Test for bug. Current stemming support for notmuch adds extra terms
to the DB which aren't removed when the file renames are detected.
When folder tags are added to a message, Xapian terms for both XFOLDER
and ZXFOLDER are generated. When one of the filenames are
renamed/removed, only the XFOLDER tags are removed, leaving it possible
for a match on a folder: tag that was previously but is no longer a
match in the maildir.
(cherry picked from commit 86e0baeb6d)
Messages in the database can have multiple files associated with a
single message-id, but until now only one filename for each message
has been reported by "notmuch search --output=files"
Signed-off-by: Mark Anderson <ma.skies@gmail.com>
(cherry picked from commit d752509abf)
Carl reports "gcc -aux-info notmuch.aux lib/notmuch.h" does not
generate notmuch.aux for him with Debian gcc 4.6.0-8. A small
modification of the original sed regular expression allows us to work
directly from lib/notmuch.h, rather than preprocessing with gcc.
As with most such simple regex based "parsing", this is quite
sensitive to the input format, and needs that each symbol to be
exported from libnotmuch should
- start with "notmuch_"
- be the first non-whitespace token on the line
- be followed by an open parenthesis.
(Cherry-picked from 51b7ab6968, with conflicts resolved by db)
This was inadvertently left over when debugging an early version of
this commit. -Carl Worth <cworth@cworth.org>
(cherry picked from commit 8bf0c1c3de)
This error occurs when `notmuch-fcc-dirs' is set to a list. The error
was in the `notmuch-fcc-dirs' format check which was changed in an
incompatible way from 0.4 to 0.5.
The fix was extracted from a bigger patch series by David
Edmondson id:"1290682750-30283-2-git-send-email-dme@dme.org".
Signed-off-by: Jameson Graef Rollins <jrollins@finestructure.net>
(cherry picked from commit ce08571428)
We exercise each of the documented values (nil, a string, and a
list). For the list, we test matching a specific entry, matching a
catch-all regular expression, and no match at all (in which case there
is no FCC set).
(cherry picked from commit 76b54f1898)
This uses dh_python2 (included with sufficiently recent versions of
the python/python-all packages). python-all brings in all of the
supported versions of python. The double calls to dh_auto_install and
friends are to avoid looping over python versions ourselves.
The worry here is that a binary linking with libnotmuch might lose
access to Xapian::Error symbols because libnotmuch hides them.
We are careful here to create ./fakedb/.notmuch in order to trigger a
Xapian exception, and not just a missing file check.
Thanks to jrollins and amddragon for suggestions.
(cherry picked from commit 66f37f5f6864a988f94ddb893e3a176af57f6c8e)
This is closely tied to gcc and particularly gnu ld, but I guess the
shared library linking code would need to be adjusted to work on a
non-gnu linker anyay.
I had to make a few not-obviously related changes to the
lib/Makefile.local to make this work: libnotmuch_modules is defined
with := and used in place of $^
(cherry picked from commit 014bf85b1c06ff49be2bde5a26433d2cf376cf70)
One of the Xapian exception symbols has gone missing. Hopefully this
doesn't mean we should be bumping the SONAME.
(cherry picked from commit e8260d63f1f5a37094801ecd5d110bf48f22ee45)
Probably there is a nicer way to do this; perhaps it should be a
configure option.
(cherry picked from commit 387dc520dd68cc805e390f3a1399f85b5d5bd83a)
After some debate, we settled on 0.6~$num as a snapshot version
number, at least until 0.6 is released. This has the needed property
of sorting after 0.5+nmu$num.
(cherry picked from commit 54f7766febc3fa3c9dc475f6ec6718c5e87a189e)
This is patch is a temporary work-around for a slight regression that
popped up in the part handling reorganization. Currently, text/plain
parts are always preferred, if present, over other non-text/plain
parts in multipart/alternative. However, this means that if there is
a blank text/plain part, no content will be displayed.
One way to get around this is to set the
"notmuch-show-all-multipart/alternative-parts" customization variable
to True ('t'), which will cause all parts to always be displayed.
Since we want to move forward with the next release, we're going to
set this variable true by default, to make sure that no content is
unretrievably hidden from the user. Once we come up with a better
solution for easy display of hidden parts we can set this back to a
default value of 'nil'.
The main() function should be written as just another function with a
return value. This allows for more reliable code reuse. Imagine that
main() grows too large and needs to be factored into multiple
functions. At that point, exit() is probably the wrong thing, yet can
also be hard to notice as it's in less-frequently-tested exceptional
cases.
This avoids patching the version file once per Debian upload.
Original version file is saved and restored.
(cherry picked from commit 2938a98bf4c4abe0426caee4555d889d655bc0df)
Each top level test (basic, corpus, etc...) is run with a fixed
timeout of 2 minutes.
The goal here is to treat a hung test as a failure. The emacs test for
sending mail is known to be problematic on the Debian
autobuilders. This is both a bandaid fix for that, and a sensible long
term feature.
(cherry picked from commit 5f99c80e02736c90495558d9b88008a768876b29)
Importing notmuch loads the notmuch shared library. When building
without a system install of notmuch, this requires e.g. setting
LD_LIBRARY_PATH for building and fails completely for cleaning.
We were still using len(self.get_replies()) for the __str__ summary of a
mail, but 1) len(Messages()) has just gone away 2) the number of replies
can not be retrieved when we got the message via search_messages()
anyway, and 3) it is likely quite expensive to pull all replies for all
messages that we display a summary of.
So we fix this by simplifying str(Message()) to omit the number of replies.
Signed-off-by: Sebastian Spaeth <Sebastian@SSpaeth.de>
libnotmuch (and python) crashed when I accidently passed in an invalid
value as path argument to the Database() instantiation.
Therefore, we now check via assert that the handed in path is actually a
real string (or None).
Signed-off-by: Sebastian Spaeth <Sebastian@SSpaeth.de>
and also maildir_flags_to_tags. The methods will be invoked by
db.add_message() and also (if not overridden via function parameter) by
add|remove_tag and remove_all_tags. Documentation on the usage has been
updated.
Signed-off-by: Sebastian Spaeth <Sebastian@SSpaeth.de>
Improve the documentation with regard to the new __cmp__ and __hash__
methods and the implications of doing set arithmetic with Messages()
objects.
Signed-off-by: Sebastian Spaeth <Sebastian@SSpaeth.de>
Before the change, the last loop in guess_from_received_header()
did not reset domain and tld variables to NULL. This leads to
double free error in some cases and possibly other bugs.
Before the change, headers and message visibility functions took
extra care to correctly set `buffer-invisibility-spec'. This was
needed because headers overlay `invisible' property had only
headers' invisibility spec. So visibility of headers was
determined only by the headers invisibility spec. The patch sets
headers overlay `invisible' property a list with both the headers
and the message invisibility spec. This makes headers invisible
if either of them is added to the `buffer-invisibility-spec' and
allows to simplify the code.
Before the patch, message, headers and hidden citation overlays
had zero priority. All these overlay have `invisible' property.
Emacs documentation says that we should not make assumptions
about which overlay will prevail when they have the same priority
[1]. It happens to work as we need, but we should not rely on
undocumented behavior.
[1] http://www.gnu.org/s/emacs/manual/html_node/elisp/Overlay-Properties.html
Before the change, message and citation invisibility overlays
conflicted: if some citation is made visible and then the whole
message is hidden, that citation remained visible. This happened
because the citation's overlay has an invisible property which
takes priority over the message overlay. The message
invisibility spec does not affect citation visibility, it is
determined solely by the citation overlay invisibility spec.
Hence, if citation is made visible, it is not hidden by message
invisibility spec.
The patch changes citation overlay invisibility property to be a
list which contains both the citation and the message
invisibility specs. This makes the citation invisible if either
of them is added to the `buffer-invisibility-spec'. Note that
all citation visibility states are "restored" when the message
hidden and shown again.
Before the change, the `notmuch-show-insert-text/plain-hook' was
given only the `depth' argument. The patch adds another one -
the message. Currently, the new message argument is not used by
any on the hooks. But it will be used later to get access to
message invisibility specs when wash buttons are inserted.
The emacs bug is that isearch cannot search through invisible text
when the 'invisible' property is a list.
The patch adds `notmuch-isearch-range-invisible' function which
is the same as `isearch-range-invisible' but with fixed Emacs bug
#8721. Advice added for `isearch-range-invisible' which calls
`notmuch-isearch-range-invisible' instead of the original
`isearch-range-invisible' when in `notmuch-show-mode'.
We can now do: if msg1 == msg2, and we can use set arithmetic on
Messages():
s1, s2= msgs1, msgs2
s1.union(s2)
s2 -= s1
Signed-off-by: Sebastian Spaeth <Sebastian@SSpaeth.de>
