Ensure we sort correctly (sort_container), since the last message in a
thread is not necessarily the last (when not sorted by thread-date).
own_date_key is redundant, so let's remove it.
This is a revival of commit c4ccaf0fdb:
"mu: sort childs of thread based on the sortfield only",
whose effect was apparently lost during the rewrite to C++.
Today when we query a find cmd with the `--threads` option, all the
children of each thread are sorted according to the most recent
message in the thread.
This patch changes how the children of a thread are sorted. Threads
are still sorted according to the thread date, but the children of each
thread are now sorted by their own date only.
Here is an example of what happened with the previous sorting:
Example with random kernel thread sorted by date:
[PATCH 0/4] drm/panel: jh057n0090: Add regulators and drop magic value in init
┣━▶[PATCH 1/4] MAINTAINERS: Add Purism mail alias as reviewer for their devkit's panel
┣━▶[PATCH 2/4] drm/panel: jh057n0090: Don't use magic constant
┣━▶[PATCH 3/4] dt-bindings: display/panel: jh057n0090: Document power supply properties
┗━▶[PATCH 4/4] drm/panel: jh057n0090: Add regulator support
If someone replies to one of these emails in the middle, its date will
bubble-up and the sorting becomes:
[PATCH 0/4] drm/panel: jh057n0090: Add regulators and drop magic value in init
┣━▶[PATCH 2/4] drm/panel: jh057n0090: Don't use magic constant
┃ ┗━▶ Re: [PATCH 2/4] drm/panel: jh057n0090: Don't use magic constant
┣━▶[PATCH 1/4] MAINTAINERS: Add Purism mail alias as reviewer for their devkit's panel
┣━▶[PATCH 3/4] dt-bindings: display/panel: jh057n0090: Document power supply properties
┗━▶[PATCH 4/4] drm/panel: jh057n0090: Add regulator support
With this patch, we will have the following output:
[PATCH 0/4] drm/panel: jh057n0090: Add regulators and drop magic value in init
┣━▶[PATCH 1/4] MAINTAINERS: Add Purism mail alias as reviewer for their devkit's panel
┣━▶[PATCH 2/4] drm/panel: jh057n0090: Don't use magic constant
┃ ┗━▶ Re: [PATCH 2/4] drm/panel: jh057n0090: Don't use magic constant
┣━▶[PATCH 3/4] dt-bindings: display/panel: jh057n0090: Document power supply properties
┗━▶[PATCH 4/4] drm/panel: jh057n0090: Add regulator support
Cc: Julien Masson <massonju.eseo@gmail.com>
Signed-off-by: Jerome Brunet <jbrun3t@gmail.com>
Since we require meson 1.3(.2), we can modernize the usage a bit,
as well as improve some logic
- use / instead of join_paths
- use option.allowed() instead of !option.disable()
- use cxx.get_supported_arguments
- use fs.copyfile instead of configure_file
- show a summary at the end of the configuration
For the (rare) messages without a Date: header, attempt to guess a date
from the top (most-recent) Received: header. Not perfect, of course, but
better than nothing.
Requires re-indexing to apply to already indexed messages.
Should help for #1083.
Move the various seq_ functions, as well as std::(stable_)sort,
std::accumulate, std::transform, std::find, std::find_if to their C++20
std::ranges counterparts.
Cleanup the query running and handling the results.
- in threading, use stable_sort
- remove dead code
- fix indentation in a few places
- don't need Option in various run_... in mu-query.cc
Support "system" properties, i.e., library versions, whether
language-detection, scm is supported.
Make these more generic (with display strings), so we can handle them
more automatic int mu-cmd-info.
Some improvements:
- Fix or/xor chains (use left-associativity).
- Fix quote handling
- Make parsing O(n) rather than quadratic; limit recursion depth
And update tests.
We already pre-fetched db file paths for the cleanup-case (since
cdb619e4f), now let's generalize this to all indexing.
For now this is mostly performance-neutral (lazy-rescan is slightly
faster); however, this simplifies the code.
Update various places for what we can do with C++20:
- ends_with / starts_with
- std::to_array
- using instead of typedef
- designated initializers
- "[[maybe_unused]]" instead of G_GNUC_UNUSED
Also remove some unnecessary 'static'
Modernize code a bit; and use a temporary dir for the working set.
Typically, this makes the tests run faster (since the temporary dir is
often in-memory), which allows focusing on the part we can most easily
influence.
Do not consider calender-invitations "attachments"; do mark as
"calendar". Do recognize application/ics as calendar messages.
Update mime-object to expose a message part's disposition.
Change the "is-attachment" heuristic to include inline parts if they
have a filename parameter in their content-disposition.
Note that this doesn't change things radically; the delta is +69 and
-202 for ~6500 attachments.
