Asking xapian to sort the messages for us causes suboptimal IO patterns. This
would be useful, if we only wanted the first few results, but since we want
everything anyway, this is pessimization.
On 2011-10-29, a measurement on a 372981 messages instance showed that wall
time can be reduced from 28 minutes (sorted by Message-ID) to 15 minutes
(unsorted).
Timings on 189605 messages:
$ time notmuch.old dump
19.48user 5.83system 12:10.42elapsed 3%CPU (0avgtext+0avgdata 110656maxresident)k
3629584inputs+22720outputs (33major+7073minor)pagefaults 0swaps
$ echo 3 > /proc/sys/vm/drop_caches
$ time notmuch.new
14.89user 1.20system 3:23.58elapsed 7%CPU (0avgtext+0avgdata 46032maxresident)k
1256264inputs+22464outputs (43major+1990minor)pagefaults 0swaps
This optimizes the user's tagging query to exclude messages that won't
be affected by the tagging operation, saving computation and IO for
redundant tagging operations.
For example,
notmuch tag +notmuch to:notmuch@notmuchmail.org
will now use the query
( to:notmuch@notmuchmail.org ) and (not tag:"notmuch")
In the past, we've often suggested that people do this exact
transformation by hand for slow tagging operations. This makes that
unnecessary.
If emacs is not available, test_expect_equal would be called with only
one argument. The patch fixes this by quoting the (possibly empty)
$(cat OUTPUT) argument.
Some tests (e.g. crypto) do a common initialization required for all
subtests. The patch adds a check for missing external dependencies
during this initialization. If any prerequisites are missing, all
subtests are skipped.
The check is run on the first call of test_reset_state_ function, so
no changes for the tests are needed.
There is existing support for general prerequisites in the test suite.
But it is not very convenient to use: every test case has to keep
track for it's dependencies and they have to be explicitly listed.
The patch aims to add better support for a particular type of external
dependencies: external executables. The main idea is to replace
missing external binaries with shell functions that have the same
name. These functions always fail and keep track of missing
dependencies for a subtest. The result reporting functions later can
check that an external binaries are missing and correctly report SKIP
result instead of FAIL. The primary benefit is that the test cases do
not need to declare their dependencies or be changed in any way.
If symbol-test is built in symbol-hiding with hardcoded g++ invokation,
it's not so easy to pass $(srcdir) which is required to find notmuch.h
when srcdir and builddir are separate directories.
Basic test 'Ensure that all available tests will be run by
notmuch-test' compares all tests that are run with listing of test/
directory. There is a growing list of exceptions for files and
directories which located in the test/ directory but are not tests.
Moreover some (probably buggy) tests do create files in the the test/
directory which may be left behind in case of failure. This makes the
basic test fail.
The patch changes the test to look only for regular executable files.
This makes the exception list much smaller. And since no tests should
create executables in the test/ directory (if there are, they should
be fixed), the basic test should not be affected by failed or
interrupted tests.
The error is easy to miss, because the test passes and stderr is not
printed. But if you run basic tests in verbose mode (./basic
--verbose), you get:
sed: can't read notmuch-test: No such file or directory
The issue is that sed command is given two files: notmuch-test and
$TEST_DIRECTORY/notmuch-test. And there is no notmuch-test file in
the current directory (test/tmp.basic/). The patch just removes the
non-existing file from the sed command.
Previous behaviour was to indent messages in a thread according
to depth by one space per level. This commit tests if setting
notmuch-indent-messages-width to `4' provides a message thread
with four spaces of indentation thread depth.
Previous behaviour was to indent messages in a thread according
to depth by one space per level. In
id:1311028119-50637-1-git-send-email-fgeller@gmail.com Felix
Geller proposed a patch in order to turn indentation off. This
commit tests if instead setting notmuch-indent-messages-width to
`0' does turn off indentation.
Previous behaviour was to indent messages in a thread according
to depth by one space per level. This is still the case with
notmuch-indent-messages-width default value `1'. This test
succeeds if output with default value is same as in "Basic
notmuch-show view in emacs".
Till now Emacs UI indents messages according to their respecive
depth of neting in the thread. The actual width of indentation
per level is hardcoded to `1' space.
This patch makes message indentation customisable by introducing
a variable `notmuch-indent-messages-width' which defaults to `1',
which is the same as before. Felix could set this variable to
`0' in order to disable indentation, I tested it with a value of
`4' for a clearer separation of messages in a thread.
When I changed the calling convention to pass in all but the zero-th
argument to subcommands, I missed this one call, resulting in a
segmentation fault.
As a bonus, the syntax "notmuch --help foo" is now equivalent to
"notmuch help foo".
Use common sed tool instead of dpkg-parsechangelog (which is usually
available on debian systems only) to verify that debian version
information is consistent with version file.
verfy-version-debian, verify-version-python and verify-version-components
checked noneqality of the comparison strings and if got "positive"
answer then made that goal fail. But in case of the test ([ ])
execution failed it never got to the 'then' part of the line (and
the 'if [ ... ] then ... fi ' construct doesn't make the script line
fail in case of problems inside [ ].
This commit inverses the "logic", so that only if the comparison for
equality succeeds the script line will exit with 0 and execution
can continue past the failure case to the next line (executed by another
shell) with '@echo done'
There's no reason to record undo information for read-only,
programmatically-constructed buffers. The undo list just chews up
memory keeping track of our calls to insert.
Since version 0.8 of dtach -n does no longer require controlling
tty to be present when executed. Currently controlling tty is not
always (if ever) present when tests are executed.
Add new customization option notmuch-saved-search-sort-function to sort
saved searches in user-defined order. Provide a sort function to sort the
saved searches in alphabetical order. Setting the search function to nil
causes the saved searches not to be sorted, as before. This also remains
the default. The function only affects display of the saved searches, not
the order in which they are stored by custom.
Signed-off-by: Jani Nikula <jani@nikula.org>
This patch breaks out much of the functionality of
notmuch-show-advance-and-archive into a new function:
notmuch-show-advance. This new function does all the advancing
through a show buffer that notmuch-show-advance-and-archive did,
without all the invasive thread archiving. The return value of
notmuch-show-advance is nil if the bottom of the thread is not
reached, and t if it is.
notmuch-show-advance-and-archive is modified to just call
notmuch-show-advance, and then call notmuch-show-archive-thread if the
return value is true. In this way the previous functionality of
notmuch-show-advance-and-archive is preserved.
This provides a way for people to rebind the space bar to a more sane
function if they don't like the default behavior.
For some reason, on my machine, the link is picking up
/usr/lib/libutil.so instead of util/libutil.a. This causes there to be
undefined symbols in libnotmuch, making it unuseable. This patch causes
the link to fail instead.
The instructions are purposely a bit coy about what files are updated,
so we don't have to update immediately when something else is plugged
into the make recipe.
