Commit 4796e823 ("bisect: introduce --no-checkout support into porcelain." Aug 4 2011)
made checking out the branch where we started depends on the "checkout" mode. But
unfortunately it lost the "|| exit" part after the checkout command.
As it makes no sense to continue if the checkout failed and as people have already
complained that the error message given when we just exit in this case is not clear, see:
http://thread.gmane.org/gmane.comp.version-control.git/180733/
this patch adds a "|| die <hopefully clear message>" part after the checkout command.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Acked-by: Jon Seymour <jon.seymour@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This enhances the support for bisecting history in bare repositories.
The "git bisect" command no longer needs to be run inside a repository
with a working tree; it defaults to --no-checkout when run in a bare
repository.
Two tests are included to demonstrate this behaviour.
Suggested-by: Junio C Hamano <gitster@pobox.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Jon Seymour <jon.seymour@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a few remaining lines that indented with spaces.
Also simplify the logic of checking out the original branch and reporting
error during "bisect reset".
Signed-off-by: Junio C Hamano <gitster@pobox.com>
All leading whitespace is now encoded with tabs.
After this patch, the following is true:
RAW=$(cat git-bisect.sh | md5sum) &&
ROUNDTRIP=$(cat git-bisect.sh | expand -i - | unexpand --first-only - | md5sum) &&
LEADING=$(sed -n "/^ */p" < git-bisect.sh | wc -l) &&
test $RAW = $ROUNDTRIP &&
test $LEADING = 0 &&
test -z "$(git diff -w HEAD~1 HEAD)"
Signed-off-by: Jon Seymour <jon.seymour@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-bisect can now perform bisection of a history without performing
a checkout at each stage of the bisection process. Instead, HEAD is updated.
One use-case for this function is allow git bisect to be used with
damaged repositories where git checkout would fail because the tree
referenced by the commit is damaged.
It can also be used in other cases where actual checkout of the tree
is not required to progress the bisection.
Improved-by: Christian Couder <chriscool@tuxfamily.org>
Improved-by: Junio C Hamano <gitster@pobox.com>
Improved-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Jon Seymour <jon.seymour@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Christian Couder pointed out that the existing eval strategy
swallows an initial non-zero return. Using && to connect
the statements should fix this.
Signed-off-by: Jon Seymour <jon.seymour@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Currently 'git bisect start' modifies some state prior to checking
that its arguments are valid.
This change moves argument validation before state modification
with the effect that state modification does not occur
unless argument validations succeeds.
An existing test is changed to check that new bisect state
is not created if arguments are invalid.
A new test is added to check that existing bisect state
is not modified if arguments are invalid.
Signed-off-by: Jon Seymour <jon.seymour@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Gettextize the "You need to start by" message in
bisect_next_check. This message assembled English output by hand so it
needed to be split up to make it translatable.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Gettextize the [Y/n] questions git-bisect presents, and leave a note
in a TRANSLATORS comment explaining that translators have to preserve
a mention of the Y/n characters since the program will expect them,
and not their localized equivalents.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Gettextize bisect_replay messages that use the $1 variable. Since it's
subroutine local we have to provide an alias for it for eval_gettext.
Since I was doing that anyway I've changed all other uses of $1
variable to use the alias variable for clarity.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
ettextize bisect_reset messages that use the $1 variable. Since it's
subroutine local we have to provide an alias for it for eval_gettext.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Gettextize bisect_run messages that use the $@ variable. Since it's
subroutine local we have to provide an alias for it for eval_gettext.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If gitk is not available in the PATH, bisect ends up
exiting with the shell's 127 error code, confusing the git
wrapper into thinking that bisect is not a git command.
We already fallback to git-log if there doesn't seem to be a
graphical display available. We should do the same if gitk
is not available in our PATH at all. This not only fixes the
ugly error message, but is a much more sensible default than
failing to show the user anything.
Reported by Maxin John.
Tested-by: Maxin B. John <maxin@maxinbjohn.info>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
'git bisect replay' has a mandatory logfile argument, but the current
implementation doesn't check whether the user has specified one. When
the user omits the logfile argument, this leads to the following
unhelpful error message:
cannot read for replaying
So, check for the mandatory argument first, and provide a more
meaningful error message when it is omitted.
Signed-off-by: SZEDER Gábor <szeder@ira.uka.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
'git bisect reset' (without the optional <commit> argument) returns to
the original HEAD from where the bisection was started. However,
when, for whatever reason, the user deleted the original HEAD before
invoking 'git bisect reset', then all he gets is an error message from
'git checkout':
fatal: invalid reference: somebranch
Let's try to be more helpful with an error message better describing
what went wrong and a suggestion about how to resolve the situation:
Could not check out original HEAD 'somebranch'. Try 'git bisect reset <commit>'.
Signed-off-by: SZEDER Gábor <szeder@ira.uka.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
'git bisect log' is implemented by a direct invocation of 'cat
"$GIT_DIR/BISECT_LOG"', without any sanity checks. Consequently,
running 'git bisect log' while not bisecting leads to an error,
because the bisect logfile doesn't exists. The accompanying error
message
cat: /path/to/repo/.git/BISECT_LOG: No such file or directory
is neither very helpful nor very friendly.
Instead of blindly trying to cat the log file, first check whether
there is a bisection going on (i.e. the bisect logfile exists), and
die with a more appropriate error message when not.
Signed-off-by: SZEDER Gábor <szeder@ira.uka.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If the user has exported the GREP_OPTIONS environment variable, the output
from "grep" and "egrep" in scripted Porcelains may be different from what
they expect. For example, we may want to count number of matching lines,
by "grep" piped to "wc -l", and GREP_OPTIONS=-C3 will break such use.
The approach taken by this change to address this issue is to protect only
our own use of grep/egrep. Because we do not unset it at the beginning of
our scripts, hook scripts run from the scripted Porcelains are exposed to
the same insanity this environment variable causes when grep/egrep is used
to implement logic (e.g. "grep | wc -l"), and it is entirely up to the
hook scripts to protect themselves.
On the other hand, applypatch-msg hook may want to show offending words in
the proposed commit log message using grep to the end user, and the user
might want to set GREP_OPTIONS=--color to paint the match more visibly.
The approach to protect only our own use without unsetting the environment
variable globally will allow this use case.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In commit ad3f9a7 (Add '--bisect' revision machinery argument) the
'--bisect' option was added to easily pass bisection refs to
commands using the revision machinery.
So it is now shorter and safer to use the new '--bisect' revision
machinery option, than to compute the refs that we must pass.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
‘git bisect reset’ accepts an optional argument specifying a branch to
check out after cleaning up the bisection state. This lets you
specify an arbitrary commit.
In particular, this provides a way to clean the bisection state
without moving HEAD: ‘git bisect reset HEAD’. This may be useful if
you are not interested in the state before you began a bisect,
especially if checking out the old commit would be expensive and
invalidate most of your compiled tree.
Clarify the ‘git bisect reset’ documentation to explain this optional
argument, which was previously mentioned only in the usage message.
Signed-off-by: Anders Kaseorg <andersk@mit.edu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch replace the "--next-exit" option of "git bisect--helper"
with a "--next-all" option that does merge base checking using
the "check_good_are_ancestors_of_bad" function implemented in
"bisect.c" in a former patch.
The new "--next-all" option is then used in "git-bisect.sh" instead
of the "--next-exit" option, and all the shell functions in
"git-bisect.sh" that are now unused are removed.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As the "sq" function was the only place using Perl in "git-bisect.sh",
this removes the Perl dependency in this script.
While at it, we also remove the sed instruction in the Makefile that
substituted @@PERL@@ with the Perl path in shell scripts, as this is
not needed anymore. (It is now only needed in "git-instaweb.sh" but
this command is dealt with separately in the Makefile.)
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
instead of "git bisect--helper --next-vars".
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When doing:
eval "git bisect--helper --next-vars" | {
while read line
do
echo "$line &&"
done
echo ':'
}
the result code comes from the last "echo ':'", not from running
"git bisect--helper --next-vars".
This patch gets rid of the need to string together the line from
the output of "git bisect--helper" with "&&" in the calling script
by making "git bisect--helper --next-vars" return output variables
already in that format.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Use the new "git bisect--helper" builtin. It should be faster and
safer instead of the old "filter_skipped" shell function. And it
is a first step to move more shell code to C.
As the output is a little bit different we have to change the code
that interpret the results. But these changes improve code clarity.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In "git-bisect.sh" the "git checkout" command is only used to
change the current branch, but it is used like this:
git checkout "$branch"
which will output the following misleading error message when
it fails:
error: pathspec 'foo' did not match any file(s) known to git.
This patch change the way we use "git checkout" like this:
git checkout "$branch" --
so that we will get the following error message:
fatal: invalid reference: foo
which is better.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch teaches "git rev-list --bisect-vars" to output an estimate
of the number of bisection step left _after the current one_ along with
the other variables it already outputs.
This patch also makes "git-bisect.sh" display this number of steps left
_after the current one_, along with the estimate of the number of
revisions left to test (after the current one).
Here is a table to help analyse what should be the best estimate for
the number of bisect steps left.
N : linear case --> probabilities --> best
-------------------------------------------------------------
1 : G-B --> 0 --> 0
2 : G-U1-B --> 0 --> 0
3 : G-U1-U2-B --> 0(1/3) 1(2/3) --> 1
4 : G-U1-U2-U3-B --> 1 --> 1
5 : G-U1-U2-U3-U4-B --> 1(3/5) 2(2/5) --> 1
6 : G-U1-U2-U3-U4-U5-B --> 1(2/6) 2(4/6) --> 2
7 : G-U1-U2-U3-U4-U5-U6-B --> 1(1/7) 2(6/7) --> 2
8 : G-U1-U2-U3-U4-U5-U6-U7-B --> 2 --> 2
9 : G-U1-U2-U3-U4-U5-U6-U7-U8-B --> 2(7/9) 3(2/9) --> 2
10: G-U1-U2-U3-U4-U5-U6-U7-U8-U9-B --> 2(6/10)3(4/10)--> 2
In the column "N", there is the number of revisions that could _now_
be the first bad commit we are looking for.
The "linear case" column describes the linear history corresponding to
the number in column N. G means good, B means bad, and Ux means
unknown. Note that the first bad revision we are looking for can be
any Ux or B.
In the "probabilities" column, there are the different outcomes in
number of steps with the odds of each outcome in parenthesis
corresponding to the linear case.
The "best" column gives the most accurate estimate among the different
outcomes in the "probabilities" column.
We have the following:
best(2^n) == n - 1
and for any x between 0 included and 2^n excluded, the probability for
n - 1 steps left looks like:
P(2^n + x) == (2^n - x) / (2^n + x)
and P(2^n + x) < 0.5 means 2^n < 3x
So the algorithm used in this patch calculates 2^n and x, and then
choose between returning n - 1 and n.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When there is nothing to be skipped, the output from
rev-list --bisect-vars was eval'ed without first being
strung together with &&; this is probably not a problem
as it is much less likely to be a bad input than the list
handcrafted by the filter_skip function, but it still is
a good discipline.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the "bad" commit was also "skip"ped and when more than one
commit was skipped, the "filter_skipped" function would have
printed something like:
bisect_rev=<hash1>|<hash2>
(where <hash1> and <hash2> are hexadecimal sha1 hashes)
and this would have been evaled later as piping "bisect_rev=<hash1>"
into "<hash2>", which would have failed.
So this patch makes the "filter_skipped" function properly quote
what it outputs, so that it will print something like:
bisect_rev='<hash1>|<hash2>'
which will be properly evaled later. The caller was not stopping
properly because the scriptlet this function returned to be evaled
was not strung together with && and because of this, an error in
an earlier part of the output was simply ignored.
A test case is added to the test suite.
And while at it, we also initialize the VARS, FOUND and TRIED
variables, so that we protect ourselves from environment variables
the user may have with these names.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It seems that Cygwin sets the variable SESSIONNAME when an interactive
desktop session is running, and does not set it when you log in via ssh.
So we can use this variable to determine whether to run gitk or git log
in git bisect view.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The patch that allows "git bisect skip" to be passed a range of
commits using the "<commit1>..<commit2>" notation is flawed because
it introduces a regression when it was passed a simple rev or commit.
"git bisect skip <commit>" doesn't work any more, because <commit>
is quoted but not properly unquoted.
This patch fixes that and add tests cases to better check when it is
passed commits and range of commits.
While at it, this patch also properly quotes the non range arguments
using the "sq" function.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
The current "git bisect skip" syntax is "git bisect skip [<rev>...]"
so it's already possible to skip a range of revisions using
something like:
$ git bisect skip $(git rev-list A..B)
where A and B are the bounds of the range we want to skip.
This patch teaches "git bisect skip" to accept:
$ git bisect skip A..B
as an abbreviation for the former command.
This is done by checking each argument to see if it contains two
dots one after the other ('..'), and by expending it using
"git rev-list" if that is the case.
Note that this patch will not make "git bisect skip" accept all
that "git rev-list" accepts, as things like "^A B" for exemple
will not work. But things like "A B..C D E F.. ..G H...I" should
work as expected.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Check to see given bad/good/skip sets are valid commit and to exit
otherwise was broken by 6a54d97 (bisect: remove "checkout_done" variable
used when checking merge bases, 2008-09-06).
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Using return values from the following functions:
- check_merge_bases
- check_good_are_ancestors_of_bad
seems simpler.
While at it, let's add some comments to better document the above
functions.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When one good revision is not an ancestor of the bad revision, the
merge bases between the good and the bad revision should be checked
to make sure that they are also good revisions.
A previous patch takes care of that, but it may check the merge bases
more often than really needed. In fact the previous patch did not try
to optimize this as much as possible because it is not so simple. So
this is the purpose of this patch.
One may think that when all the merge bases have been checked then
we can save a flag, so that we don't need to check the merge bases
again during the bisect process.
The problem is that the user may choose to checkout and test
something completely different from what the bisect process
suggested. In this case we have to check the merge bases again,
because there may be new merge bases relevant to the bisect
process.
That's why, in this patch, when we detect that the user tested
something else than what the bisect process suggested, we remove
the flag that says that we don't need to check the merge bases
again.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Before this patch, "git bisect", when it was given some good revs that
are not ancestor of the bad rev, didn't check if the merge bases were
good. "git bisect" just supposed that the user knew what he was doing,
and that, when he said the revs were good, he knew that it meant that
all the revs in the history leading to the good revs were also
considered good.
But in pratice, the user may not know that a good rev is not an
ancestor of the bad rev, or he may not know/remember that all revs
leading to the good rev will be considered good. So he may give a good
rev that is a sibling, instead of an ancestor, of the bad rev, when in
fact there can be one rev becoming good in the branch of the good rev
(because the bug was already fixed there, for example) instead of one
rev becoming bad in the branch of the bad rev.
For example, if there is the following history:
A--B--C--D
\
E--F
and we launch "git bisect start D F" then only C and D would have been
considered as possible first bad commit before this patch. This could
invite user errors; F could be the commit that fixes the bug that exists
everywhere else.
The purpose of this patch is to detect when "git bisect" is passed
some good revs that are not ancestors of the bad rev, and then to first
ask the user to test the merge bases between the good and bad revs.
If the merge bases are good then all is fine, we can continue
bisecting. Otherwise, if one merge base is bad, it means that the
assumption that all revs leading to the good one are good too is
wrong and we error out. In the case where one merge base is skipped we
issue a warning and then continue bisecting anyway.
These checks will also catch the case where good and bad have been
mistaken. This means that we can remove the check that was done latter
on the output of "git rev-list --bisect-vars".
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Given that users are supposed to type 'git bisect' now, make the output
of 'git bisect log' consistent with this.
Signed-off-by: Miklos Vajna <vmiklos@frugalware.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It seems simpler and safer to use the BISECT_START file everywhere
to decide if we are bisecting or not, instead of using it in some
places and BISECT_NAMES in other places.
In commit 6459c7c678 (Nov 18 2007,
Bisect: use "$GIT_DIR/BISECT_NAMES" to check if we are bisecting.),
we decided to use BISECT_NAMES but code changed a lot and we now
have to check BISECT_START first in the "bisect_start" function
anyway.
This patch also makes things a little bit safer by creating
the BISECT_START file first and deleting it last, and also by
adding checks in "bisect_clean_state".
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "git bisect" was first written, it was not possible to
checkout a detached HEAD. The detached feature appeared latter.
That's why before this patch the "git bisect" process used a
"bisect" branch to checkout new revisions to be tested (and also
a "new-bisect" one to check if the checkouts could work).
This patch makes "git bisect" checkout revisions to be tested on
a detached HEAD. This simplifies the code a bit.
The tests to check that "git bisect" does not start if a
"bisect" or a "new-bisect" branch exists are removed as they
are not relevant any more.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Before this patch, when using "git bisect start" with mistaken revs
or when the checkout of the branch we want to test failed, we exited
after having written files like ".git/BISECT_START",
".git/BISECT_NAMES" and after having written "refs/bisect/bad" and
"refs/bisect/good-*" refs.
With this patch we trap all errors that can happen when writing the
new state and when we are in "bisect_next". So that we can try to
clean up everything in case of problems, using "bisect_clean_state".
This patch also contains a "bisect_write" cleanup to make it exit
on error and return 0 otherwise.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>