In xapian terms, convert folder: prefix from probabilistic to boolean
prefix, matching the paths, relative from the maildir root, of the
message files, ignoring the maildir new and cur leaf directories.
folder:foo matches all message files in foo, foo/new, and foo/cur.
folder:foo/new does *not* match message files in foo/new.
folder:"" matches all message files in the top level maildir and its
new and cur subdirectories.
This change constitutes a database change: bump the database version
and add database upgrade support for folder: terms. The upgrade also
adds path: terms.
Finally, fix the folder search test for literal folder: search, as
some of the folder: matching capabilities are lost in the
probabilistic to boolean prefix change.
We will need this for improved folder search tests, but having some
folders should exercise our code paths better anyway.
Modify the relevant test accordingly to make it pass.
This reorganization triggers a bug in the test suite, namely that it
expects the output of --output=files to be in a certain order. So we
add the fix for that into the same commit.
This mainly involves sorting, although the case --duplicate=$n
requires more subtlety.
Sanitize tabs and newlines to spaces rather than question marks in
--output=summary --format=text output.
This will also hide any difference in unfolding a header that has been
folded with a tab. Our own header parser replaces tabs with spaces,
while gmime would retain the tab.
The printf builtin "%(fmt)T" specifier (which allows time values
to use strftime-like formatting) is introduced in bash 4.2.
Trying to execute this in pre-4.2 bash will fail -- and if this
happens execute the fallback piece of perl code to do the same thing.
The test names assigned to NOTMUCH_SKIP_TESTS variable can now be given
with or without the Tddd- prefix for tester convenience:
The test name without Tddd -prefix stays constant even when test filenames
are renumbered.
The test name with Tddd -prefix is printed out when tests run.
According the semantics of make, the expansion of $(dir) in recipes
uses dynamic scope, i.e. the value at the time the recipe is run. This
means if test/Makefile.local is not the last sub-makefile included,
all heck breaks loose.
Previously, we stripped the "Tnnn-" part from the test name when
printing its description at the beginning of each test. However, this
makes it difficult to find the source script for a test (e.g., when a
test fails). Put this prefix back.
Searching by Message-Id no longer works via the old mail-archive.com
API, though I have contacted them in hopes that they restore it to
prevent dead links. Anyway, the new API is cleaner.
Acked-by: Austin Clements <amdragon@MIT.EDU>
Script `notmuch-test` expects the results file have T\d\d\d- part
intact so the results files (and some test output files) are now
name as such.
Without this change `notmuch-test` will exit in case the test
script it was executing exited with nonzero value.
The T\d\d\d- part is dropped in new variable $this_test_bare which is
used in progress informational messages and when loading .el files in
emacs tests (whenever $this_test_bare.el exists).
If there is a syntax error in the emacs test library, it causes other
tests to hang or crash without a useful error message.
This test could be eliminated if the error reporting for emacs tests
was somehow improved.
All test scripts to be executed are now named as T\d\d\d-name.sh,
numers in increments of 10.
This eases adding new tests and developers to see which are test scripts
that are executed by test suite and in which order.
When naming test scripts in format 'T\d\d\d-name.sh' the list of
tests to run are created dynamically. This makes test
'Ensure that all available tests will be run by notmuch-test'
in test/basic obsolete.
Previously the tags on each line in tree view were separarted by ", "
not just " ". This is different from show and search views.
This patch removes this comma. This is a large patch as essentially
every line of each of the expected outputs in the tree tests needs
updating.
Apart from aesthetic reasons this simplifies the switch to
notmuch-tag-format-tags in the next patch.
In some environments (at least Hurd), process-attributes is
unimplimented and always returns nil. This ends up causing test
failures (see e.g. id:87a9ffofsc.fsf@zancas.localnet).
Historically and according to POSIX 1003.1-2001, a signal of 0 can be
used to check the validity of a pid. This seems less heinous than
parsing the output of ps(1).
This fixes a non-deterministic failure in "Ignore files and
directories specified in new.ignore (multiple occurrences)". The test
assumed that all directories would be scanned, even though nothing
updated the mtime of ${MAIL_DIR}. It *usually* worked nevertheless
because the tests run quickly enough that the directory mtime is
usually the same as the current time, so notmuch new does not update
the mtime in the database (because more changes could occur in the
same second). However, when it occasionally did update the mtime in
the database, the notmuch new call in this test would (correctly) skip
"pass 2" of scanning ${MAIL_DIR}, causing it to skip the following
expected lines:
(D) add_files_recursive, pass 2: explicitly ignoring ${MAIL_DIR}/.git
(D) add_files_recursive, pass 2: explicitly ignoring ${MAIL_DIR}/.ignored_hidden_file
(D) add_files_recursive, pass 2: explicitly ignoring ${MAIL_DIR}/ignored_file
This patch fixes this problem by touching ${MAIL_DIR} to ensure it
gets scanned and by rearranging the test to ensure the directories are
touched immediately before the main notmuch new call in the test.
There is an obscure bug in notmuch-hello that very occasionally causes
emacs_deliver_message to fail. Since it it doesn't serve any actual
purpose in the function we delete it, and leave tracking down the the
bug for another day.
Most of the tests previously using emacs_deliver_message do not use
the actual transmitted message, so we replace it with a simpler (and
presumably more reliable function) that only saves (and indexes) an
fcc copy of the message.
When NOTMUCH_TEST_QUIET environment variable is set to non-null value
messages when new test script starts and when test PASSes are disabled.
This eases picking the cases when tests FAIL (as those are still printed).
In preparation for quiet mode print empty line before writing the
test description. This is done now in function designed for it --
it will also be called when test fails.
test-lib.sh sometimes did equivalent of `basename "$0" .sh`, sometimes
skipping the basename part and sometimes .sh part. This worked as
we never had path components in $0 (more than ./) nor .sh ending.
Now the equivalent of `basename "$0" .sh` is done once and used
everywhere. In the future we may have .sh suffix in test names
-- removing those is a good idea.
The choice of decreasing timestamps is a hack which reduces the number
of existing tests which fail. This can be changed to increasing
if/when somebody wants update another 47 tests.
add a new function notmuch_date_sanitize for rfc822-ish things. Add
date sanitization to notmuch_show_sanitize_all and use it more places.
This is all in aid of a transition to unique timestamps on messages.
Eventually we want test messages to have distinct dates to avoid
reproducability problems. This sanitization will prevent some test
failures when that change is made.
Replace the use of a local function in maildir-sync with
notmuch_json_show_sanitize
This was causing test failures because version strings varied in
length between GNU/Linux and GNU/KFreeBSD. One can also imagine
different versions of gnupg causing the same failure.
When executed command line is written to *Notmuch errors* buffer,
shell-quote-argument will backslash-escape any char that is not in
"POSIX filename characters" (i.e. matching "[^-0-9a-zA-Z_./\n]").
Currently in two emacs tests shell has expanded $PWD as part of
emacs variable, which will later be fed to #'shell-quote-argument
and finally written to ERROR file. If $PWD contained non-POSIX
filename characters, data in ERROR file will not match $PWD when
later comparing in shell. Therefore, in these two particular cases
the escaped $PWD is replaced with YYY in ERROR file and expected
content is adjusted accordingly.
This fixes races in thread-local and global tagging in notmuch-search
(e.g., "+", "-", "a", "*", etc.). Previously, these would modify tags
of new messages that arrived after the search. Now they only operate
on the messages that were in the threads when the search was
performed. This prevents surprises like archiving messages that
arrived in a thread after the search results were shown.
This eliminates `notmuch-search-find-thread-id-region(-search)'
because these functions strongly encouraged racy usage.
This fixes the two broken tests added by the previous patch.
These tests check that both thread-local and global search tagging
operations are race-free. They are currently known-broken because
they aren't race-free.
These queries will match exactly the set of messages currently in the
thread, even if more messages later arrive. Two queries are provided:
one for matched messages and one for unmatched messages.
This can be used to fix race conditions with tagging threads from
search results. While tagging based on a thread: query can affect
messages that arrived after the search, tagging based on stable
queries affects only the messages the user was shown in the search UI.
Since we want clients to be able to depend on the presence of these
queries, this ushers in schema version 2.
(Unfortunately, it's difficult to first demonstrate this problem with
a known-broken test because modern Linux kernels have argument length
limits in the megabytes, which makes Emacs really slow!)
It was looking in completely the wrong place for the backup and the
(test) xapian database. Unfortunately test_begin_subtest hides the
relevant errors.
Currently notmuch-show looks at the prefix-arg directly via
current-prefix-arg. This changes it to use the interactive
specification.
One test (for elide-toggle functionality) set the prefix arg
directly. Update this test to set the new argument directly.
One test (reply to encrypted message in the crypto test) recently
started failing on some systems. The failure I saw were two extra
lines of the form
<87d2nbc5xg.fsf@host.i-did-not-set--mail-host-address--so-tickle-me>
The test pipes the output through
grep -v -e '^In-Reply-To:' -e '^References:'
which would normally these two ids but it does not, in this case,
because they are so long they get put on a separate line in the output.
To fix this we set mail-host-address for emacs deliver. example.com
seems a sensible address to use. This is short enough that we don't
get the line breaks above and the tests then all pass.
As explained by Jeffrey Stedfast, the author of GMime, quoted in [1]:
> Passing the GMIME_ENABLE_RFC2047_WORKAROUNDS flag to g_mime_init()
> *should* solve the decoding problem mentioned in the thread. This
> flag should be safe to pass into g_mime_init() without any bad side
> effects and my unit tests do test that code-path.
The thread being referred to is [2].
[1] id:87bo56viyo.fsf@nikula.org
[2] id:08cb1dcd-c5db-4e33-8b09-7730cb3d59a2@gmail.com
Some common broken RFC 2047 encodings that we currently let gmime
parse strictly. We could tell gmime to be forgiving in what it accepts
as RFC 2047 encoding, making these tests pass.
When 'xpg_echo' bash shell option is unset (usually the default)
echo builtin does not expand backslash-escape sequences by default
(i.e. '\n' is echoed as '\n' instead of newline). Not all bash
installations have this feature we depend on activated by default.
Note that the feature is bash (and GNU /bin/echo) specific. It is used
as it is convenient. If portability is needed (elsewhere) use printf(1)
(also often available as a shell builtin).
If any of the tests in our test system is not passing the execution
of the test suite completes with nonzero exit value.
It is better to rely on the exit value of the test system instead
of some arbitrary strings in test output (or use both).
notmuch_message_tags_to_maildir_flags() unconditionally moves messages from
maildir directory "new/" to maildir directory "cur/", which makes messages lose
their "new" status in the MUA. However some users want to keep this "new"
status after, for instance, an auto-tagging of new messages.
However, as Austin mentioned and according to the maildir specification,
messages living in "new/" are not allowed to have flags, even if mutt allows it
to happen. For this reason, this patch prevents moving messages from "new/" to
"cur/", only if no flags have to be changed. It's hopefully enough to satisfy
mutt (and maybe other MUAs showing the "new" status) users checking the "new"
status.
Changelog:
* v2: Fix bool type as well as NULL returned despite having no errors (Austin
Clements)
* v4: Tag the related test (contributed by Michal Sojka) as working
Signed-off-by: Louis Rilling <l.rilling@av7.net>
[Condition for keeping messages in new/ was extended to satisfy all
tests from the previous patch. -Michal Sojka]
[Added by David Bremner, to keep the tests passing at each commit]
update insert tests for new maildir synchronization rules
As of id:1355952747-27350-4-git-send-email-sojkam1@fel.cvut.cz
we are more conservative about moving messages from ./new to ./cur.
This updates the insert tests to match
As mentioned by Jani Nikula in id:87vcccp4y3.fsf@nikula.org, some cases
of maildir synchronization are not covered by our tests. Let's add the
missing tests.
Some MUA's like mutt show the difference between "new" emails living in maildir
directory new/, and "old" emails living in maildir directory cur/. However
notmuch tag unconditionally moves selected messages from new/ to cur/, even if
no maildir synchronized tag is changed.
While maildir specification forbids messages with tags living in new/, there is
no need to move messages to cur/ when no maildir synchronized tag is changed.
Thus notmuch can remain transparent with respect to other MUA's.
[ Edited commit log to better describe the intended changes, and tag the
test as broken until the actual changes are implemented -- Louis Rilling ]
Signed-off-by: Louis Rilling <l.rilling@av7.net>
[ Converted to use test_subtest_known_broken, David Bremner ]
RFC 2047 states that the encoding and charset in an encoded word are
case-insensitive, so force them to lower case in the reply test. This
fixes an issue caused by GMime versions (somewhere between 2.6.10 and
2.6.16), which changed the capitalization of the encoding.
Previously, reply's default text format used an odd mix of RFC 2045
MIME encoding for the reply template's body and some made-up RFC
2822-like UTF-8 format for the headers. The intent was to present the
headers to the user in a nice, un-encoded format, but this assumed
that whatever ultimately sent the email would RFC 2047-encode the
headers, while at the same time the body was already RFC 2045 encoded,
so it assumed that whatever sent the email would *not* re-encode the
body.
This can be fixed by either producing a fully decoded UTF-8 reply
template, or a fully encoded MIME-compliant RFC 2822 message. This
patch does the latter because it is
a) Well-defined by RFC 2822 and MIME (while any UTF-8 format would be
ad hoc).
b) Ready to be piped to sendmail. The point of the text format is to
be minimal, so a user should be able to pop up the template in
whatever editor they want, edit it, and push it to sendmail.
c) Consistent with frontend capabilities. If a frontend has the
smarts to RFC 2047 encode the headers before sending the mail, it
probably has the smarts to RFC 2047 decode them before presenting
the template to a user for editing.
Also, as far as I know, nothing automated consumes the reply text
format, so changing this should not cause serious problems. (And if
anything does still consume this format, it probably gets these
encoding issues wrong anyway.)
Previously, the References header code seemed to assume
notmuch_message_get_header would return NULL if the header was not
present, but it actually returns "". As a result of this, it was
inserting an unnecessary space when concatenating an empty or missing
original references header with the new reference.
This shows up in only two tests because the text reply format later
passes the whole reply template through g_mime_filter_headers, which
has the side effect of stripping out this extra space.
This switches `notmuch-mua-reply' and `notmuch-query-get-threads' to
the S-exp format. These were the last two uses of the JSON format in
the Emacs frontend.
While looked good on paper, its attempted use caused confusion, complexity,
and potential for information leak when passed through wrapper scripts.
For slimmer code and to lessen demand for maintenance/support the set of
commits which added top level --stderr= option is now reverted.
The find option syntax `-perm +111` is deprecated gnu find feature.
The replacement `( -perm -100 -o -perm -10 -o -perm 1 )` should also
work outside of the GNU domain.
In addition to being the Right Thing to do, this noticeably improves
the time taken to display the first page of search results, since it's
roughly an order of magnitude faster than the JSON parser.
Interestingly, it does *not* significantly improve the time to
completely fill a large search buffer because for large search
buffers, the cost of creating author invisibility overlays and
inserting text (which slows down with more overlays) dominates.
However, the time required to display the first page of results is
generally more important to the user experience.
Previously, search started the async notmuch process directly. Now,
it uses `notmuch-start-notmuch'. This simplifies the process sentinel
a bit and means that we no longer have to worry about errors
interleaved with the JSON output.
We also update the tests of Emacs error handling, since the error
output is now separated from the search results buffer.
We now check error handling more carefully in the last test in
test/emacs and we're about to add more error handling tests. (This
was also a strange place for this test, since it had nothing to do
with large search buffers.)
This unifies the part button actions and the underlying part action
functions into single interactive command that simply applies to the
part containing point using the just-added part p-list text property
instead of button properties. Since all part actions can be performed
by applying the appropriate mm function to an mm-handle, this patch
abstracts out the creation of mm handles, making the implementations
of the part commands trivial. This also eliminates our special
handling for part save in favor of using the appropriate mm function.
This necessarily modifies the way we handle the default part button
action, but in a way that does not change the meaning of the
notmuch-show-part-button-default-action defcustom.
Since these commands are no longer specific to buttons, this patch
eliminates the extra metadata stored with each button. This also
eliminates one rather special-purpose macro for a collection of
general purpose part handling utilities.
--stderr=FILE tests were added to test/help-test as it is the one
doing most global option testing. Also, it was simplest to test
this new option using `notmuch help` command.
Presently, the code which finds the parent of a message as it is being
added to the database assumes that the first Message-ID-like substring
of the In-Reply-To header is the parent Message ID. Some mail clients,
however, put stuff other than the Message-ID of the parent in the
In-Reply-To header, such as the email address of the sender of the
parent. This can fool notmuch.
The updated algorithm prefers the last Message ID in the References
header. The References header lists messages oldest-first, so the last
Message ID is the parent (RFC2822, p. 24). The References header is
also less likely to be in a non-standard
syntax (http://cr.yp.to/immhf/thread.html,
http://www.jwz.org/doc/threading.html). In case the References header
is not to be found, fall back to the old behavior.
V2 of this patch, incorporating feedback from Jani and (indirectly)
Austin.
The use of realpath(3) in
commit 58ed67992d
Author: Jani Nikula <jani@nikula.org>
Date: Sun Apr 7 20:15:03 2013 +0300
cli: config: do not overwrite symlinks when saving config file
broke config file save when the file does not exist, which results in
'notmuch setup' always failing to create a new config file.
Fix by checking ENOENT from realpath(3).
Use realpath to canonicalize the config path before writing.
Previously 'notmuch setup' and 'notmuch config set' overwrote the
config file even if it was a symbolic link.
We now have a notmuch_config_is_new() function to query whether a
config was created or not. Change the notmuch_config_open() is_new
parameter into boolean create_new to determine whether the function
should create a new config if one doesn't exist. This reduces the
complexity of the API.
When execution of tests is interrupted by signal coming outside of the
test system itself, output just one line "interrupted by signal <num>"
message to standard output. This distinguishes the case from internal
exit and reduces noise.
Set the variable '$test_subtest_name' in all functions which starts
a new test and use that variable in all functions that output
test results.
Additionally output the latest '$test_subtest_name' in case of
abnormal exit, to avoid confusion.
Instead of checking immediately for the watched process, delay a
minute, or in the case that process-attributes returns nil, for two
minutes. This is intended to cope with the case that
process-attributes is unimplimented, and returns always returns nil.
In this case, the watchdog check is the same as the two minute limit
imposed by timeout.
The TERM environment variable is set to 'dumb' when running tests, but
the original value of it is stored for echoing colors and running emacs
(somewhat interactively) in detached session. Emacs requires some
terminal control sequences to be available for interactive operation.
In case original TERM is (also) 'dumb' (or unset/empty) emacs cannot
run interactively. To fix this problem dtach (and emacs as it's child
process) is run with TERM=vt100 in case original TERM was unset, empty
or 'dumb'. This way there is a chance to run emacs tests with different
user terminals and potentially find problems there.
This test also serves as documentation of the quoting
requirements. The comment lines are so that it exactly matches the man
page. Nothing more embarrassing than having an example in the man page
fail.
The (now fixed) bug that this test revealed is that unquoted
message-ids with whitespace or other control characters in them are
split into several tokens by the Xapian query parser.
This is based on the similar test for notmuch restore, but the parser
in batch tagging mode is less tolerant of a few cases, in particular
those tested by illegal_tag.
We recently switched to popping up a buffer to report CLI errors, but
this was too intrusive, especially for transient errors and especially
since we made fewer things ignore errors. This patch changes this to
display a basic error message in the minibuffer (using Emacs' usual
error handling path) and, if there are additional details, to log
these to a separate error buffer and reference the error buffer from
the minibuffer message. This is more in line with how Emacs typically
handles errors, but makes the details available to the user without
flooding them with the details.
Given this split, we pare down the basic message and make it more
user-friendly, and also make the verbose message even more detailed
(and more debugging-oriented).
This switches the new batch-tag format away from using a home-grown
hex-encoding scheme for message IDs in the dump to simply using Xapian
queries with Xapian quoting syntax.
This has a variety of advantages beyond presenting a cleaner and more
consistent interface. Foremost is that it will dramatically simplify
the quoting for batch tagging, which shares the same input format.
While the hex-encoding is no better or worse for the simple ID queries
used by dump/restore, it becomes onerous for general-purpose queries
used in batch tagging. It also better handles strange cases like
"id:foo and bar", since this is no longer syntactically valid.
When we switch to using regular Xapian queries in the dump format, \n
will cause problems, so we disallow it. Specially, while Xapian can
quote and parse queries containing \n without difficultly, quoted
queries containing \n still span multiple lines, which breaks the
line-orientedness of the dump format. Strictly speaking, we could
still round-trip these, but it would significantly complicate restore
as well as scripts that deal with tag dumps. This complexity would
come at absolutely no benefit: because of the RFC 2822 unfolding
rules, no amount of standards negligence can produce a message with a
message ID containing a line break (not even Outlook can do it!).
Hence, we simply disallow it.
This patch corrects several undesirable behaviours:
1) Empty files were not detected, leading to buffer read overrun.
2) An initial blank line cause restore to silently abort
3) Initial comment line caused format detection to fail
notmuch_json_show_sanitize replaced "filename" field values even in part
structures, where the value is predictable. Make it only normalize the
filename value if it is an absolute path (begins with slash), which is
true of the Maildir filenames that were intended to be normalized away.
This slightly changes the output of an existing test since we now
report non-zero exits with a pop-up buffer instead of at the end of
the search results.
* test/emacs:
- Rename subtests "{Add,Remove} tag from notmuch-show view" to
"notmuch-show: {add,remove} single tag {to,from} single message"
to be consistent with the following tests.
- New subtest "notmuch-show: add multiple tags to single message":
`notmuch-show-add-tag' ("+") can add multiple tags to a message.
- New subtest "notmuch-show: remove multiple tags from single message":
`notmuch-show-remove-tag' ("-") can remove multiple tags from a message.
We want to test both that error/warning messages are generated when
they should be, and not generated when they should not be. This varies
between restore and batch tagging.
These one need the completed functionality in notmuch-restore. Fairly
exotic tags are tested, but no weird message id's.
We test each possible input to autodetection, both explicit (with
--format=auto) and implicit (without --format).
The quoting for ${SEARCH} is broken when it's supposed to be '*', and
it seems tricky to get it right. Just drop the variable and use '*'
directly. Before this, none of the messages ever matched, and the test
was comparing zeros.
test_expect_equal_json uses json.tool from the system Python. While
Python 2 wasn't picky about the encoding of stdin, Python 3 decodes
stdin strictly according to the environment. Since we set LC_ALL=C
for the tests, Python 3's json.tool was assuming stdin would be in
ASCII and aborting when it couldn't decode the UTF-8 characters from
some of the JSON tests. This patch sets the PYTHONIOENCODING
environment variable to utf-8 when invoking json.tool to override
Python's default encoding choice.
Without this change, GCC complains as follows:
gcc test/random-corpus.o test/database-test.o notmuch-config.o command-line-arguments.o lib/libnotmuch.a util/libutil.a parse-time-string/libparse-time-string.a -o test/random-corpus -lgmime-2.6 -lgio-2.0 -lgobject-2.0 -lglib-2.0 -Wl,-rpath,/usr/lib -ltalloc -lxapian
/usr/bin/ld: lib/libnotmuch.a(database.o): undefined reference to symbol '_ZNSs4_Rep10_M_destroyERKSaIcE@@GLIBCXX_3.4'
/usr/bin/ld: note: '_ZNSs4_Rep10_M_destroyERKSaIcE@@GLIBCXX_3.4' is defined in DSO /usr/lib/libstdc++.so.6 so try adding it to the linker command line
/usr/lib/libstdc++.so.6: could not read symbols: Invalid operation
collect2: error: ld returned 1 exit status
make: *** [test/random-corpus] Error 1
We demonstrate the current notmuch restore parser being confused by
message-id's and tags containing non alpha numeric characters
(particularly space and parentheses are problematic because they are
not escaped by notmuch dump).
We save the files as hex escaped on disk so that terminal emulators
will not get confused if the test fails (as we mostly expect it to do).
Initial use case is testing dump and restore, so we only have
message-ids and tags.
The message ID's are nothing like RFC compliant, but it doesn't seem
any harder to roundtrip random UTF-8 strings than RFC-compliant ones.
Tags are UTF-8, even though notmuch is in principle more generous than
that.
updated for id:m2wr04ocro.fsf@guru.guru-group.fi
- talk about Unicode value rather some specific encoding
- call talloc_realloc less times
Initially, provide a way to create "stub" messages in the notmuch
database without corresponding files. This is essentially cut and
paste from lib/database.cc. This is a seperate file since we don't
want to export these symbols from libnotmuch or bloat the library with
non-exported code.
Previously, the test framework generated a variable name for each
external prereq as a poor man's associative array. Unfortunately,
prereqs names may not be legal variable names, leading to
unintelligible bash errors like
test_missing_external_prereq_emacsclient.emacs24_=t: command not found
Using proper associative arrays to track prereqs, in addition to being
much cleaner than generating variable names and using grep to
carefully construct unique string lists, removes restrictions on
prereq names.
Previously, if a test script aborted (e.g., because it passed too few
arguments to a test function), the test driver loop would simply
continue on to the next test script and the final results would
declare that everything passed (except that the test count would look
suspiciously low, but maybe you just misremembered how many tests
there were).
Now, if a test script exits with a non-zero status and did not produce
a final results file, we propagate that failure out of the driver loop
immediately.
To keep this simple, this patch removes the PID from the test-results
file name. This PID was inherited from the git test system and seems
unnecessary, since the file name already includes the name of the test
script and the test-results directory is created anew for each run.
And require that if TEST_EMACS is specified, so is TEST_EMACSCLIENT.
Previously, the test framework always used "emacsclient", even if the
Emacs in use was overridden by TEST_EMACS. This causes problems if
both Emacs 23 and Emacs 24 are installed, the Emacs 23 emacsclient is
the system default, but TEST_EMACS is set to emacs24. Specifically,
with an Emacs 24 server and an Emacs 23 client, emacs tests that run
very quickly may produce no output from emacsclient, causing the test
to fail.
The Emacs server uses a very simple line-oriented protocol in which
the client sends a request to evaluate an expression and the server
sends a request to print the result of evaluation. Prior to Emacs bzr
commit 107565 on March 11th, 2012 (released in Emacs 24.1), if
multiple commands were sent to the emacsclient between when it sent
the evaluation command and when it entered its receive loop, it would
only process the first response command, ignoring the rest of the
received buffer. This wasn't a problem with the Emacs 23 server
because it sent only the command to print the evaluation result.
However, the Emacs 24 server first sends an unprompted command
specifying the PID of the Emacs server, then processes the evaluation
request, then sends the command to print the result. If the
evaluation is fast enough, it can send both of these commands before
emacsclient enters the receive loop. Hence, if an Emacs 24 server is
used with an Emacs 23 emacsclient, it may miss the response printing
command, ultimately causing intermittent notmuch test failures.
Previously, many tests in emacs-subject-to-filename didn't quote the
$output argument to test_expect_equal. As a result, if $output was
empty, test_expect_equal would be passed only one argument and would
abort the entire test script. By quoting the argument, we ensure
test_expect_equal will always receive two arguments.
We now test for user ignore patterns before attempting to determine if
a directory entry is itself a directory. As a result, we no longer
abort for broken symlinks if the user has explicitly ignored them.
This fixes the test added in the previous patch. It also slightly
changes the debug output checked by another test of ignores.
The macro with-current-notmuch-show-message executes command
`notmuch show --format=raw id:...` which just outputs the contents
of the mail file verbatim (into temporary buffer). In case e.g. utf-8
locale is used the temporary buffer has buffer-file-coding-system as
utf-8. In this case Emacs converts the data to multibyte format, guessing
that input is in utf-8.
However, the "raw" (MIME) message may contain octet data in any other
8bit format, and as no (MIME-)content spesific handling to the message
is done at this point, conversion to other formats may lose information.
By setting coding-system-for-read 'no-conversion drops the conversion part
and makes this handle input as notmuch-get-bodypart-internal() does.
This marks the broken test in previous change fixed.
Previously, we would treat multi-message mboxes as one giant email,
which, besides the obvious incorrect indexing, often led to
out-of-memory errors for archival mboxes. Now we explicitly reject
multi-message mboxes. For historical reasons, we retain support for
single-message mboxes, but official deprecate this behavior.
This test is currently broken. Note that its brokenness cascades and
causes the next test to fail as well (because notmuch incorrectly
indexes the mbox file).
There are currently 45 TESTS scripts. 36 of those load
test-lib.sh using '. ./test-lib.sh' and 9 '. test-lib.sh'.
In latter case test-lib.sh is first searched from directories
in PATH (posix) and then from current directory (bash feature).
Changed the 9 files to execute '. ./test-lib.sh'. The test-lib.sh
should never be loaded from directory in PATH.
Previously, this would simply indicate that the grep failed without
any indication of the Emacs output it failed on. Now we take
advantage of the test framework's handling of stdout to display the
incorrect Emacs output if the test fails.
It seems we have never tested the case that restore --accumulate
actually adds tags. I noticed this when I started optimizing and no
tests failed.
The bracketing with "restore --input=dump.expected" are to make sure
we start in a known state, and we leave the database in a known state
for the next test.
The test designed to exercise Emacs' rendering of HTML emails
containing images inadvertently assumed w3m was available under Emacs
23. The real point of this test was to check that Emacs 24's shr
renderer didn't crash when given img tags, so use shr if it's
available, html2text otherwise (which is built in), and do only a
simple sanity check of the result.
This regexp agrees with Xapian query syntax much more closely, though
we specifically disallow various cases that would be confusing in the
context of an email body (e.g., punctuation at the end of an id: link
is not considered part of the id: link because it's probably part of
the surrounding text).
In particular, this handles id: links that are not surrounded by
quotes much better, which stash is much more likely to generate now
that we don't quote id's that don't need to be quoted. It also
handles quoted id: links better.
We update the buttonization test to reflect the new pattern.
This matches the current behavior of the buttonizer, so it passes, but
many of these cases are not what you'd want (and some of them aren't
even valid Xapian queries). The next patch will fix the handling of
these cases and update the test.
Over time, maintaining this very long regex has become irritating,
especially when resolving conflicts.
This patch replaces the call to sed with multiple extra arguments to
find. Since each test binary is now on it's own line, this should
make resolving conflicts easier.
Test the date/time parser module directly, independent of notmuch,
using the parse-time test tool.
Credits to Michal Sojka <sojkam1@fel.cvut.cz> for writing most of the
tests.
Add a smoke testing tool to support testing the date/time parser
module directly and independent of the rest of notmuch.
Credits to Michal Sojka <sojkam1@fel.cvut.cz> for the stdin parsing
idea and consequent massive improvement in testability.
Currently, we only properly escape stashed id queries, but there are
other places where the Emacs UI constructs queries for boolean terms.
Since this escaping function is meant to be used in other places, it
avoids escaping strings that don't need escaping.
This disallows adding empty tags, since nothing but confusion follows
in their wake, and disallows adding tags that begin with "-" because
they are also confusing, the tag "-" is impossible to remove using the
CLI, and because the syntax for removing such tags conflicts with long
argument syntax.
This does not place any restrictions on what tags can be removed, as
that would make it difficult for people who have the misfortune of
already having malformed tags to remove these tags.
Although messages are created in a particular order, it seems that
when they are created on a tmpfs, they do not always come back in the
same order, leading to the same files being ignored but being output
in a different order. This causes the test to fail because the outputs
being compared are the same.
Fix the failures by sorting the output of notmuch --debug and
comparing this to a hand-sorted version of its output.
Signed-off-by: Ethan Glasser-Camp <ethan@betacantrips.com>
The use of --background option (instead of shell '&') ensures that
smtp-dummy is listening its server socket until execution of shell
script can continue, thus the client will always have socket where
to connect.
smtp-dummy outputs smtp_dummy_pid variable in shell assignment format;
eval'ing that output makes that variable available for the shell.
As the smtp-dummy instance is no longer child process of the script
the SIGKILL signal sent to it will ensure it is going away in case
the mail sender fails to connect to smtp-dummy.
When shell executes background process using '&' the scheduling of
that new process is arbitrary. It could be that smtp-dummy doesn't
get execution time to listen() it's server socket until some other
process attempts to connect() to it. The --background option in
smtp-dummy makes it to go background *after* it started to listen
its server socket.
When --background option is used, the line "smtp_dummy_pid='<pid>'"
is printed to stdout from where shell can eval it.
Demonstrates that *every* file/directory which matches one of the values
in 'new.ignore' will be ignored, independent of its depth/location in
the mail store.
Signed-off-by: Ethan Glasser-Camp <ethan@betacantrips.com>
Obviates the need to create a 'NOTMUCH_NEW' clone which runs
'notmuch new --debug'. This will be used in a later patch.
Doesn't cause any issues for other tests.
* test/emacs:
- New subtest "notmuch-show: collapse all messages in thread":
`notmuch-show-open-or-close-all' with prefix arg ("C-u M-RET")
collapses all messages in thread.
- New subtest "notmuch-show: uncollapse all messages in thread":
`notmuch-show-open-or-close-all' without prefix arg ("M-RET")
uncollapses all messages in thread.
* test/emacs:
- New subtest "notmuch-show: show message headers":
Setting `notmuch-message-headers-visible' to t causes all headers
defined in `notmuch-message-headers' to be shown.
- New subtest "notmuch-show: hide message headers":
Setting `notmuch-message-headers-visible' to nil causes all headers
defined in `notmuch-message-headers' to be hidden.
("Subject:" may be an exception; See the use of `headers-start' in
`notmuch-show-insert-msg')
- New subtest "notmuch-show: hide message headers (w/ notmuch-show-toggle-headers)":
Setting `notmuch-message-headers-visible' to t causes all headers
defined in `notmuch-message-headers' to be shown, but they can be
hidden for the current message by running `notmuch-show-toggle-headers'.
This requires changing the contents of the crypto tests, as one thread
that was marked read by the earlier tests in test/emacs is no longer
marked read.
This moves tests for:
- 09d19ac "test: emacs: toggle eliding of non-matching messages in
`notmuch-show'", which should have actually read: "test: emacs:
toggle processing of cryptographic MIME parts in `notmuch-show'".
See commit 19ec74c5.
- 5ea1dbe "test: emacs: toggle eliding of non-matching messages in
`notmuch-show'"
- 345faab "test: emacs: toggle thread content indentation in
`notmuch-show'"
Signed-off-by: Ethan Glasser-Camp <ethan@betacantrips.com>
Since $TEST_DIRECTORY is an absolute path, any filenames generated
with it will be complete paths. Only use the basename to generate
suffixes for filenames.
Signed-off-by: Ethan Glasser-Camp <ethan@betacantrips.com>
Most Emacs tests end with a call to (test-output), which saves the
buffer to a filed called OUTPUT. Previously, if the test code failed
with an exception before this call, the test framework would then
compare against the OUTPUT file from the last Emacs test, resulting in
confusing diffs.
This requires one tweak to an emacs test that made two calls to
test_emacs and expected an OUTPUT file from the first call. We simply
reverse the order of the test_emacs calls.
On FreeBSD, and probably anywhere else someone installed xapian to
some other prefix, we need to use XAPIAN_LDFLAGS to make the linker can
actually find libxapian.
Before the change, test_expect_equal_file() function treated the first
argument as "actual output file" and the second argument as "expected
output file". When the test fails, the files are copied for later
inspection. The first files was copied to "$testname.output" and the
second file to "$testname.expected". The argument order for
test_expect_equal_file() is often wrong which results in confusing
diff output and incorrectly named files.
The patch solves the issue by changing test_expect_equal_file() to
treat arguments just as two files, without any special properties
(like "actual" and "expected"). The file names for copying is now
based on the given file name: "$testname.$file1" and
"$testname.$file2". E.g. if test_expect_equal_file() is called with
"OUTPUT" and "EXPECTED", the copied files can be named
"emacs.1.OUTPUT" and "emacs.1.EXPECTED".
The down side of this approach is that diff argument order depends on
test_expect_equal_file() argument order. So sometimes we get diff
from expected to actual results, and sometimes the other way around.
But the files are always named correctly.
The behaviour of "emacsclient --eval nil" changed from emacs23 to
emacs24, and in emacs24 it prints 'nil' rather than an empty string.
(format "%S" foo) produces a sexpr form of foo, and is consistent
between the two versions.
The version of message.el in emacs24 omits the charset=us-ascii,
causing the current version of this test to fail. With this patch, we
accept either option. According to RFC 2046, they are semantically
equivalent.
When running emacs tests using emacs 23.1.1 the tests block (until timeout)
when emacs function (notmuch-test-wait) is called.
There is an emacs bug #2930 titled:
23.0.92; `accept-process-output' and `sleep-for' do not run sentinel
It seems this is present in emacs 23.1.
Calling list-processes after accept-process-output seems work around
this problem; in case Emacs version is 23.1 a defadvice is activated
to do just that.
notmuch-test-wait called sleep-for in a loop to wait unconditionally 0.1
seconds while waiting for process to exit.
accept-process-output returns as soon as there is any data available
from process, so using it avoids unnecessary fixed delays.
Both of these functions run process sentinels.
The string function in a sprinter may be called with a NULL string
pointer (eg if a header is absent). This causes a segfault. We fix
this by checking for a null pointer in the string functions and update
the sprinter documentation.
At the moment some output when format=text is done directly rather than
via an sprinter: in that case a null pointer is passed to printf or
similar and a "(null)" appears in the output. That behaviour is not
changed in this patch.
The syntax --output=filename is a smaller change than deleting the
output argument completely, and conceivably useful e.g. when running
notmuch under a debugger.
Format canonicalization of JSON output is no longer necessary, so
remove it. Value canonicalization (e.g., normalizing thread IDs) is
still necessary, so all of the sanitization functions remain.
Previously, we used a variety of ad-hoc canonicalizations for JSON
output in the test suite, but were ultimately very sensitive to JSON
irrelevancies such as whitespace. This introduces a new test
comparison function, test_expect_equal_json, that first pretty-prints
*both* the actual and expected JSON and the compares the result.
The current implementation of this simply uses Python's json.tool to
perform pretty-printing (with a fallback to the identity function if
parsing fails). However, since the interface it introduces is
semantically high-level, we could swap in other mechanisms in the
future, such as another pretty-printer or something that does not
re-order object keys (if we decide that we care about that).
In general, this patch does not remove the existing ad-hoc
canonicalization because it does no harm. We do have to remove the
newline-after-comma rule from notmuch_json_show_sanitize and
filter_show_json because it results in invalid JSON that cannot be
pretty-printed.
Most of this patch simply replaces test_expect_equal and
test_expect_equal_file with test_expect_equal_json. It changes the
expected JSON in a few places where sanitizers had placed newlines
after commas inside strings.
These extra directories cause problems for building on Debian
twice in a row.
In order to remove directories, we need to us "rm -rf" instead of
"rm -f". So now we should be extra careful what we add to the
variable CLEAN.
This patch switches from the current ad-hoc printer to the structured
formatters in sprinter.h, sprinter-text.c and sprinter-json.c.
The JSON tests are changed slightly in order to make them PASS for the
new structured output formatter.
The text tests pass without adaptation.
The JSON format eliminates the complex escaping issues that have
plagued the text search format. This uses the incremental JSON parser
so that, like the text parser, it can output search results
incrementally.
This slows down the parser by about ~4X, but puts us in a good
position to optimize either by improving the JSON parser (evidence
suggests this can reduce the overhead to ~40% over the text format) or
by switching to S-expressions (evidence suggests this will more than
double performance over the text parser). [1]
This also fixes the incremental search parsing test.
This has one minor side-effect on search result formatting.
Previously, the date field was always padded to a fixed width of 12
characters because of how the text parser's regexp was written. The
JSON format doesn't do this. We could pad it out in Emacs before
formatting it, but, since all of the other fields are variable width,
we instead fix notmuch-search-result-format to take the variable-width
field and pad it out. For users who have customized this variable,
we'll mention in the NEWS how to fix this slight format change.
[1] id:"20110720205007.GB21316@mit.edu"
This advises the search process filter to make it process one
character at a time in order to test the pessimal case for incremental
search output parsing.
The text parser fails this test because it gets tricked into thinking
a parenthetical remark in a subject is the tag list.
There didn't seem to be these basic tests for --format=text,
as there are for --format=json. These are just the tests from
the `json' script, with adjusted expected outputs.
