I am attaching the first part of the current draft release notes to
request help from the developer community, so that we can avoid a fiasco
like we had at 1.6.0 when we switched from "git-foo" to "git foo".
The 1.6.0 change was widely advertised but the way it was introduced
allowed users to _ignore_ the issue until the last minute, and as the
consequence of having ignoring the issue, they were _forced_ to scramble
and adjust once their system administrator decided to install the new
version.
I think contributors who pushed for the updated safety valve for "git
push" did a lot better, learning from the bitter 1.6.0 experience, to
prevent users from ignoring the issue and instead prepare them for smooth
transition. I wanted to make sure all the users who read Release Notes to
the upcoming release but otherwise do not follow this list regularly will
not be surprised when 1.7.0 happens, and I feel the text needs polishing
for that purpose. We might want to deliver the announcement to a wider
audience than usual as well, and volunteers to help doing so are very
welcomed.
Two things and a half to note:
- I do not think we have such an anti-procrastination measure for
send-email's --[no-]chain-reply-to change. We might want to have one
before 1.6.6 ships; namely, if the code decided to use chain-reply-to
behaviour by default because there was no sendemail.chainreplyto (or
sendemail.$identity.chainreplyto) configured, nor --no-chain-reply-to
or --chain-reply-to given from the command line, we issue a big fat
warning just like we warn against unconfigured push.denyCurrentBranch
when allowing pushing to a checked-out branch without being told.
- I do not think of a sane way to cover "diff -b/-w" changes, as this is
a "bugfix --- but there may be some scripts that have been relying on
the bug", and a configuration option that retains the buggy behaviour
does not make much sense. But I may be mistaken and somebody can come
up with an easy patch to allow both behaviour, in which case we should
add similar anti-procrastination measure to this change.
- Recent proposal to add "grep.fulltree" configuration variable may fall
into the same category as sendemail.chainreplyto, as it is not about
"buggy behaviour" nor "risky default that can easily hurt users" but is
more about "there are two competing and equally sane defaults and it is
purely an issue of user preference". So if we were to decide flip the
default in a far future, we may add such anti-procrastination measure
there as well (but not before we decide we will flip it someday).
This last item is obviously outside the scope of 1.7.0, though.
----------------------------------------------------------------
Git v1.6.6 Release Notes
========================
Notes on behaviour change
-------------------------
* In this release, "git fsck" defaults to "git fsck --full" and
checks packfiles, and because of this it will take much longer to
complete than before. If you prefer a quicker check only on loose
objects (the old default), you can say "git fsck --no-full". This
has been supported by 1.5.4 and newer versions of git, so it is
safe to write it in your script even if you use slightly older git
on some of your machines.
Preparing yourselves for compatibility issues in 1.7.0
------------------------------------------------------
In git 1.7.0, which is planned to be the release after 1.6.6, there will
be a handful of behaviour changes that will break backward compatibility.
These changes were discussed long time ago and existing behaviours have
been identified as more problematic to the userbase than keeping them for
the sake of backward compatibility.
When necessary, transition strategy for existing users has been designed
not to force them running around setting configuration variables and
updating their scripts in order to keep the traditional behaviour on the
day their sysadmin decides to install the new version of git. When we
switched from "git-foo" to "git foo" in 1.6.0, even though the change had
been advertised and the transition guide had been provided for a very long
time, the users procrastinated during the entire transtion period, and
ended up panicking on the day their sysadmins updated their git.
For changes decided to be in 1.7.0, we have been much louder to strongly
discourage such procrastination. If you have been using recent versions
of git, you would have already seen warnings issued when you exercised
features whose behaviour will change, with the instruction on how to keep
the existing behaviour if you choose to. You hopefully should be well
prepared already.
Of course, we have also given "this and that will change in 1.7.0; prepare
yourselves" warnings in the release notes and announcement messages.
Let's see how well users will fare this time.
* "git push" into a branch that is currently checked out (i.e. pointed by
HEAD in a repository that is not bare) will be refused by default.
Similarly, "git push $there :$killed" to delete the branch $killed
in a remote repository $there, when $killed branch is the current
branch pointed at by its HEAD, will be refused by default.
Setting the configuration variables receive.denyCurrentBranch and
receive.denyDeleteCurrent to 'ignore' in the receiving repository
can be used to override these safety features. Versions of git
since 1.6.2 have issued a loud warning when you tried to do them
without setting the configuration, so repositories of people who
still need to be able to perform such a push should already been
future proofed.
Please refer to:
http://git.or.cz/gitwiki/GitFaq#non-bare
http://thread.gmane.org/gmane.comp.version-control.git/107758/focus=108007
for more details on the reason why this change is needed and the
transition process that already took place so far.
* "git send-email" will not make deep threads by default when sending a
patch series with more than two messages. All messages will be sent as
a reply to the first message, i.e. cover letter. It has been possible
to configure send-email to do this by setting sendemail.chainreplyto
configuration variable to false. The only thing the new release will
do is to change the default when you haven't configured that variable.
* "git status" will not be "git commit --dry-run". This change does not
affect you if you run the command without pathspec.
Nobody sane found the current behaviour of "git status Makefile" useful
nor meaningful, and it confused users. "git commit --dry-run" has been
provided as a way to get the current behaviour of this command since
1.6.5.
* "git diff" traditionally treated various "ignore whitespace" options
only as a way to filter the patch output. "git diff --exit-code -b"
exited with non-zero status even if all changes were about changing the
ammount of whitespace and nothing else. and "git diff -b" showed the
"diff --git" header line for such a change without patch text.
In 1.7.0, the "ignore whitespaces" will affect the semantics of the
diff operation itself. A change that does not affect anything but
whitespaces will be reported with zero exit status when run with
--exit-code, and there will not be "diff --git" header for such a
change.
Subject: prepare send-email for smoother change of --chain-reply-to default
Give a warning message when send-email uses chain-reply-to to thread the
messages because of the current default, not because the user explicitly
asked to, either from the command line or from the configuration.
This way, by the time 1.7.0 switches the default, everybody will be ready.
Signed-off-by: Nanako Shiraishi <redacted>
---
Quoting Junio C Hamano [off-list ref]
> - I do not think we have such an anti-procrastination measure for
> send-email's --[no-]chain-reply-to change. We might want to have one
> before 1.6.6 ships; namely, if the code decided to use chain-reply-to
> behaviour by default because there was no sendemail.chainreplyto (or
> sendemail.$identity.chainreplyto) configured, nor --no-chain-reply-to
> or --chain-reply-to given from the command line, we issue a big fat
> warning just like we warn against unconfigured push.denyCurrentBranch
> when allowing pushing to a checked-out branch without being told.
Like this?
git-send-email.perl | 19 +++++++++++++++++--
t/t9001-send-email.sh | 40 ++++++++++++++++++++++++++++++++++++++++
2 files changed, 57 insertions(+), 2 deletions(-)
diff --git a/git-send-email.perl b/git-send-email.perl
index 4f5da4e..2afed76 100755
--- a/git-send-email.perl
+++ b/git-send-email.perl
@@ -187,9 +187,11 @@ my ($identity, $aliasfiletype, @alias_files, @smtp_host_parts);
my ($validate, $confirm);
my (@suppress_cc);
+my $not_set_by_user = "true but not set by the user";
+
my %config_bool_settings = (
"thread" => [\$thread, 1],
- "chainreplyto" => [\$chain_reply_to, 1],
+ "chainreplyto" => [\$chain_reply_to, $not_set_by_user],
"suppressfrom" => [\$suppress_from, undef],
"signedoffbycc" => [\$signed_off_by_cc, undef],
"signedoffcc" => [\$signed_off_by_cc, undef], # Deprecated@@ -214,6 +216,19 @@ my %config_settings = (
"from" => \$sender,
);
+# Help users prepare for 1.7.0
+sub chain_reply_to {
+ if (defined $chain_reply_to &&
+ $chain_reply_to eq $not_set_by_user) {
+ print STDERR
+ "In git 1.7.0, the default will be changed to --no-chain-reply-to\n" .
+ "Set sendemail.chainreplyto configuration variable to true if\n" .
+ "you want to keep --chain-reply-to as your default.\n";
+ $chain_reply_to = 1;
+ }
+ return $chain_reply_to;
+}
+
# Handle Uncouth Termination
sub signal_handler {
@@ -1157,7 +1172,7 @@ foreach my $t (@files) {
# set up for the next message
if ($thread && $message_was_sent &&
- ($chain_reply_to || !defined $reply_to || length($reply_to) == 0)) {
+ (chain_reply_to() || !defined $reply_to || length($reply_to) == 0)) {
$reply_to = $message_id;
if (length $references > 0) {
$references .= "\n $message_id";diff --git a/t/t9001-send-email.sh b/t/t9001-send-email.sh
index 84a7f03..4372774 100755
--- a/t/t9001-send-email.sh
+++ b/t/t9001-send-email.sh
@@ -769,4 +769,44 @@ test_expect_success 'threading but no chain-reply-to' '
grep "In-Reply-To: " stdout
'
+test_expect_success 'no warning with an explicit --chain-reply-to' '
+ git send-email \
+ --dry-run \
+ --from="Example <nobody@example.com>" \
+ --to=nobody@example.com \
+ --chain-reply-to \
+ outdir/000?-*.patch 2>errors >out &&
+ ! grep "no-chain-reply-to" errors
+'
+
+test_expect_success 'no warning with an explicit --no-chain-reply-to' '
+ git send-email \
+ --dry-run \
+ --from="Example <nobody@example.com>" \
+ --to=nobody@example.com \
+ --no-chain-reply-to \
+ outdir/000?-*.patch 2>errors >out &&
+ ! grep "no-chain-reply-to" errors
+'
+
+test_expect_success 'no warning with sendemail.chainreplyto = false' '
+ git config sendemail.chainreplyto false &&
+ git send-email \
+ --dry-run \
+ --from="Example <nobody@example.com>" \
+ --to=nobody@example.com \
+ outdir/000?-*.patch 2>errors >out &&
+ ! grep "no-chain-reply-to" errors
+'
+
+test_expect_success 'no warning with sendemail.chainreplyto = true' '
+ git config sendemail.chainreplyto true &&
+ git send-email \
+ --dry-run \
+ --from="Example <nobody@example.com>" \
+ --to=nobody@example.com \
+ outdir/000?-*.patch 2>errors >out &&
+ ! grep "no-chain-reply-to" errors
+'
+
test_done
--
Nanako Shiraishi
http://ivory.ap.teacup.com/nanako3/
Junio C Hamano [off-list ref] writes:
- I do not think of a sane way to cover "diff -b/-w" changes, as this is
a "bugfix --- but there may be some scripts that have been relying on
the bug", and a configuration option that retains the buggy behaviour
does not make much sense. But I may be mistaken and somebody can come
up with an easy patch to allow both behaviour, in which case we should
add similar anti-procrastination measure to this change.
The patch to enable/disable this feature would look like the attached,
which doesn't look too bad. This apply to a merge between two topics:
jc/1.7.0-diff-whitespace-only-status (97bf2a0)
gb/1.7.0-diff-whitespace-only-output (3e97c7c)
An anti-procrastination measure must be done in at least two, but probably
in three steps on top of this patch:
#1. Flip the default in this patch back to 1, as the traditional
behaviour is to treat ignore-whitespace options as purely affecting
output of patch text, not affecting status nor the header;
#2. Add extra logic to detect if a particular invocation may trigger a
different behaviour once the default is changed to 0, and issue an
warning if there is no configuration explicitly set;
#3. Flip the default to 0, perhaps still keeping the warning when there
is no configuration.
Make a major release after the second step is done, and wait for at least
one but preferrably two releases, and then ship the result of the third
step in the "flag day" release. In a much later version we would remove
the "no configuration warning".
Do we need an anti-procrastination measure for this? If so, I think we
probably need to postpone the "diff" semantics changes after 1.7.0, as I
am not very confident that we can solidly finish the second step before
the 1.6.6 final.
Also, I am not very keen on having this configuration, as its only purpose
is to ask for a broken semantics from "diff", even though it has been the
traditional behaviour for the past four years.
Comments?
---
diff.c | 15 +++++++++++----
1 files changed, 11 insertions(+), 4 deletions(-)
diff --git a/diff.c b/diff.c
index da90e6e..6e7c47c 100644
--- a/diff.c
+++ b/diff.c
@@ -29,6 +29,7 @@ static const char *diff_word_regex_cfg;
static const char *external_diff_cmd_cfg;
int diff_auto_refresh_index = 1;
static int diff_mnemonic_prefix;
+static int diff_b_w_output_only;
static char diff_colors[][COLOR_MAXLEN] = {
GIT_COLOR_RESET,@@ -133,6 +134,11 @@ int git_diff_basic_config(const char *var, const char *value, void *cb)
return 0;
}
+ if (!strcmp(var, "diff.bwoutputonly")) {
+ diff_b_w_output_only = git_config_bool(var, value);
+ return 0;
+ }
+
return git_color_default_config(var, value, cb);
}
@@ -1668,7 +1674,7 @@ static void builtin_diff(const char *name_a,
struct emit_callback ecbdata;
const struct userdiff_funcname *pe;
- if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS)) {
+ if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || diff_b_w_output_only) {
fprintf(o->file, "%s", header.buf);
strbuf_reset(&header);
}@@ -2537,9 +2543,10 @@ int diff_setup_done(struct diff_options *options)
* inside contents.
*/
- if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
- DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
- DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
+ if (!diff_b_w_output_only &&
+ (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
+ DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
+ DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL)))
DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
else
DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
--
1.6.6.rc0.61.g41d5b.dirty