can do whatever they want and if they think having a BOM
in these files is a bad idea, I'd encourage them to use whatever
means to ensure that.  The code and history hygiene is a good thing.

But you should realize that $HOME/.gitconfig, $GIT_DIR/info/exclude,
$GIT_DIR/config, etc.  are not even committed files in the first
place.  These are not even defined to be "UTF-8 only" by us.  Their
contents is entirely up to the end users.

Here with these changes, we are only being nice to the users by
stripping a well-known two-byte sequence that is known to be left
commonly by some tools users would use. In a sense, this is the same
degree of niceness that we strip the CR at the end of the line
before LF.  Just like you _could_ have said these files must be
encoded in UTF-8 and must not have BOM at the beginning, we _could_
have defined that these files must be recorded with LF end-of-line.
But obviously we don't, as there is no need to make lives of end
users unnecessarily more complex, and it is easy to help users use
both LF and CRLF with simply stripping on our reader's side.  We do
this BOM stripping for the same reason to make it easier for users.