| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Chrome 70 has deprecated Shadow DOM v0
and displayed a warning when it's used.
And accroding to https://www.chromestatus.com/features/4507242028072960,
Shadow DOM v0 will be removed since Chrome 73.
The support for Shadow DOM v1 has been there since Chrome 53 and
is enabled by default on Firefox 63,
so it's time to use it.
Old support is still needed because Vimium's min_chrome_version is 51.
|
|
Fixes #3154 (I think).
|
|
use Element::scrollBy if it exists
|
|
When yanking an email URL with `yf`, it's likely to be more useful to
strip the leading "mailto:".
Fixes #3123.
|
|
This reverts commit aff380669445594e7566a835d4b27f11da26ea6a.
This breaks scrolling on Twitter.
Needs more thought.
|
|
If Vimium becomes disabled, then we remove this listener.
So we better add the actual function we want to (later) remove, rather
than creating a new function (which will never be removed.
|
|
This selects the body of a message, if such is displayed.
|
|
1. Use verb phrase for function name.
2. Add `Scroller.reset()` method. This *only* resets the activated element.
3. Reset the scroller only if the URL has changed. (Previously, in #3119, the scroller was also being reset when the tab gained the focus.)
Based on a suggestion from @marcotc.
|
|
|
|
|
|
|
|
|
|
Twitter is an important site and Vimium's scrolling is currently broken
when a tweet is expanded.
In my opinion, the existing bahviour is so bad and that a site-specific
hack is warranted.
Fixes #3045.
|
|
Launching link hints can sometimes be slow. For filtered hints, the
user already knows what to type - but must nevertheless wait until the
hints have been calculated and rendered.
Here, we cache intervening keydown events, and replay them once
hints-mode proper is launched. This should make improve usability (in
the case of filtered hints).
Fixes #3050.
|
|
First, I *very rarely* use local marks. They just don't seem as useful
in a browser as they are in a text editor.
However, on a page like GMail, I do often want to jump back and forward
between my labels. These are different locations hashes (anchors).
The idea here is to make local marks useful for this case.
If the scroll positions (X and Y) are both 0, and the hash (anchor) is
present in the mark and non-empty, then change the hash instead of
scrolling.
On first tests, this appears to work nicely for changing labels in
GMail. I'm sure there are other uses.
This change shouldn't interfere with use cases where the user has
scrolled within the page.
|
|
|
|
Chrome has never searched in all frames, so this is no loss.
However, find on Firefox Quantum is now broken with the "all frames"
flag set.
Fixes #2962.
H/t: @gdh1995 (https://github.com/philc/vimium/issues/2962#issuecomment-367991086).
|
|
Move the `reload` command to the background page and make it accept a
count prefix.
Fixes #674.
Althought the demand for this is low and the use cases limited, it is a
very vim-ish extension to the existing command.
|
|
It looks like we'll be initialising the Vomnibar on demand (see #2939,
merged). In this case, we should also initialise the flash frame on
demand. This means that *none* of Vimium's UI elements are now created
until they are needed.
|
|
This is #2941.
|
|
|
|
|
|
Disable pre-loading of Vomnibar iframe.
|
|
It appears that nobody is using this (and it is broken). So this PR
removes this feature.
Fixes #2937 (kind of).
|
|
Pre-loading the Vomnibar iframe seems to be causing some issues with the
dev console. This PR disables pre-loading of the Vomnibar.
I'm going to try this out for a while. If it doesn't seem to sluggish,
then I'll merge it.
|
|
This handler is undefined at the point that it is installed. The
definition is at the bottom of the file.
|
|
This sounds like #2035, #2305 is unrelated.
|
|
... use literals instead.
For some reason, on Firefox, using `vimiumOnClickAttributeName` was
causing link hints to fail silently (on Facebook). It's not at all
clear why this was happening.
Simply repeating the literal fixes the issue.
Fixes #2879.
|
|
This tweaks the scoring of hints for filtered hints.
Specifically, it scores matches at the start of the text or the start of
a word higher than whole-word matches.
This is intended avoid the phenomenon whereby adding matching characters
to the filter causes the selected hint to change.
|
|
Since these are executing in the page's context, they could break the
page if they were to fail (for whatever reason). So we add some
(possibly unnecessary) guards, just in case.
|
|
We should rotate on `<Shift-Space>` too.
|
|
Detect click listeners for link hints.
|
|
`forTrusted()` for `checkEnabledAfterURLChange()` after web navigation
is being called with a request (not an event). So make the request
trusted.
The symptom was that pass keys and the enabled state was not being
correctly updated on web navigation.
|
|
This is a simpler version of #1167. It detects clickable elements with
listeners added with `addEventListener()`.
It includes some of @mrmr1993's ideas from #1167 (in fact, it's mostly
those ideas tweaked into a slightly different form).
|
|
Add "swap" command option for marks.
|
|
|
|
This removes @insertModeLock and instead always checks whether we are
in insert mode dynamically (as we previously did only for contentEditable).
@mrmr1993... Would you be able to check this, please? It is intended to
be mostly a no-op.
Fixes #2841.
|
|
Tiny tweaks...
1. The test is clearer like this.
2. Use a more CoffeeScript-ish approach to sub-string.
|
|
Use HTML labels accessor to get link hint labels for input elements
|
|
Tweak of #2601.
|
|
Enable Firefox clipboard commands
|
|
|
|
|
|
|
|
|
|
Merging #2792 (mention @mrmr1993).
|
|
Previously, @tabCount was reset on the `<Shift>` `keydown` event, which
defeats the purpose of #2792. Also, the @tabCount was reset on other
non-printable characters such as `<LeftArrow>`.
|
|
|
|
This fixes #2830: it stops timesofindia.indiatimes.com from failing due
to FF bug 1408996.
|
