Age | Commit message (Collapse) | Author |
|
|
|
Check $XDG_USER_DATA/feedreader/GrabberConfig for grabber configuration
files first. This allows for user-local testing/development of site
grabber configurations without needing root access nor enabling the
grabber for all users.
|
|
Unless you’re going to reimplement all the functionality from
g_log_writer_default() (or deliberately don’t want it), you should not
set a custom writer function. g_log_writer_default() likely does what
you want.
In particular, setting the writer function to
g_log_writer_standard_streams() meant that `G_MESSAGES_DEBUG` was
ignored, resulting in all debug messages being logged unconditionally
and sending gigabytes of data to the disk over the course of several
days. This noticeably slowed down people’s systems.
It also meant that fatal messages did not actually abort the program,
which hides bugs.
g_log_writer_default() handles those things, then tries sending log
messages to the journal. If that fails, it actually calls
g_log_writer_standard_streams().
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Fixes: https://github.com/jangernert/FeedReader/issues/970
|
|
|
|
On at least the DecSync backend, `addFeed` involves downloading the
specified feed, and that's slow. Doing it synchronously means that the
UI hangs for a while. I've never written any Vala before but this patch
still lets me add feeds without stopping the UI, so hopefully it's good?
|
|
|
|
is it OK to replace other typos in this file
|
|
This makes it so we can use the escaped-HTML output of Htmlclean
directly as titles, and things like <pre> will show as <pre>
in titles.
See #918
|
|
- Fix double-negative in delete query
- Change logic to use days instead of weeks
- Use to_start_date() consistently
|
|
The NextCloud plugin was doing this instead of using last-sync. This way
all of the plugins do this behavior.
|
|
|
|
new feed
This is a better user experience because articles only close when a user explicitly
closes them.
Fixes #864
|
|
This adds a new "close" button in the top left of the article view
to close the current article without needing to switch views.
For now, I'm not displaying this in fullscreen mode.
|
|
|
|
Previously, we used the RSS <link> to determine if a feed was a duplicate, but
it's common for multiple feeds to exist for a single site. This changes the
logic to just deduplicate based on the actual link given to us by a user.
I also fixed a bug where we checked more than 1 duplicate instead of just
checking for any duplicates.
See #899 and #883
|
|
This was previously restricted to just the Unread articles tab, but
the hover text for this was non-obvious and it seems strange to have
this setting only apply on one tab.
Fixes #901
|
|
|
|
This lets us restore it when people go from one list to another, when
the article is still visible on the new list.
See #864
|
|
This isn't perfect but it's way better than what Uncrustify generated
|
|
This will make it easier to programatically fix the indentation.
|
|
- Use configured font and size if applicable
- Fallback to system font
- Fallback to Sans 12
This makes it so we will always display something reasonable no matter
how broken the system settings are.
|
|
Previously we defaulted to Droid fonts, but we shouldn't be trying to
make assumptions about what fonts users have installed or what they
want to use here.
This makes it so we use the system document font by default and we
still allow users to override it at-will in the settings.
See #877
|
|
Previously it also called addFeeds(), but I moved that into the
function that calls it.
|
|
For functions with obvious implementations where a plugin won't be
broken if they're missing, this adds default implementations.
|
|
This just calls addFeed() in a loop. I removed all empty implementations
and implementations that had essentially the same logic.
|
|
This will let us deduplicate code in cases where a function can be implemented
in terms of another.
|
|
https://gitlab.gnome.org/GNOME/Initiatives/issues/4
|
|
|
|
|
|
This makes it easier to add a feed and matches the behavior users expect from
working with browsers.
|
|
This shouldn't happen, but it does sometimes. For now, at least make us not crash.
|
|
A commit a long time ago removed the tags column from articles and replaced it with
the taggings table, but it looks like the code to rename and delete tags, and also
to show all tagged articles was broken.
This fixes all of the references to the old tags column and switches them to
reference taggings instead. I also inlined the all-tags subquery.
This should fix #840
|
|
These cases don't involve any threading so this should speed things
up a little without any risk.
|
|
This is a really heavy-handed way to fix thread safety in our plugins, by
using a different SQLite handle every time we do anything with the DB.
|
|
- Fix grabber to do DB inserts in the main thread
- Remove unsafe threading throughout the UI
|
|
|
|
This uses the changes in these two Uncrustify pull requests:
https://github.com/uncrustify/uncrustify/pull/2142
https://github.com/uncrustify/uncrustify/pull/2137
|
|
|
|
Fixes #762
|
|
|
|
|
|
|
|
Copy article url from ArticleRow
|
|
When a category is removed, uncategorize feeds/categories explicitly to plugins
|
|
|
|
|
|
|
|
|
|
This has two improvements over the previous version:
- We don't try to unlock the keyring if we're not using it (i.e. the local plugin)
- We will unlock the keyring again if it gets locked
|
|
|