notmuch.git
10 years agodoc: quiet doxygen warnings
David Bremner [Sun, 6 Jul 2014 14:23:42 +0000 (11:23 -0300)]
doc: quiet doxygen warnings

remove some obsolete tags for XML output (which we don't currently
generate in any case)

10 years agodoc: build and install doxygen api docs
David Bremner [Tue, 1 Jul 2014 16:40:33 +0000 (13:40 -0300)]
doc: build and install doxygen api docs

In order to support out of tree builds and avoid hardcoding version
number, generate `doc/config.dox` from configure.

10 years agoconfigure: use cc/c++ instead of gcc/g++
Fraser Tweedale [Thu, 22 May 2014 10:10:06 +0000 (20:10 +1000)]
configure: use cc/c++ instead of gcc/g++

Some systems (e.g. FreeBSD 10) do not ship with the GNU Compiler
Collection.  Use generic cc/c++ instead of gcc/g++ (unless the
CC/CXX environment variables are used).

10 years agoMerge tag '0.18.1'
David Bremner [Wed, 25 Jun 2014 11:01:17 +0000 (08:01 -0300)]
Merge tag '0.18.1'

notmuch 0.18.1 release

10 years agoversion: bump to 0.18.1 0.18.1 debian/0.18.1-1
David Bremner [Wed, 25 Jun 2014 10:24:57 +0000 (07:24 -0300)]
version: bump to 0.18.1

Also add precis of NEWS to debian changelog

10 years agoNEWS: Improve and correct "Fix for phrase indexing" entry
Austin Clements [Mon, 23 Jun 2014 14:25:28 +0000 (10:25 -0400)]
NEWS: Improve and correct "Fix for phrase indexing" entry

This improves the description of the fix, fixes some typos, and
changes "(re)-indexed" to "indexed" because we have no particular
notion of "re-indexing" a message.

10 years agoNEWS: quibbles from Tomi
David Bremner [Mon, 23 Jun 2014 09:36:30 +0000 (06:36 -0300)]
NEWS: quibbles from Tomi

See id:m2vbrsgi7n.fsf@guru.guru-group.fi for details

10 years agoMerge tag '0.18.1_rc0'
David Bremner [Mon, 23 Jun 2014 01:25:10 +0000 (22:25 -0300)]
Merge tag '0.18.1_rc0'

notmuch 0.18.1~rc0 release

10 years agoversion: bump to 0.18.1~rc0
David Bremner [Mon, 23 Jun 2014 00:23:22 +0000 (21:23 -0300)]
version: bump to 0.18.1~rc0

Also bump the python bindings version, the NEWS version and the Debian
version.

Since the changelog is (slightly dubiously) metadata, we have to
change it to upload a release candidate.

10 years agoNEWS: add news for 0.18.1
David Bremner [Sat, 21 Jun 2014 20:09:01 +0000 (17:09 -0300)]
NEWS: add news for 0.18.1

This is my summary of the commits from 0.18.

10 years agoemacs: Forwarded messages should not have modified buffers
David Edmondson [Mon, 16 Jun 2014 05:30:11 +0000 (06:30 +0100)]
emacs: Forwarded messages should not have modified buffers

When the user begins forwarding a message, the resulting composition
buffer should not be marked as modified, in order that it can
immediately be killed without prompting.

10 years agoMerge branch 'release'
David Bremner [Sun, 22 Jun 2014 09:53:21 +0000 (06:53 -0300)]
Merge branch 'release'

Austin's termpos patches and Felipe's zlib.pc workaround

10 years agoconfigure: add workaround for systems without zlib.pc
Felipe Contreras [Mon, 12 May 2014 03:09:32 +0000 (22:09 -0500)]
configure: add workaround for systems without zlib.pc

Some systems (e.g. FreeBSD) might not have installed the appropriate
pkg-config file as they should. We can workaround the issue by creating
the .pc file they should have distributed.

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
10 years agolib: Separate all phrases indexed by _notmuch_message_gen_terms
Austin Clements [Mon, 16 Jun 2014 02:40:34 +0000 (22:40 -0400)]
lib: Separate all phrases indexed by _notmuch_message_gen_terms

This adds a 100 termpos gap between all phrases indexed by
_notmuch_message_gen_terms.  This fixes a bug where terms from the end
of one header and the beginning of another header could match together
in a single phrase and a separate bug where term positions of
un-prefixed terms overlapped.

This fix only affects newly indexed messages.  Messages that are
already indexed won't benefit from this fix without re-indexing, but
the fix won't make things any worse for existing messages.

10 years agotest: Known-broken test for overlapping/adjacent termpos
Austin Clements [Mon, 16 Jun 2014 02:40:33 +0000 (22:40 -0400)]
test: Known-broken test for overlapping/adjacent termpos

This adds two known-broken tests and one working test related to the
term positions assigned to terms from different headers or MIME parts.
The first test fails because we don't create a termpos gap between
different headers.  The second test fails because we don't adjust
termpos at all when indexing multiple parts.

10 years agolib: Index name and address of from/to headers as a phrase
Austin Clements [Mon, 16 Jun 2014 02:40:32 +0000 (22:40 -0400)]
lib: Index name and address of from/to headers as a phrase

Previously, we indexed the name and address parts of from/to headers
with two calls to _notmuch_message_gen_terms.  In general, this
indicates that these parts are separate phrases.  However, because of
an implementation quirk, the two calls to _notmuch_message_gen_terms
generated adjacent term positions for the prefixed terms, which
happens to be the right thing to do in this case, but the wrong thing
to do for all other calls.  Furthermore, _notmuch_message_gen_terms
produced potentially overlapping term positions for the un-prefixed
copies of the terms, which is simply wrong.

This change indexes both the name and address in a single call to
_notmuch_message_gen_terms, indicating that they should be part of a
single phrase.  This masks the problem with the un-prefixed terms
(fixing the two known-broken tests) and puts us in a position to fix
the unintentionally phrases generated by other calls to
_notmuch_message_gen_terms.

10 years agotest: Add search tests for combined name/address queries
Austin Clements [Mon, 16 Jun 2014 02:40:31 +0000 (22:40 -0400)]
test: Add search tests for combined name/address queries

Two of these are currently known-broken.  We index the name and
address parts in two separate calls to _notmuch_message_gen_terms.
Currently this has the effect of placing the term positions of the
prefixed terms from the second call right after those of the first
call, but screws up the term positions of the non-prefixed terms.

10 years agotest: Fix from/to search test queries
Austin Clements [Mon, 16 Jun 2014 02:40:30 +0000 (22:40 -0400)]
test: Fix from/to search test queries

Two of the search tests for "from" and "to" queries were clearly
trying to search for prefixed phrases, but forgot to shell quote the
phrases.  Fix this by quoting them correctly.

10 years agoMerge branch 'release'
David Bremner [Sun, 15 Jun 2014 18:52:18 +0000 (15:52 -0300)]
Merge branch 'release'

still painfully slowly assembling 0.18.1

10 years agoemacs install: make sure all components to be installed are there
Tomi Ollila [Sat, 10 May 2014 21:36:43 +0000 (00:36 +0300)]
emacs install: make sure all components to be installed are there

`make install-emacs` will copy $(emacs_sources), $(emacs_images) and
$(emacs_bytecode) to their target directories. $(emacs_bytecode) was
already a prerequisite of make install-emacs as these obviously needed
to be build. Until a while ago all of $(emacs_sources) was available
in the repository, but now it includes `notmuch-version.el` which
is generated. In the future we may have generated emacs images too.

10 years agodebian: update notmuch-emacs for emacsen-common 2.0.8
David Bremner [Sat, 14 Jun 2014 10:50:52 +0000 (07:50 -0300)]
debian: update notmuch-emacs for emacsen-common 2.0.8

- redo install/remove scripts from new samples
- symlink .el files properly
- add depends instead of conflicts

10 years agodebian: tighten dependency of python packages on libnotmuch.
David Bremner [Sat, 14 Jun 2014 10:20:56 +0000 (07:20 -0300)]
debian: tighten dependency of python packages on libnotmuch.

Using (>= ${source:Version}) allows newer versions of the library with
the same SONAME.

10 years agoMerge branch 'release'
David Bremner [Sat, 14 Jun 2014 02:03:49 +0000 (23:03 -0300)]
Merge branch 'release'

Merge in a few more commits towards 0.18.1

10 years agolib: resurrect support for single-message mbox files
Jani Nikula [Thu, 5 Jun 2014 06:34:09 +0000 (08:34 +0200)]
lib: resurrect support for single-message mbox files

This is effectively a revert of

commit 6812136bf576d894591606d9e10096719054d1f9
Author: Jani Nikula <jani@nikula.org>
Date:   Mon Mar 31 00:21:48 2014 +0300

    lib: drop support for single-message mbox files

The intention was to drop support for indexing new single-message mbox
files (and whether that was a good idea in the first place is
arguable). However this inadvertently broke support for reading
headers from previously indexed single-message mbox files, which is
far worse.

Distinguishing between the two cases would require more code than
simply bringing back support for single-message mbox files.

10 years agotest: use --quick when starting emacs.
David Bremner [Thu, 29 May 2014 00:26:27 +0000 (21:26 -0300)]
test: use --quick when starting emacs.

At least in emacs24, this removes the "site-lisp" directories from the
load path in addition to enforcing --no-site-lisp --no-init-file.

This works around a slightly mysterious bug on Debian that causes
test-lib.el not to load when there is cl-lib.el(c) in some site-lisp
directory.  It should be harmless in general since we really don't
want to load any files from addon packages to emacs.

10 years agobuild: add dataclean
David Bremner [Sun, 4 May 2014 06:10:49 +0000 (15:10 +0900)]
build: add dataclean

It turns out to be inconvenient to delete the downloaded datafiles with
distclean, so I propose a new target which does that instead.

The closest conventional target is 'maintainer-clean'; the difference
here is that having the original source tarball is not enough to
reconstruct these files.

10 years agobuild: remove .tar.gz.tmp files in clean
David Bremner [Sun, 4 May 2014 06:10:48 +0000 (15:10 +0900)]
build: remove .tar.gz.tmp files in clean

Most people won't see these, but they annoy anyone running release
or pre-release targets.

10 years agobuild: don't overwrite DISTCLEAN in performance-test
David Bremner [Sun, 4 May 2014 06:10:47 +0000 (15:10 +0900)]
build: don't overwrite DISTCLEAN in performance-test

This might have worked once for some fixed order of including
makefiles, but it is clearly wrong now.

10 years agoMerge branch 'release'
David Bremner [Wed, 28 May 2014 12:50:32 +0000 (09:50 -0300)]
Merge branch 'release'

A point release is slowly being built on branch release. Merge those
changes into master.

10 years agobuild: fix order of rpath
Felipe Contreras [Mon, 12 May 2014 02:22:25 +0000 (21:22 -0500)]
build: fix order of rpath

In my system `pkg-config --libs talloc` returns
'Wl,-rpath,/usr/lib -ltalloc' (probably wrongly) which causes the final
LDFLAGS to be something like '-Wl,-rpath,/usr/lib
-Wl,-rpath,/opt/notmuch/lib', which causes the RUNPATH to be
'/usr/lib:/opt/notmuch/lib', so basically defeating the whole purpose of
RUNPATH.

I noticed this when my /opt/notmuch/bin/notmuch (0.17) started updating
the database after I updated the system (which updated the system's
notmuch). This shouldn't happen.

Let's move the RUNPATH flags before other external flags have a chance of
screwing the build.

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
10 years agoUnset html_static_path in Python bindings docs
Amadeusz Żołnowski [Mon, 12 May 2014 10:35:41 +0000 (12:35 +0200)]
Unset html_static_path in Python bindings docs

html_static_path is a kind of source directory and it was set to
destination directory (../html) which caused infinite recursion with
Sphinx 1.2 and above.

10 years agodoc/conf.py: Remove _static from html_static_path
W. Trevor King [Sat, 10 May 2014 15:55:21 +0000 (08:55 -0700)]
doc/conf.py: Remove _static from html_static_path

Avoid:

  $ make HAVE_SPHINX=1 sphinx-html
  python ./doc/mkdocdeps.py ./doc doc/_build doc/docdeps.mk
  sphinx-build -b html -d doc/_build/doctrees -q ./doc doc/_build/html
  Making output directory...
  WARNING: html_static_path entry '/home/wking/src/notmuch/notmuch/doc/_static' does not exist

because we have no static source in doc/_static.

10 years agoemacs: make sure tagging on an empty query is harmless
Mark Walters [Wed, 21 May 2014 09:58:50 +0000 (10:58 +0100)]
emacs: make sure tagging on an empty query is harmless

Currently notmuch-tag throws a "wrong-type-argument stringp nil" if
passed a nil query-string. Catch this and provide a more useful error
message. This fixes a case in notmuch-tree (if you try to tag when at
the end of the buffer).

Secondly, as pointed out by David (dme)
`notmuch-search-find-stable-query-region' can return the query string
() if there are no messages in the region. This gets passed to notmuch
tag, and due to interactions in the optimize_query code in
notmuch-tag.c becomes, in the case tag-change is -inbox, "( () ) and
(tag:inbox)". This query matches some strange collection of messages
which then get archived. This should probably be fixed, but in any
case make `notmuch-search-find-stable-query-region' return a nil
query-string in this case.

This avoids data-loss (random tag removal) in this case.

10 years agoemacs: Correct the documentation for `notmuch-search-line-faces'.
David Edmondson [Tue, 6 May 2014 10:03:30 +0000 (11:03 +0100)]
emacs: Correct the documentation for `notmuch-search-line-faces'.

The implementation and documentation for `notmuch-search-line-faces'
disagreed in how elements in the list were merged. Correct the
documentation to match the implementation (that is, the earlier
elements in the list have precedence over later elements).

10 years agoruby: Add wrapper for notmuch_query_count_threads
Wael M. Nasreddine [Sat, 10 May 2014 21:40:11 +0000 (14:40 -0700)]
ruby: Add wrapper for notmuch_query_count_threads

10 years agoMerge branch 'release'
David Bremner [Sat, 17 May 2014 21:37:48 +0000 (06:37 +0900)]
Merge branch 'release'

bugfix release being built up on release.

10 years agotest: allow pending break points in atomicity script.
David Bremner [Tue, 6 May 2014 13:06:27 +0000 (22:06 +0900)]
test: allow pending break points in atomicity script.

This seems to fix problems with the symbol rename not being defined
at startup on at least OS/X and some Debian Linux architectures.

10 years agotest/Makefile.local: Added configured TALLOC_LDFLAGS.
Charles Celerier [Tue, 6 May 2014 17:02:24 +0000 (13:02 -0400)]
test/Makefile.local: Added configured TALLOC_LDFLAGS.

The linking to talloc is hard-coded in the testing Makefile. This patch
causes the linking to talloc to be done according to how TALLOC_LDFLAGS
was configured.

Signed-off-by: Charles Celerier <cceleri@cs.stanford.edu>
10 years agonotmuch-dump: use fsync instead of fdatasync
David Bremner [Thu, 8 May 2014 11:57:07 +0000 (20:57 +0900)]
notmuch-dump: use fsync instead of fdatasync

Since the file size will have changed, there is no performance benefit
to calling fdatasync.  Somewhat surprisingly, using fdatasync
apparently causes portability problems on FreeBSD.

10 years agoMerge in a few Debian commits.
David Bremner [Thu, 8 May 2014 21:38:42 +0000 (06:38 +0900)]
Merge in a few Debian commits.

10 years agodebian: disable atomicity tests on armel.
David Bremner [Thu, 8 May 2014 05:28:13 +0000 (14:28 +0900)]
debian: disable atomicity tests on armel.

Another temporary fix for build problems.

10 years agodebian: disable atomicity tests on armhf
David Bremner [Wed, 7 May 2014 23:29:15 +0000 (08:29 +0900)]
debian: disable atomicity tests on armhf

In fact a fix for these tests is currently being reviewed, but I want
to roll it together with a few other portability fixes for a point
release.

10 years agoMerge tag '0.18'
David Bremner [Tue, 6 May 2014 08:14:26 +0000 (17:14 +0900)]
Merge tag '0.18'

notmuch 0.18 release

10 years agodebian: changelog stanza for 0.18 0.18
David Bremner [Tue, 6 May 2014 07:27:29 +0000 (16:27 +0900)]
debian: changelog stanza for 0.18

plagiariaze my own summary from upstream NEWS

10 years agoversion: bump to 0.18
David Bremner [Tue, 6 May 2014 07:24:17 +0000 (16:24 +0900)]
version: bump to 0.18

debian changelog to be done seperately.

10 years agoemacs: allow functions in notmuch-show-stash-mlarchive-link-alist
Jani Nikula [Thu, 1 May 2014 12:35:11 +0000 (15:35 +0300)]
emacs: allow functions in notmuch-show-stash-mlarchive-link-alist

Some archives may use a more complicated scheme for referring to
messages than just concatenated url and message-id. In particular,
patchwork requires a query to translate message-id to a patchwork
patch id. Allow functions in notmuch-show-stash-mlarchive-link-alist
to facilitate this.

For example, one could use something like this for patchwork.

(lambda (message-id)
  (concat
   "http://patchwork.example.com/patch/"
   (nth 0
(split-string
 (car (last (process-lines "pwclient" "search" "-n" "1"
   "-m" (concat "<" message-id ">"))))))))

10 years agoMerge tag '0.18_rc1'
David Bremner [Sat, 3 May 2014 23:32:05 +0000 (08:32 +0900)]
Merge tag '0.18_rc1'

notmuch 0.18~rc1 release

10 years agobuild: use UPSTREAM_TAG and not VERSION to generate tarball 0.18_rc1 debian/0.18_rc1-1
David Bremner [Sat, 3 May 2014 23:11:35 +0000 (08:11 +0900)]
build: use UPSTREAM_TAG and not VERSION to generate tarball

The latter can have "~" in it, which is not legal for a git tag.

10 years agodoc/doxygen.cfg: bump version to 0.18
David Bremner [Sat, 3 May 2014 22:38:21 +0000 (07:38 +0900)]
doc/doxygen.cfg: bump version to 0.18

10 years agoversion: bump for 0.18~rc1
David Bremner [Sat, 3 May 2014 22:32:43 +0000 (07:32 +0900)]
version: bump for 0.18~rc1

Doing all of the needed version bumps in one commit, and do a
complete, if minimal debian changelog entry

10 years agovim: improve the way messages are sent
Felipe Contreras [Thu, 1 May 2014 22:57:54 +0000 (17:57 -0500)]
vim: improve the way messages are sent

We want the proper encoding and content-type to be set when sending the
mail, but human-readable plain-text for composing. So split the code in
two parts: the presentation and the transport conversion.

This fixes an issue while sending non-ascii mails to strict servers; the
mail needs to be encoded.

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
10 years agovim: make the html handler configurable
Paul Roberts [Thu, 1 May 2014 22:57:53 +0000 (17:57 -0500)]
vim: make the html handler configurable

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
10 years agovim: fix count_threads variable check
Felipe Contreras [Thu, 1 May 2014 22:57:52 +0000 (17:57 -0500)]
vim: fix count_threads variable check

It never really worked; in Ruby only 'nil' and 'false' evaluate to
false, therefore the statement '0 : true ? false' returns true, so it
doesn't matter if notmuch_folders_count_threads = 0, count_threads would
be true.

We need to check specifically if the value is 1 or 0.

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
10 years agodoc: add *.pyc to doc/.gitignore
Jani Nikula [Fri, 2 May 2014 14:51:01 +0000 (17:51 +0300)]
doc: add *.pyc to doc/.gitignore

10 years agotest: add have-man and have-compact in test/.gitignore
Jani Nikula [Fri, 2 May 2014 14:51:00 +0000 (17:51 +0300)]
test: add have-man and have-compact in test/.gitignore

Sort the file while at it.

10 years agoNEWS: make it explicit that wildcard matching is no longer supported
Jani Nikula [Fri, 2 May 2014 16:28:38 +0000 (19:28 +0300)]
NEWS: make it explicit that wildcard matching is no longer supported

Wildcard matching was a feature of the probabilistic prefix, and we no
longer have it for the boolean prefix. Also note that top-level folder
can now be searched.

10 years agoNEWS: insert and new refuse invalid tags
Jani Nikula [Thu, 1 May 2014 14:42:02 +0000 (17:42 +0300)]
NEWS: insert and new refuse invalid tags

10 years agoNEWS: message piping working directory
Jani Nikula [Thu, 1 May 2014 14:42:01 +0000 (17:42 +0300)]
NEWS: message piping working directory

10 years agoNEWS: emacs: push mark before signature on reply
Jani Nikula [Thu, 1 May 2014 14:41:59 +0000 (17:41 +0300)]
NEWS: emacs: push mark before signature on reply

10 years agoNEWS: cli exit status codes
Jani Nikula [Thu, 1 May 2014 14:42:00 +0000 (17:42 +0300)]
NEWS: cli exit status codes

10 years agoNEWS: notmuch insert respects maildir.synchronize_flags
Jani Nikula [Thu, 1 May 2014 14:41:58 +0000 (17:41 +0300)]
NEWS: notmuch insert respects maildir.synchronize_flags

10 years agoNEWS: add overview section for 0.18
David Bremner [Wed, 30 Apr 2014 23:13:44 +0000 (08:13 +0900)]
NEWS: add overview section for 0.18

This mentions the things I (subjectively) thought were most likely to
either annoy people or make it worth upgrading.

10 years agoNEWS: message header parser changes
Jani Nikula [Sun, 27 Apr 2014 21:09:21 +0000 (00:09 +0300)]
NEWS: message header parser changes

10 years agoNEWS: mbox files are no longer supported
Jani Nikula [Sun, 27 Apr 2014 20:54:24 +0000 (23:54 +0300)]
NEWS: mbox files are no longer supported

10 years agoNEWS: folder:, path:, and database upgrade
Jani Nikula [Sun, 27 Apr 2014 14:46:17 +0000 (17:46 +0300)]
NEWS: folder:, path:, and database upgrade

Related news together.

10 years agoNEWS: replies to encrypted messages encrypted by default
Jani Nikula [Sun, 27 Apr 2014 16:30:47 +0000 (19:30 +0300)]
NEWS: replies to encrypted messages encrypted by default

10 years agoNEWS: Ido initialization for Emacs 23.[123]
Tomi Ollila [Sat, 26 Apr 2014 09:59:23 +0000 (12:59 +0300)]
NEWS: Ido initialization for Emacs 23.[123]

10 years agobuilding from git: use --abbrev=7 for version string
Tomi Ollila [Tue, 22 Apr 2014 15:19:44 +0000 (18:19 +0300)]
building from git: use --abbrev=7 for version string

Users may have set core.abbrev=n, where n != 7 in their git config
file(s) which would give them different than expected version strings
when building notmuch from git. This fixes the commit hash part of
version string to 7 hexadecimal values.

10 years agoMerge tag '0.18_rc0'
David Bremner [Tue, 22 Apr 2014 05:40:56 +0000 (14:40 +0900)]
Merge tag '0.18_rc0'

notmuch 0.18~rc0 release

10 years agodoc: remove conf.pyc on clean 0.18_rc0 debian/0.18_rc0-1
David Bremner [Tue, 22 Apr 2014 05:24:04 +0000 (14:24 +0900)]
doc: remove conf.pyc on clean

This build artifict messes up the packaging process for (at least)
Debian if not removed on clean.

10 years agopython: bump version
David Bremner [Tue, 22 Apr 2014 00:36:26 +0000 (09:36 +0900)]
python: bump version

10 years agoNEWS: bump version
David Bremner [Tue, 22 Apr 2014 00:34:44 +0000 (09:34 +0900)]
NEWS: bump version

As usual, we'll just edit the version in place for release candidates.

10 years agodebian: NEWS item about database upgrade
David Bremner [Tue, 22 Apr 2014 00:33:16 +0000 (09:33 +0900)]
debian: NEWS item about database upgrade

10 years agodebian: changelog stanza for 0.18~rc0-1
David Bremner [Tue, 22 Apr 2014 00:27:55 +0000 (09:27 +0900)]
debian: changelog stanza for 0.18~rc0-1

- make versions match to pacify release-checks.sh
- close a few more bugs.
- fix one HTMLism

10 years agoversion: bump to 0.18~rc0
David Bremner [Mon, 21 Apr 2014 22:00:29 +0000 (07:00 +0900)]
version: bump to 0.18~rc0

Start the promised feature freeze

10 years agodoc: Simplify and clarify notmuch show --format=sexp description
Austin Clements [Fri, 18 Apr 2014 22:42:08 +0000 (18:42 -0400)]
doc: Simplify and clarify notmuch show --format=sexp description

Previously, this was a verbatim copy of the --format=json text.
Change it to instead reference the JSON text and actually describe how
the S-expression format works.

10 years agodoc: Clarify charset encoding of JSON output
Austin Clements [Fri, 18 Apr 2014 22:42:07 +0000 (18:42 -0400)]
doc: Clarify charset encoding of JSON output

10 years agodoc: Fix minor formatting issues in notmuch-show.rst
Austin Clements [Fri, 18 Apr 2014 22:42:06 +0000 (18:42 -0400)]
doc: Fix minor formatting issues in notmuch-show.rst

There were some extra line breaks and missing periods.

10 years agodoc: Clarify notmuch show --format=raw description
Austin Clements [Fri, 18 Apr 2014 22:42:05 +0000 (18:42 -0400)]
doc: Clarify notmuch show --format=raw description

In addition to being generally more precise, this is explicit that
there is no charset conversion.

10 years agoemacs: hello: bugfix for saved searches defcustom
Mark Walters [Wed, 16 Apr 2014 21:21:53 +0000 (22:21 +0100)]
emacs: hello: bugfix for saved searches defcustom

The recent changes for saved searches introduced a bug when notmuch
was loaded after the saved search was defined. This was caused by a
utility function not being defined when the defcustom was loaded.

Fix this by moving some code around: the defcustom is moved into
notmuch-hello (which is a more natural place anyway), and the utility
functions are moved before the defcustom in notmuch-hello. We are
rather constrained as the defcustom for saved searches is the first
variable in the notmuch-hello customize window; to avoid moving this
customize the defcustom needs to be the first defcustom in
notmuch-hello, and the utility functions come before that.

This patch also renames one of the utility functions from
notmuch--saved-searches-to-plist to
notmuch-hello--saved-searches-to-plist (as it is purely local to
notmuch-hello) and corrects a couple of typo/spelling mistakes pointed
out by Tomi.

10 years agodoc/prerst2man.py: Fix 'os.system' -> 'system' typo
W. Trevor King [Sat, 5 Apr 2014 17:31:08 +0000 (10:31 -0700)]
doc/prerst2man.py: Fix 'os.system' -> 'system' typo

Avoid:

  $ make HAVE_SPHINX=0 HAVE_RST2MAN=1 build-man
  python ./doc/prerst2man.py ./doc doc/_build/man
  Traceback (most recent call last):
    File "./doc/prerst2man.py", line 65, in <module>
      os.system('set -x; rst2man {0} {1}/{2}.{3}'
  NameError: name 'os' is not defined
  make: *** [doc/_build/man/man1/notmuch.1] Error 1

by using system directly.  We don't need the 'os.' namespacing,
because the function was imported with:

  from os import makedirs, system

10 years agodoc/prerst2man.py: Use Python-3-compatible octal notation
W. Trevor King [Sat, 5 Apr 2014 17:31:07 +0000 (10:31 -0700)]
doc/prerst2man.py: Use Python-3-compatible octal notation

Python 3 only supports the 0oXXX notation for octal literals [1,2],
which have also been supported in 2.x since 2.6 [2].

[1]: https://docs.python.org/3.0/whatsnew/3.0.html#integers
[2]: http://legacy.python.org/dev/peps/pep-3127/

10 years agodoc/mkdocdeps.py: Use "with" statement for the output file
W. Trevor King [Sat, 5 Apr 2014 17:31:06 +0000 (10:31 -0700)]
doc/mkdocdeps.py: Use "with" statement for the output file

Before this patch, the open was unnecessarily early and relied on the
process cleanup to close.  Neither one of these was a real problem,
but PEP 343's context managers (which landed in Python 2.5) make
proper cleanup very easy.

[1]: http://legacy.python.org/dev/peps/pep-0343/

10 years agodoc/mkdocdeps.py: Convert execfile to import
W. Trevor King [Sat, 5 Apr 2014 17:31:05 +0000 (10:31 -0700)]
doc/mkdocdeps.py: Convert execfile to import

excefile is gone in Python 3 [1].  Instead of exec-ing the
configuration, it's easier to insert the source directory in Python's
path [2], and just import the configuration.  With this change,
mkdocdeps.py is compatible with both Python 2 and 3.

[1]: https://docs.python.org/3.0/whatsnew/3.0.html#builtins
[2]: https://docs.python.org/3/library/sys.html#sys.path

10 years agoNEWS for displaying tag changes
Mark Walters [Sat, 19 Apr 2014 08:52:35 +0000 (09:52 +0100)]
NEWS for displaying tag changes

10 years agodoc: make notmuch-new summary line more generic
David Bremner [Sat, 19 Apr 2014 05:29:06 +0000 (14:29 +0900)]
doc: make notmuch-new summary line more generic

Since 'notmuch new' now takes multiple options, it's confusing to show
only one of them in the summary.

10 years agoNews for changes from Austin Clements
Austin Clements [Sat, 19 Apr 2014 03:31:03 +0000 (23:31 -0400)]
News for changes from Austin Clements

10 years agoemacs: Honor debug-on-error for part renderers
Austin Clements [Fri, 18 Apr 2014 22:57:17 +0000 (18:57 -0400)]
emacs: Honor debug-on-error for part renderers

Previously, even if debug-on-error was non-nil, the debugger would not
trap on part renderer errors.  This made debugging part renderer bugs
frustrating, so let the debugger trap these errors.

10 years agoconfigure: fix comment, pass HAVE_CANONICALIZE_FILE_NAME to build
David Bremner [Thu, 17 Apr 2014 22:24:21 +0000 (07:24 +0900)]
configure: fix comment, pass HAVE_CANONICALIZE_FILE_NAME to build

Apparently omitting it is not fatal, but let's be consistent with the
other compat functions.

10 years agobuild: add canonicalize_file_name to symbols exported from libnotmuch.so
David Bremner [Wed, 9 Apr 2014 11:24:02 +0000 (08:24 -0300)]
build: add canonicalize_file_name to symbols exported from libnotmuch.so

This is needed for our compat version of canonicalize_file_name to be used.

10 years agodoc: Fix parallel build of roff files
Austin Clements [Thu, 17 Apr 2014 20:34:57 +0000 (16:34 -0400)]
doc: Fix parallel build of roff files

The roff build rule builds all of the roff files in a single command.
Previously, this was expressed as a multi-target rule, but since this
is equivalent to specifying a copy of the rule for each target, make
-jN could start up to N parallel instances of this command.  Fix this
by bottlenecking this rule through a single stamp file.

This also removes the unused man.stamp from CLEAN.

10 years agoNEWS: notmuch-init-file and notmuch-emacs-version related news
Tomi Ollila [Mon, 14 Apr 2014 18:32:35 +0000 (21:32 +0300)]
NEWS: notmuch-init-file and notmuch-emacs-version related news

Along with those, removed trailing space from subsection title.

10 years agoNEWS: Document the recent 'nmbug clone' and @{upstream} changes
W. Trevor King [Tue, 15 Apr 2014 21:03:56 +0000 (14:03 -0700)]
NEWS: Document the recent 'nmbug clone' and @{upstream} changes

The changes landed with c200167 (nmbug: Add 'clone' and replace
FETCH_HEAD with @{upstream}, 2014-03-09).

The preferred markup language for NEWS seems to be Markdown, which is
parsed by devel/news2wiki.pl into Markdown chunks for rendering by
ikiwiki [1].

[1]: http://notmuchmail.org/news/

10 years agotest: use test_expect_equal for PATH test, update message
David Bremner [Sun, 13 Apr 2014 12:42:49 +0000 (09:42 -0300)]
test: use test_expect_equal for PATH test, update message

- The old test was quite impossible to debug; the new one shows the difference
  between the two directories, if any.

- "repository" doesn't make sense for out of tree builds. Or tarball
  builds, for that matter.

10 years agonmbug: mark repository as bare on clone
David Bremner [Sat, 12 Apr 2014 16:30:32 +0000 (13:30 -0300)]
nmbug: mark repository as bare on clone

If a git repository is non-bare, and core.worktree is not set, git
tries to deduce the worktree. This deduction is not always helpful, e.g.

% git --git-dir=$HOME/.nmbug clean -f

would likely delete most of the files in the current directory

10 years agoNEWS: document possible breakage from saved-search format change
Mark Walters [Sat, 12 Apr 2014 12:46:27 +0000 (13:46 +0100)]
NEWS: document possible breakage from saved-search format change

If the user has unsorted or alphabetically sorted saved-searches these
should continue to work. If they have some custom lisp sort function
then it will need updating to work with the new saved-sort
format. Document this, and how the updating should be done.

Also roll in fixes for the markdown in the first part of the NEWS
suggested by Tomi: change `just work' to *just work* and removed
periods from the end of subtitle lines.

10 years agoemacs: hello: bugfix: make alphabetically sorted saved searches work
Mark Walters [Sat, 12 Apr 2014 12:46:26 +0000 (13:46 +0100)]
emacs: hello: bugfix: make alphabetically sorted saved searches work

My recent changes to the saved search format broke the alphabetically
sorted saved sort option. This makes it work again.

Also update docs for saved-search sort defcustom to match the new
format.

Finally, since the saved-search list is no longer an alist change the
names in the sort function to avoid confusion.

10 years agoemacs: remove auto-signing of replies to signed messages
Jameson Graef Rollins [Mon, 14 Apr 2014 19:40:50 +0000 (12:40 -0700)]
emacs: remove auto-signing of replies to signed messages

It was decided that auto-signing is potentially too troublesome for the
apparently common case of users who enable crypto processing for the
purpose of checking signature validity but who are not in a position to
sign out-going messages.  Users can still manually invoke signing as needed.

Encrypting replies to encrypted messages is more of a security issue
so we leave it in place.

10 years agoemacs: add $(srcdir) to notmuch-version.el.tmpl dependency
Tomi Ollila [Sat, 12 Apr 2014 11:59:46 +0000 (14:59 +0300)]
emacs: add $(srcdir) to notmuch-version.el.tmpl dependency

This fixes out-of-tree build when generating emacs/notmuch-version.el.

10 years agoconfigure: add $(ZLIB_CFLAGS) to CONFIGURE_CFLAGS
Tomi Ollila [Sat, 12 Apr 2014 16:30:22 +0000 (19:30 +0300)]
configure: add $(ZLIB_CFLAGS) to CONFIGURE_CFLAGS

As it is defined in CONFIGURE_CXXFLAGS.