Re: [PATCH v3] post-receive-email: explicitly set Content-Type header

Alexey Shumkin [off-list ref] writes:

quoted

Some email clients (e.g. claws-mail) incorrectly display
message body when there is no Content-Type header and charset
explicitly defined.
So, set explicitly Content-Type header. Its charset
can be defined with hooks.emailcharset config variable.

NB: This above-mentioned charset may differ from
i18n.logOutputEncoding, because e.g. gitweb expects (for now)
i18n.logOutputEncoding set to UTF-8 to display logs correctly.

Also, introduce hooks.gitopts config variable
with the default '-c core.quotepath=false'.
This takes into account that we want to see pretty email-message
with well-looking messages and list of changed filenames.
And usually non-ASCII filenames are in the same
encoding that commit messages are.

(style) Why such an extremely ragged looking line-wrap of paragraphs?

I'm not good enough in English spelling ;(

quoted

Signed-off-by: Alexey Shumkin <redacted>
---

In this space, please describe what happened during v1 and v2, and
how is this round different to help reviewers. Pointers to list
archive, e.g.
http://thread.gmane.org/gmane.comp.version-control.git/181737, would
be helpful.

People involved in v1/v2 discussion are missing from the Cc: line.
Please do not give a false impression that you are hiding from them.

Oh, I've missed that moment, sorry all

quoted

diff --git a/contrib/hooks/post-receive-email

b/contrib/hooks/post-receive-email index ba077c1..913be89 100755

--- a/contrib/hooks/post-receive-email
+++ b/contrib/hooks/post-receive-email

@@ -65,6 +65,14 @@
 #   Default is "--stat --summary --find-copies-harder". Add -p to

those #   options to include a unified diff of changes in addition
to the usual #   summary output.
+# hooks.gitopts
+#   git options for the git diff-tree invocation that shows
changes. +#   Default is '-c core.quotepath=false' to be able to
see non-ASCII filenames +#   used in a project.

We do not particularly appreciate a patch that does two unrelated
things ("they are both related to post-receive-email" is not an
argument). Wouldn't this be useful even if the change to add
hooks.emailcharset turned out to be unwanted, or vice versa?

The main reason was that using core.quotepath=false leads to showing
file names "correctly" according to commit messages encoding to make
email-message look pretty.

quoted

+# hooks.emailcharset
+#   The charset used in Content-Type header. UTF-8, if not
specified. +#   It can differ from i18n.logOutputEncoding (not to
mess-up with gitweb +#   which expects i18n.logOutputEncoding to be
set to UTF-8)

Why "UTF-8" instead of "i18n.logoutputencoding" if not specified?

Well, you're right.
For the explanation:
AFAIU, such hooks are used on central servers to notify involved people
about changes. And AFAIU the same server repos are used with gitweb
(which AFAIK requires i18n.logOutputEncoding=UTF-8)

But in common case "i18n.logoutputencoding" is more suitable.

quoted

@@ -234,6 +242,9 @@ generate_email_header()
 	cat <<-EOF
 	To: $recipients
 	Subject: ${emailprefix}$projectdesc $refname_type

$short_refname ${change_type}d. $describe
+	MIME-Version: 1.0
+	Content-Type: text/plain; charset=$emailcharset
+	Content-Transfer-Encoding: 8bit
 	X-Git-Refname: $refname
 	X-Git-Reftype: $refname_type
 	X-Git-Oldrev: $oldrev
...

@@ -730,6 +734,19 @@ custom_showrev=$(git config hooks.showrev)
 maxlines=$(git config hooks.emailmaxlines)
 diffopts=$(git config hooks.diffopts)
 : ${diffopts:="--stat --summary --find-copies-harder"}
+gitopts=$(git config hooks.gitopts || echo '-c

core.quotepath=false') +emailcharset=$(git config
hooks.emailcharset || echo 'UTF-8') +
+projectdesc=$(sed -ne '1p' "$GIT_DIR/description" 2>/dev/null)
+# Check if the description is unchanged from it's default, and
shorten it to +# a more manageable length if it is
+if expr "$projectdesc" : "Unnamed repository.*$" >/dev/null
+then
+	projectdesc="UNNAMED PROJECT"
+fi
+# Leave description in UTF-8 to be used in the Subject header
+# But convert it to an hooks.emailcharset encoding to be used in a
message body +projectdesc_e=$(echo $projectdesc | iconv -f UTF-8 -t
$emailcharset 2>/dev/null)

Hmm, this generates a piece of e-mail whose subject line is in UTF-8
(without B/Q quoting) and message body is in totally different
encoding. Is it what mailers really want to see?

Here you're right, too. Windows email clients may interpret Subject
header without B/Q quoting in its default Windows charset, and as far as
it may contain non-English project description, so Subject would look
ugly. But I'll try to test with some clients.

It almost seems backwards; converting the payload to UTF-8 and always
sending UTF-8 would be a simpler approach, methinks.

Sounds reasonable