code minings drawn more consistently always via drawAsLeftOf1stCharacter #2243

tobiasmelcher · 2024-09-05T13:28:59Z

this improves the selection behavior for the scenario where the selection end offset contains a LineContentAnnotation. Selection is no longer extended with the code mining.

The selection behavior before this change always included the code mining text which is located at the end selection offset. The following screenshot shows the behavior. By including the last character "o" of "echo" in the selection, the selection is extended with the code mining text next to the "o" character.

This is no longer the case with this change. The code mining text is no longer added to the selection when "echo" is selected.

Is the current behavior of drawing the code mining texts intended? Is this behavior needed for a feature (which I am not aware of)?

Let me try to explain why I request this change.
In the ABAP Development Tools in Eclipse we use the code minings to show inline keyword completions while the user is typing a keyword. If a matching keyword is available, it is shown as code mining and user can take over the text by pressing the tab key. Here a screencast showing the scenario when the user is typing the keyword "definition".

cursor_jumps.mp4

It works quite well. Users don't like the way how the cursor always jumps to the end of the code mining showing the rest of the keyword "definition". The cursor jumps to the end of the code mining because method drawAsRightOfPreviousCharacter is used to render the code mining.

The cursor does not jump to the end of the code mining with this pull request by using method drawAsLeftOf1stCharacter to render the code mining.

cursor_does_not_jump.mp4

The question now is if we would break any existing functionality regarding rendering of code minings by always using method drawAsLeftOf1stCharacter?
An alternative approach would be to provide an API with which the user can decide how to draw the code mining. What would be your preferred solution? Could you please give any guidance?

Thanks a lot,
Tobias

github-actions · 2024-09-05T13:40:44Z

Test Results

1 815 files ±0 1 815 suites ±0 1h 39m 32s ⏱️ + 5m 21s
7 702 tests +2 7 461 ✅ +1 228 💤 ±0 2 ❌ +1 11 🔥 ±0
24 267 runs +6 23 507 ✅ +5 747 💤 ±0 2 ❌ +1 11 🔥 ±0

For more details on these failures and errors, see this check.

Results for commit 8c67165. ± Comparison against base commit f9b743f.

♻️ This comment has been updated with latest results.

BeckerWdf · 2024-09-05T13:49:57Z

"pr-head" says:
Failed to execute goal org.eclipse.tycho.extras:tycho-p2-extras-plugin:4.0.8:compare-version-with-baselines (compare-attached-artifacts-with-release) on project org.eclipse.jface.text: Only qualifier changed for (org.eclipse.jface.text/3.25.200.v20240905-1304). Expected to have bigger x.y.z than what is available in baseline (3.25.200.v20240816-1151)

Pls add a second mitt that increases the version of org.eclipse.jface-text to "3.25.300". This should fix this issue.

BeckerWdf · 2024-09-06T06:31:12Z

There seems to be an infrastructure issue:

06:14:56.854 [ERROR] Error executing Maven.
06:14:56.855 [ERROR] Extension org.eclipse.tycho:tycho-build:4.0.6 or one of its dependencies could not be resolved: Failed to read artifact descriptor for org.eclipse.tycho:tycho-build:jar:4.0.6
06:14:56.856 [ERROR] Caused by: Failed to read artifact descriptor for org.eclipse.tycho:tycho-build:jar:4.0.6
06:14:56.856 [ERROR] Caused by: Failed to read artifact descriptor for org.eclipse.tycho:tycho-build:jar:4.0.6
06:14:56.856 [ERROR] Caused by: The following artifacts could not be resolved: org.eclipse.tycho:tycho-build:pom:4.0.6 (absent): Could not transfer artifact org.eclipse.tycho:tycho-build:pom:4.0.6 from/to eclipse.maven.central.mirror (https://repo.eclipse.org/content/repositories/maven_central/): status code: 403, reason phrase: Forbidden (403)
06:14:56.856 [ERROR] Caused by: Could not transfer artifact org.eclipse.tycho:tycho-build:pom:4.0.6 from/to eclipse.maven.central.mirror (https://repo.eclipse.org/content/repositories/maven_central/): status code: 403, reason phrase: Forbidden (403)
06:14:56.857 [ERROR] Caused by: status code: 403, reason phrase: Forbidden (403)

BeckerWdf · 2024-09-12T08:06:28Z

@mickaelistria: What are your thoughts on this PR?

BeckerWdf · 2024-09-13T07:59:15Z

org.eclipse.jface.text.tests also needs a version bump:

Failed to execute goal org.eclipse.tycho.extras:tycho-p2-extras-plugin:4.0.8:compare-version-with-baselines (compare-attached-artifacts-with-release) on project org.eclipse.jface.text.tests: Only qualifier changed for (org.eclipse.jface.text.tests/3.13.600.v20240913-0631). Expected to have bigger x.y.z than what is available in baseline (3.13.600.v20240816-1151) -> [Help 1]

Can you pls. ammend your version change commit?

BeckerWdf · 2024-09-13T08:01:19Z

Some of the code mining tests are failing:

org.eclipse.jface.text.tests.codemining.CodeMiningTest.testCodeMiningOnEmptyLine
org.eclipse.jface.text.tests.codemining.CodeMiningTest.testCodeMiningAtEndOfLine

BeckerWdf · 2024-09-13T13:52:18Z

org.eclipse.jface.text.tests also needs a version bump:

Failed to execute goal org.eclipse.tycho.extras:tycho-p2-extras-plugin:4.0.8:compare-version-with-baselines (compare-attached-artifacts-with-release) on project org.eclipse.jface.text.tests: Only qualifier changed for (org.eclipse.jface.text.tests/3.13.600.v20240913-0631). Expected to have bigger x.y.z than what is available in baseline (3.13.600.v20240816-1151) -> [Help 1]

Can you pls. ammend your version change commit?

Thanks

mickaelistria · 2024-09-13T14:10:01Z

I didn't try this PR yet.
But the history here is that the main use-case for inline code minings in the Eclipse SDK are currently method parameter annotation. With those, user want for example System.arraycopy((src: array1, srcPos: |0, dest: array2, destPos: 0, n). If we change it the other way round, then we break this case, the cursor before 0 would be ~80 pixels to the left. Both this legacy use-case and the one you're sharing here seem antagonist. We most likely need some extra flag on codeMinings somewhere to declare whether it's a prefix or a suffix.

tobiasmelcher · 2024-09-13T14:32:09Z

Thanks a lot for the feedback @mickaelistria. I fully agree. I just tested the java method parameter code minings with this change and selection behavior looks ugly.

Without this change, one can select the method parameter value 0 and the code mining text is not included.

With this change, this is no longer possible. Selecting 0 always includes the code mining which does not feel right.

I will make a proposal with an API enhancement so that the selection behavior of the code minings can be configured. For sure in a compatible way.

BeckerWdf · 2024-09-13T14:34:08Z

I will make a proposal with an API enhancement so that the selection behavior of the code minings can be configured. For sure in a compatible way.

Yes. That's a good idea.

mickaelistria

I have not tested it, but the code looks already good.
Just a comment inline discussing whether a new CodeMining type is useful here.

mickaelistria · 2024-09-30T09:45:54Z

...rg.eclipse.jface.text/src/org/eclipse/jface/text/codemining/SuffixLineContentCodeMining.java

+ * Line content code mining which is drawn right to the given position. Cursor and selection
+ * handling are not including the code mining.
+ */
+public abstract class SuffixLineContentCodeMining extends LineContentCodeMining {


I would rather imagine a new flag on LineContentCodeMning than a whole new method here.
IMO, a new constructor LineContentCodeMining(Position, boolean afterPosition /* default being false in other constructors that don't have this arg */, provider) would be simpler to both consume and maintain.

Thanks a lot for the fast reply. I changed the API as requested and added a further constructor in LineContentCodeMining. I have put "boolean afterPosition" as last contructor argument for compatibility reasons. Do you think it is better to place it as second argument?

For compatibility, you only need to keep the current constructors. You're free to design your new ones freely.
So yes, I think it's better to place it as 2nd argument if it's not conflicting with any other existing constructor.

ok, I added it as 2nd argument. Is it because first argument is "position" and next argument "afterPosition" is then influencing the position?

Yes, the position where the codemining is rendered is a function of those 2 arguments. Grouping them together makes things faster to understand.

ok, understood. Makes absolutely sense.

examples/org.eclipse.jface.text.examples/META-INF/MANIFEST.MF

mickaelistria

Some notes about probable complaints from API tools about new API methods.

mickaelistria · 2024-09-30T13:06:45Z

bundles/org.eclipse.jface.text/src/org/eclipse/jface/text/codemining/LineContentCodeMining.java

+	 * @param afterPosition if true code mining is treated as suffix code mining where cursor and
+	 *            selection is not including the mining
+	 */
+	public LineContentCodeMining(Position position, boolean afterPosition, ICodeMiningProvider provider, Consumer<MouseEvent> action) {


Since this is new API, this will likely require some @since 3.26 and a version increase in the MANIFEST.

mickaelistria · 2024-09-30T13:08:35Z

.../org.eclipse.jface.text/src/org/eclipse/jface/text/source/inlined/LineContentAnnotation.java

@@ -158,7 +158,7 @@ StyleRange updateStyle(StyleRange style, FontMetrics fontMetrics, ITextViewer vi
 		return style;
 	}

-	static boolean drawRightToPreviousChar(int widgetOffset, StyledText textWidget) {
+	protected boolean drawRightToPreviousChar(int widgetOffset, StyledText textWidget) {


Since this is new API, this will likely require some @since 3.26 and a version increase in the MANIFEST.
Or better if you can find a way to not make this method API.

thanks a lot for the help. I tried to fix the API tools messages with 80b1736. I added an additional argument to package method LineContentAnnotation#updateStyle . Is this allowed? If not, I would need to introduce method "protected boolean drawRightToPreviousChar" as protected instance method.

mickaelistria · 2024-09-30T14:49:49Z

Is this allowed?

Yes, it is allowed because this method is package visible only (not API then).

mickaelistria · 2024-09-30T14:55:16Z

bundles/org.eclipse.jface.text/src/org/eclipse/jface/text/codemining/LineContentCodeMining.java

+	 *
+	 * @since 3.26
+	 */
+	protected boolean afterPosition= false;


As there is a getter, couldn't this one be private final too?

I fully agree, thanks for the feedback.

which allows to render a code mining where the cursor selection does not include the code mining at the given source position.

mickaelistria

Code looks very good and the use-case is pretty interesting. Let's merge it as soon as CI reports a successful build on Jenkins.

BeckerWdf · 2024-10-01T06:28:53Z

failed test are unrelated to this PR

mickaelistria · 2024-10-01T06:52:47Z

Thank you for this good code and for making the IDE capable of achieving 1 new use-case!

tobiasmelcher requested a review from mickaelistria September 5, 2024 13:48

tobiasmelcher force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch 3 times, most recently from 371ac78 to 5f67aff Compare September 6, 2024 06:13

BeckerWdf force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch from 5f67aff to 28f94ae Compare September 12, 2024 08:05

BeckerWdf force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch from 28f94ae to 8c61750 Compare September 13, 2024 06:31

tobiasmelcher force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch 2 times, most recently from c0f3414 to 0b85447 Compare September 13, 2024 13:48

BeckerWdf assigned tobiasmelcher Sep 23, 2024

tobiasmelcher force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch from 0b85447 to fdabddd Compare September 30, 2024 09:36

mickaelistria reviewed Sep 30, 2024

View reviewed changes

tobiasmelcher force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch from fdabddd to fbdbd21 Compare September 30, 2024 11:58

BeckerWdf reviewed Sep 30, 2024

View reviewed changes

examples/org.eclipse.jface.text.examples/META-INF/MANIFEST.MF Outdated Show resolved Hide resolved

increase version numbers for next release

dfb9e06

tobiasmelcher force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch from fbdbd21 to 46e6ea3 Compare September 30, 2024 12:56

mickaelistria reviewed Sep 30, 2024

View reviewed changes

tobiasmelcher force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch from 46e6ea3 to 80b1736 Compare September 30, 2024 14:34

mickaelistria reviewed Sep 30, 2024

View reviewed changes

additional constructor param afterPosition in LineContentCodeMining

8c67165

which allows to render a code mining where the cursor selection does not include the code mining at the given source position.

tobiasmelcher force-pushed the codeMiningsAlwaysDrawnAsLeftOf1stCharacter branch from 80b1736 to 8c67165 Compare September 30, 2024 15:00

mickaelistria approved these changes Sep 30, 2024

View reviewed changes

BeckerWdf added this to the 4.34 M2 milestone Oct 1, 2024

mickaelistria merged commit 9b21618 into eclipse-platform:master Oct 1, 2024
14 of 17 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

code minings drawn more consistently always via drawAsLeftOf1stCharacter #2243

code minings drawn more consistently always via drawAsLeftOf1stCharacter #2243

tobiasmelcher commented Sep 5, 2024 •

edited

Loading

github-actions bot commented Sep 5, 2024 •

edited

Loading

BeckerWdf commented Sep 5, 2024

BeckerWdf commented Sep 6, 2024

BeckerWdf commented Sep 12, 2024

BeckerWdf commented Sep 13, 2024

BeckerWdf commented Sep 13, 2024

BeckerWdf commented Sep 13, 2024

mickaelistria commented Sep 13, 2024

tobiasmelcher commented Sep 13, 2024

BeckerWdf commented Sep 13, 2024

mickaelistria left a comment

mickaelistria Sep 30, 2024

tobiasmelcher Sep 30, 2024

mickaelistria Sep 30, 2024

tobiasmelcher Sep 30, 2024

mickaelistria Sep 30, 2024

tobiasmelcher Sep 30, 2024

mickaelistria left a comment

mickaelistria Sep 30, 2024

mickaelistria Sep 30, 2024

tobiasmelcher Sep 30, 2024

mickaelistria commented Sep 30, 2024

mickaelistria Sep 30, 2024

tobiasmelcher Sep 30, 2024

mickaelistria left a comment

BeckerWdf commented Oct 1, 2024

mickaelistria commented Oct 1, 2024

code minings drawn more consistently always via drawAsLeftOf1stCharacter #2243

code minings drawn more consistently always via drawAsLeftOf1stCharacter #2243

Conversation

tobiasmelcher commented Sep 5, 2024 • edited Loading

github-actions bot commented Sep 5, 2024 • edited Loading

Test Results

BeckerWdf commented Sep 5, 2024

BeckerWdf commented Sep 6, 2024

BeckerWdf commented Sep 12, 2024

BeckerWdf commented Sep 13, 2024

BeckerWdf commented Sep 13, 2024

BeckerWdf commented Sep 13, 2024

mickaelistria commented Sep 13, 2024

tobiasmelcher commented Sep 13, 2024

BeckerWdf commented Sep 13, 2024

mickaelistria left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mickaelistria left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mickaelistria commented Sep 30, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mickaelistria left a comment

Choose a reason for hiding this comment

BeckerWdf commented Oct 1, 2024

mickaelistria commented Oct 1, 2024

tobiasmelcher commented Sep 5, 2024 •

edited

Loading

github-actions bot commented Sep 5, 2024 •

edited

Loading