Re: [PATCH] git-svn: Add a svn-remote.<name>.pushurl config key

[Jeff King <peff@peff.net>]

On Fri, Apr 08, 2011 at 01:13:47PM -0700, Junio C Hamano wrote:

> A Tangent.
> 
> >> X-Mailer: git-send-email 1.7.4.2.1.gd6f1f
> >> In-Reply-To: <1302102336-8800-1-git-send-email-asedeno@mit.edu>
> 
> This is not about this particular patch, but the From: address
> git-send-email generates for you does not seem to quote the human readable
> part, even though the name has a "." in it.
> 
> Your mails seem to reach the recipients fine, but I saw my reply to you
> bounce, because "To:" or "Cc:" in my reply end up having the "R." part not
> quoted, like this:
> 
>   (wrong)  To: Alejandro R. Sedeño <asedeno@mit.edu>
>  (correct) To: "Alejandro R. Sedeño" <asedeno@mit.edu>

Hmm. His case has an extra level of confusion, though, because the
non-ascii characters all need rfc2047 encoding. So two emails I've seen
from him have:

  From: =?UTF-8?B?IkFsZWphbmRybyBSLiBTZWRlw7FvIg==?= <asedeno@MIT.EDU>
  From: =?UTF-8?q?Alejandro=20R=2E=20Sede=C3=B1o?= <asedeno@mit.edu>

where the first is from Icedove (i.e., Thunderbird) and the second is
from git-send-email.

The first one contains double-quotes embedded in the encoded portion.
The second one (send-email) does not.

But I'm not clear on if that is necessary. I thought that rfc2047 could
only encode a "word" in the "phrase" portion in an address header,
meaning the parsing should be unambiguous.

That being said, I think we are not quoting in the non-rfc2047 case,
anyway, and that is a bug. rfc5322 says this (section 4.1, Miscellaneous
obsolete tokens):

  Note: The "period" (or "full stop") character (".") in obs-phrase is
  not a form that was allowed in earlier versions of this or any other
  specification.  Period (nor any other character from specials) was not
  allowed in phrase because it introduced a parsing difficulty
  distinguishing between phrases and portions of an addr-spec (see
  section 4.4).  It appears here because the period character is
  currently used in many messages in the display-name portion of
  addresses, especially for initials in names, and therefore must be
  interpreted properly.

which recognizes this situation. But being in the obsolete section, I
think it is saying "you still need to interpret these, but don't
generate them". IOW, we should still be generating quotes now.

I think format-patch is totally lacking in this type of quoting. If I
do:

  $ git init
  $ git config user.name '<bogus> with "quotes"'
  $ echo contents >foo && git add . && git commit -m foo
  $ git format-patch --stdout --root
  ...
  From: bogus with "quotes <peff@peff.net>

So some of my magic characters are just stripped, and some of them get
included, making the output bogus (the stripping of <> actually happens
within git, so the commit itself is missing them).

Not that I think a name like that is sane, but probably we should be
double-quoting properly anyway, and then the "." case would just fall
out.

-Peff