Fixes: #3865
Currently works by allowing the exact strings "DEV_NULL" or "INHERIT" to
be passed to return_std*.
Signed-off-by: aarondill <aaronsacks2006@gmail.com>
* Stop using top level properties for vertical/horizontal
* No abbreviations
* Use height/width instead of size
* Don't use methods where properties can be used
This is long overdue. A bit of historical context. The grid API
is losely somewhat based on the old `radical` module, but was
heavily improved by @getzze. That version had row_span and col_span.
This made the way the previous implementation coded the border
incompatible. I spent some time back then trying to bolt it back on,
but the complexity is quite high and never made it work right.
This commit goes in another direction. Rather than draw the border,
it creates a mask where the border should *not* be, then bucket fill
the widget. This is the equivalent of CSS `border-collapse`.
It also support custom borders. This allows dashed lines and partial
borders.
The main use case will be to add border support to the calendar. It
was previously possible to partially do it using custom cell painters,
but was pretty hacky. Now that the calendar will deprecate the custom
painters in favor of `widget_template`s, a more robust alternative was
required.
The drawback of this commit is obviously the added complexity to the
most complex layout. This is why it adds many tests to cover the various
corner cases.
This is already used in the `wibox.layout.manual` layout. It makes
the widget easier to use. Previously, using the imperative syntax
was necessary for most grids.
Previously, it would raw_set properties like floating. This means the
signals and the result of calling `awful.client.property.get` was
different. Some older code uses `awful.client.property.get` or
`awful.client.object.get_floating` rather than `c.floating`.
With this change, all native properties should be handled as so
and all non native properties set at the end, after `__newindex`
is defined.
The list of properties was extracted using
cat client.c
| tail -n100
| grep -Eo '["][^"]+["]'
| grep -Eo '[^"]+'
In tests/examples/wibox/widget/textbox/line_spacing.lua, there is a typo
It's supposed to be "has", but it's "shas", this is obviously not a grammar issue, so it's a typo.
This can also act as an auto-save feature if the delay is zero. It
also adds more signals. These signals are intended for creating
notifications. `awful` cannot safely depend on `naughty`, so this
intergration will have to be done by the users.
* Fix all warnings
* Make indentation consistent across the file
* Simplify/unify the validation (use `error()` rather
than try to silently fix problems, move to setters)
and fallback code (move to the getters rather than...
everywhere)
* Write the documentation
* Finish the tests
* Remove unnecessary constructors
* Remove path builder because it belongs in
`gears.filesystem` and wasn't really used anyway
* Add more properties and a beautiful variable to
replace `set_defaults` and hardcoded values.
* Replace callbacks with signals (common pattern
in modern AwesomeWM APIs)
* Moved from `os.date` to GLib and some boilerplate
code to make Debian devs less angry
* Changed the way the snipping tool selection wibox works.
Now it freeze the screenshot before the selection.
The old way had a bunch of side effects for things
like Qt/GTK comboboxes. It also could crash.
There will still be one more commit to replace some of the screenshot
module code with available functionality from gears (e.g. the filesystem
module).
So far, it worked fine if both `w`/`h` were enforced or neither.
If one was, then there was a chance one axis would end up 9999
pixel wide or tall. This worked fine when the `.svg` was inserted
using `<img>`, but now that it's using `<object>`, it now scales
down to zero.
The goal is to catch cases where the return value exists, but is
forgotten. There was a large enough number of them to turn this
into a real check. Initially, I just wanted to implement it to fix
the problems, then delete the code. But since this is so common, I
think it is worth the annoyance.
Right now the UML template write many files which were not tracked
by CMake. This caused them to be missing from the doc since the
post-processing was added.