In addition to the same type of changes as converting other
subcommands, add the possibility of creating a database at the top
level. It would probably make sense to use this for insert as well.
This will allow client code to provide more meaningful diagnostics. In
particular it will enable "notmuch new" to continue suggsting the user
run "notmuch setup" to create a config after "notmuch new" is
transitioned to the new configuration framework.
This will need some cleanup when the transition completes, and we stop
passing notmuch_config_t structs to the subcommands.
Unlike the general case, we open the database in the subcommand, since
we don't know whether it should be opened read/write until we parse
the command line arguments.
Add a test to make sure passing config file on the command line is not
broken by these or future config related changes.
This is intended for use in temporary code transitioning to the new
configuration system. The name is chosen to avoid cluttering the
notmuch_config_* namespace further with non-library functions.
Since we are already passing a search context around as a kind of
parameter block, add a new talloc context to that to replace relying
on 'config'.
Convert notmuch-search and notmuch-address at the same time, because
they share some code.
Add a test to make sure we don't break passing configuration as a
command line argument.
The new talloc context is needed to run the hook at the very end of
the function. That in turn is needed so that this process gives up the
write lock on the database.
This conversion is trivial because the only configuration information
accessed by dump is that stored in the database (in order to dump
it). We do need to be careful to keep the write lock on the database
to ensure dump consistency.
The main effort is changing from the old argv style config list
iterators to the new more opaque ones provided by the library (and
backed by the database+file config cache).
This will allow transitioning individual subcommands to the new
configuration framework. Eventually when they are all converted we can
remove the notmuch_config_t * argument.
For now, live with the parameter shadowing in some some subcommands;
it will go away when they are converted.
By using an enum we can have better error detection than copy pasting
key strings around.
The question of what layer this belongs in is a bit
tricky. Historically most of the keys are defined by the CLI. On the
other hand features like excludes are supported in the
library/bindings, and it makes sense to configure them from the
library as well.
The somewhat long prefix for notmuch_config_t is to avoid collisions
with the existing usage in notmuch-client.h.
The renaming and extra values will make sense when we start to convert
subcommands to the new configuration framework. It will also avoid
collisions with a new enum for configuration keys to be introduced in
a future commit.
Fill in the remainder of the documented functionality for
n_d_open_with_config with respect to config file location. Similar
searching default locations of the database file still needs to be
added.
The main goal is to allow configuration information to be temporarily
overridden by a separate config file. That will require further
changes not in this commit.
The performance impact is unclear, and will depend on the balance
between number of queries and number of distinct metadata items read
on the first call to n_d_get_config.
When prompting for one or more tags to add or remove to/from one or
more threads, ensure that the set of tags offered for completion
contains no duplicates.
Some completion packages (e.g. selectrum) will include every member of
the offered list, resulting in the same tag being indicated as a
possibility several times.
In ee897cab8b the upgrade tests from pre v3 databases were
removed. The reasons for that are still valid, but we should still
test the code paths that do the upgrade, and it is relatively
straightforward to do that for v3 to v3 upgrades.
The output of "notmuch show --format=sexp --format-version=4"
may contain `:content-type' entries with `nil' as the value,
when it fails to detect the correct value. Account for that
in a few places where we would otherwise risk a type error.
Note that `string=' does not choke on `nil' because it uses
the `symbol-name' when encountering a symbol.
Users may type some text into the buffer on an address line, before
actually invoking address completion. We now use that text as the
initial input when we begin address completion.
Previously we did knowingly replace the actual initial input with some
completion candidate that happens to match. Which candidate is used is
essentially random, at least when the actual initial input is short.
As a result users very often had to begin completion by deleting the
less than helpful "initial input".
IMO Notmuch should not override the default completion mechanism by
default, at least not globally. But since users are already used to
this behavior it is probably too late to change it. Do the next best
thing and at least allow users to opt out.
When called from code, then this function returns non-nil when the
message at point is a matched message. However it does nothing at all
to present that information to the user when it called interactively.
It is therefore safe to conclude that nobody is using this as a
command.
Like `cl-lib' and `pcase', which are already available in all
libraries, `subr-x' also provided many useful functions that
we would like to use.
Making `subr-x' available in every library from the get-go means
that we can use the functions it defines without having to double
check every single time, whether the feature is already available
in the current library.
We need to load `cl-lib' at run-time because we use more from it than
just macros. Never-the-less many, but not all libraries required it
only at compile-time, which we got away with because at least some
libraries already required it at run-time as well.
We use `cl-lib' and (currently to a lesser extend) `pcase' throughout
the code-base, which means that we should require these features in
most libraries.
In the past we tried to only require these features in just the
libraries that actually need them, without fully succeeding. We did
not succeed in doing so because that means we would have to check
every time that we use a function from these features whether they
are already being required in the current library.
An alternative would be to add the `require' forms at the top of every
library but that is a bit annoying too.
In order to make sure that these features are loaded when needed but
also to keep the noise down we only require them in "notmuch-lib.el",
which most other libraries require, and in most of the few libraries
that do not do so, namely "notmuch-draft.el", "notmuch-message.el" and
"notmuch-parser.el". ("coolj.el", "make-deps.el", various generated
libraries, and "notmuch-compat.el" are left touched.)
To some extend this is a personal preference, but the preference is
strongly dependent on whether one is used to a language that makes it
necessary to use variables like this.
This makes it perfectly clear that we are first getting and then using
a "foo":
(use-foo (get-foo))
Sure this has to be read "inside out", but that's something one better
gets used to quickly when dealing with lisp. I don't understand why
one would want to write this instead:
(let ((the-foo (get-foo)))
(use-foo the-foo))
Both `get-foo' and `use-foo' are named in a way that make it very
clear that we are dealing with a "foo". Storing the value in an
additional variable `the-foo' does not make this any more clear.
On the contrary I makes the reader wonder why the author choose to
use a variable. Is the value used more than once? Is the value
being retrieved in one context and then used in another (e.g. when
the current buffer changes)?
