| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
| |
we now store images ourselves for caching rather than relying on
url-automatic-caching.
|
| |
|
|
|
|
|
|
| |
when we fetch images, check if they are cached, and if so use the cached
version.
for now, images aren't cached explicitly, but this should work if the user has
`url-automatic-caching' enabled.
|
| | |
|
| | |
|
| |
|
|
| |
and put point at point-min on profile load
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
images are tab stops. click or RET runs shr-browse-image. a prefix arg
copies the URL.
images use the mastodon-tl--shr-image-map-replacement for extra functions like zoom
image, save image, rotate image, etc.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
Instead of making it search the whole buffer every time to find images to load, give it a range where this work should be done.
We then call this immediately after inserting a single status, notification, ...
There should be no big noticible difference - images might load a tiny bit sooner although I doubt you can see that.
This should be more efficient on large buffers although Alex didn't notice any problems when testing streamed buffers.
We should still do it as it make things easier to understand. I was always worried about these global operations.
|
| |
|
|
|
|
| |
* Bump version numbers to 0.7.2
* Adjusted functions to bring line length below 90.
|
| |
|
|
|
|
|
| |
the minibuffer.
These happen during async loads, i.e. when the user might be doing something else completely.
Especially with auto-update that are coming this is becoming annoying.
|
| |
|
|
|
|
| |
- Small logic fix in mastodon-tl--media to remove redundant newline
- Replaced switch-to-buffer with with-current-buffer in mastodon-media--process-image-response
- Squashed with merge to develop
|
| |
|
|
|
|
| |
exists.
This should fix https://github.com/jdenen/mastodon.el/issues/156.
|
| | |
|
| |
|
|
|
| |
On Emacs24 I've been able to reliably fail url fetching which exposed issues in loading a timeline (it aborts the loading).
This catches any errors, marking the image load as failed so that we won't retry (retries are a TODO item I guess) and then succeeds the function so the rest of the timeline loading can proceed.
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
We do this by
- moving vars into the files where they are (mostly) used
- "declaring" vars used elsewhere with the (defvar <var-name>) pattern,
- declaring functions defined in others functions rather than loading the file via require.
|
| | |
|
| |
|
|
|
| |
We only support Emacs 24 and 25 so are free to use sane bindings by default.
To keep the linter happy, we need to declare dependency on Emacs 24 in all files.
|
| |
|
|
| |
This also includes tweaks to make Travis happy — tests previously did pass on my laptop but Travis's environment is different.
|
| |
|
|
|
|
|
| |
images.
Having all the logic in one file reduces interdependencies.
Having default images is more pleasing during the incremental loading.
|
| |
|
|
|
|
|
| |
Now that we are also loading avatars there is a lot of image loading to do to show the timeline.
We can do the loading asynchronously to let the user have a look at the toots already while image loading is incrementally proceeding.
We can no longer enforce caching of avatar loading since the variable is consulted when the response parsing happens at which point the dynamic binding we had used so far has gone out of scope again.
|
| |
|
|
|
|
|
|
|
| |
- Shows users' avatars (makes only sense if Emacs is built with imagemagick)
- Scales media attachement previews to a max size (if Emacs is built with imagemagick)
- Enable cacheing of image fetches
Known issues:
- We should really cache the avatars to avoid having multiple identical images in memory.
|
| | |
|
| |
|
|
| |
(Besides, just "()" would have been enough anyway. No need to quote "()" or "nil".)
|
| |
|
|
|
| |
Re-do it as a loop rather than raising an error when the search fails
and trying to catch that wholesale.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
