Those fields have a slightly different meaning on each object that
extends DownloadableContribution and having them grouped in
DownloadableContribution only increase confusion in change of a
(very) tiny code reuse.
Moreover:
- the `readOnly` field has been renamed to `builtIn`
- predicates have been replaced by lambdas
- DownloadableContributionBuiltInAtTheBottomComparator has been replaced
with a singleton instance
- replaced the logic to check if an installed-library is a
builtin library by reusing the same method available in
ContributedLibraryReleases
- renamed some local vars to better reflect their contents:
uninstalledLibraries -> notInstalled
uninstalledNewerReleases -> notInstalledNewer
uninstalledPreviousReleases -> notInstalledPrevious
This class allows to attach a Location property to a folder
path, this way we directly know if a library is in the sketchbook,
core, referenced-core or bundled in the IDE.
This simplify a lot of logic in the IDE.
ContributedLibrary is used to decode library_index.json and it's
intended to keep data coming only from the index.
Now, when the library_index is synced with the filesystem, the
metadata about installed libraries are kept in a separate list
to not mess up with the main index.
Previously the filtering was made on ContributedLibrary, but the
objects that is actually displayed is a ContributeLibraryReleases,
so it comens natural to filter on this class of objects.
Apparently the functionality is not affected by this commit, but
there may be some side-effect that may probably fix some
visualization bug.
Optional<T> helps to not forget to check about nullness where it is
needed.
This commit should be equivalent and shouln't fix any bug, BTW the
Optional<T> semantic turns out to be useful in the next commits.
Possibly all nullable values will be replaced by Optional in the
future.
- we ensure that there is always a filter selected, no more need
to check for non-nullness
- the filters are always the same, no need to pass a
Stream<Predicate<T>> around where Predicate<T> suffices
User testing on skilled devs showed that "buttons appear on click" behaviour is far from being understood.
Accessibility features (like moving with Arrow keys) should be untouched.
This commit makes the Preferences dialog more usable by accessibility devices like screen readers.
Previously, a screen reader would only read the content of the text boxes. By using JLabel's setLabelFor() feature (and in some extreme cases, setAccessibleName()), a screen reader will know that a label is assigned to a text field. For example, instead of just reading "12", it will now read "Editor font size: 12".
This commit makes this changes:
- SplashScreenHelper is now local in Base constructor
- if SplashScreenHelper is instantiated with a null SplashScreen
instance then it outputs progress in console and avoid to make
calls to Swing toolkit
- The parsing of command line arguments is anticipated so we can
determine if we are in command line or GUI mode early and setup
objects that produces output to not use graphics toolkits.
- In this case the SplashScreenHelper is initialized with a real
splashscreen only if we are in GUI mode
The snippet:
boolean wrapNeeded = false;
if (wrap && nextIndex == -1) {
// if wrapping, a second chance is ok, start from the end
wrapNeeded = true;
}
Can be moved inside the `if (nextIndex == -1)` that follows, this way:
if (nextIndex == -1) {
boolean wrapNeeded = false;
if (wrap) {
// if wrapping, a second chance is ok, start from the end
wrapNeeded = true;
}
[...CUT...]
if (wrapNeeded) {
nextIndex = backwards ? text.lastIndexOf(search) : text.indexOf(search, 0);
}
}
but since `wrapNeeded` is used only at the very end of the `if` statement
we can move it forward:
if (nextIndex == -1) {
[...CUT...]
boolean wrapNeeded = false;
if (wrap) {
// if wrapping, a second chance is ok, start from the end
wrapNeeded = true;
}
if (wrapNeeded) {
nextIndex = backwards ? text.lastIndexOf(search) : text.indexOf(search, 0);
}
}
and finally simplify it by removing `wrapNeeded` altogether:
if (nextIndex == -1) {
[...CUT...]
if (wrap) {
nextIndex = backwards ? text.lastIndexOf(search) : text.indexOf(search, 0);
}
}
The snippet:
boolean wrapNeeded = false;
if (wrap && nextIndex == -1) {
// if wrapping, a second chance is ok, start from the end
wrapNeeded = true;
}
is present on both sides of the `if` statement so it can be factored out.
When searching through all tabs, the order was accidentally reversed.
This was broken by commit d2bac86 (Remove tab switching logic from
Sketch).
This also fixes a problem where "replace all" would only work on the
first and last tab (since it would search backwards from the first tab
to the last tab and then conclude it was done).
This fixes a part of #5380.
`EditorTab.setText()` now already preserves the caret position. The code
used during auto-format tried a bit harder to preserve the position
correctly, and probably worked better in a few specific cases, but for
most cases they would both end up approximating the caret position
anyway. To make the code simpler, better just stick to the simpler
approach.
When completely replacing all text, all highlights will end up at the
start of the file. Since keeping them at the right place is tricky (even
impossible in some circumstances), just remove them now.
This already happened in the autoformat code, so that part can be
removed.
