| Commit message (Collapse) | Author | Age |
|---|
| | |
|
| |\ |
|
| | |\ |
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | | |
Esc key, 2 second wait after scroll, and whenever the messages inChannel
change. Kinda gross set of things, but, so it goes. This does offer us
the option of extending this to include "when you click 'unread from
here' on a message" in future.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This requires both the channels and messages stores to be available, so
we do it in the page file where we've got those stores handy. It's a bit
inefficient (it recreates the entire enrichedChannels array every time
one of them changes), but it'll do for now.
There's also a TODO hinting that we might want an idea of unread-count.
I think that would require a distinct idea of mentions, because we just
want "fact of unread" for general-availability channels, not "this many
messages since you were last here".
|
| | | |
| | |
| | |
| | | |
Even when they get mapped to snake_case searchParams.
|
| | | |
| | |
| | |
| | | |
I dunno, I like the fleuron. Maybe it's too twee?
|
| | | |
| | |
| | |
| | |
| | |
| | | |
This includes jamming the "at" of a message into a data- attribute on
the Message component, so that it can later be used by parent components
via Plain Old Javascript and the .dataset attribute of an HTML node.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
I tried to have a custom class for Channel objects, but Svelte's
automatic proxy logic works only on bare objects, as far as I could
tell. So that broke everything. I resorted to a function that would
build the bare objects, but we still lack methods that I think would
make life easier ("touch last read" etc).
|
| | |/
| |
| |
| |
| | |
This handles datetimes better than the browser-builtin, and is in many
ways a successor to Moment, the previous standard choice.
|
| | | |
|
| |\| |
|
| | |\ |
|
| | | |
| | |
| | |
| | | |
the unified build.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | | |
* It now _always_ runs both the backend and the frontend.
* The frontend is now _always_ exposed on both `localhost` and on the machine's local network interfaces.
The role of running a consolidated server is now handled by `cargo run` (this has been possible since e7d4b6d7ddbcd0128e47476e6cd1d824a1929f3c anyways).
|
| | | | |
|
| | | |
| | |
| | |
| | | |
Svelte's lint complains if you add interaction to a div but don't give that div an ARIA role.
|
| | | | |
|
| | | |
| | |
| | |
| | | |
`test-all` instead of `test` to avoid colliding with a shell builtin.
|
| | |\ \
| | |/
| |/| |
|
| | | |
| | |
| | |
| | | |
If you're setting it to a static value, use set.
|
| | | |
| | |
| | |
| | | |
Just use state, don't derive from a non-state local variable.
|
| | |/
| |
| |
| | |
If you're setting it to a static value, use set.
|
| | | |
|
| | |\ |
|
| | | |
| | |
| | |
| | |
| | | |
Hide and show channel menu thus. It doesn't gradually pull it out, which
is less than ideal, but it's good enough for now.
|
| | | |\ |
|
| | | |\ \ |
|
| | | |\ \ \ |
|
| | | |\ \ \ \ |
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
This includes a change to tools/run that exposes the dev server on my
local network. This change should not make it into the final form of
this branch.
This is so I can use actual for real touch events on my actual for real
phone, hooked up for remote debugging to my computer so I can see
console events etc.
|
| | | | | | | | |
|
| | |\ \ \ \ \ \ |
|
| | | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
Just basic "will it mount?" tests for more components.
|
| | | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
Turns out a number of our components are a pain to run via a mounting
test, but Svelte does suggest that this is the dispreferred way to test
anyway. Using use:, triggering navigation after logout, and
scrollIntoView all make the headless node test process with the janky
in-memory mount have a Very Bad Day.
But this is still progress! We will press on.
|
| | | |\ \ \ \ \ \
| | | | |_|_|_|/
| | | |/| | | | |
|
| | | | | | | | | |
|
| | | | | | | | | |
|
| | | | |_|_|_|/
| | |/| | | | |
|
| | |\ \ \ \ \ \
| | | |/ / / /
| | |/| | | | |
|
| | | |\| | | | |
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
The original retention values were loosely based on Slack's retention, for lack of a more specific motivator. Today's election results have changed my views; the service now defaults to retention more in line with the needs of communities for which deep message history may be a risk:
* Unused channels expire after 7 days.
* Used channels expire when their last message expires (as before).
* Deleted channels are purged after 6 hours (which is in line with the purge behaviour of messages).
* Messages expire after 15 days.
* Deleted messages are purged after 6 hours (as before).
No changes have been made to token expiry.
|
| | | |/ / / /
| |/| | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
rendered.
~16% of the `hi development` channel's rendering time was taken up on this.
|
| |\| | | | | |
|
| | | | | | | |
|
| | | | | | | |
|
| | |\ \ \ \ \ |
|
| | | |\ \ \ \ \
| | |/ / / / /
| |/| | | | | |
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Service worker basics I guess.
|
| | | | | | | | |
|
