Regression: Line Numbers are broken in different ways on Linux/MacOS

[BeckerWdf]

Let's make sure issue is not already fixed in latest builds first.

• I verified I can reproduce this issue against latest Integration Build of Eclipse SDK

Steps to reproduce

From a fresh installation and clean workspace:

I tried

• Open a Java Class and scroll up and down slowly

I expected: line number to be sharp

But got: partly blurry line numbers at the top or the bottom. See

I think this is a regression from #2722

Tested under this environment:

• OS & version: macOS
• Eclipse IDE/Platform version (as shown in Help > About): newest i-build

Community

• I understand reporting an issue to this OSS project does not mandate anyone to fix it. Other contributors may consider the issue, or not, at their own convenience. The most efficient way to get it fixed is that I fix it myself and contribute it back as a good quality patch to the project.

[BeckerWdf]

@akoch-yatta and @HeikoKlare can you please have a look?

[akoch-yatta]

We have a look

[iloveeclipse]

OMG. I see on Linux line numbers completely broken after scrolling up/down few times:

This must be fixed ASAP.

[HeikoKlare]

Looks to me as if it might got worse with the recent changes, but it seems like this is long broken on GTK. I just tested with a 1,5-year-old SDK and got this:
Version: 2023-06 (4.28)
Build id: I20230605-0440

[iloveeclipse]

Whatever issues your I20230605-0440 version has is not related to recent regression. It is not the first time the line number ruler got a regression.

It works fine with reverted 57666c4 and is broken with current master.

So please provide a fix or we should revert 57666c4 as it makes line number ruler on Linux unusable.

[HeikoKlare]

Sure, we'll do that. But we need to understand the reasons for the issue and actually all SDKs I am running on Linux (no matter whether the 2023-06 one, a recent but older I-Build one, or any other) show this behavior to a certain degree. I am able to produce broken line number rulers in all of them. So it's currently quite hard to understand what part of the behavior needs to be fixed, given that it was (at least on some environments) also broken before yesterday's change to a certain degree.
But as said, we are of course investigating the issue and will come up with a fix.

Edit: In particular if I revert 57666c4, the issue is not fixed on my system but just occurs less often.

[iloveeclipse]

all SDKs I am running on Linux (no matter whether the 2023-06 one, a recent but older I-Build one, or any other) show this behavior to a certain degree

Ugh. I'm using RHEL 9.2 and used to run RHEL 7.9 with almost every nightly SDK build till 4.35 broke compatibility to older Linux distros and I saw no issues with line number ruler (except few regressions clearly identified as such like this one here).

I'm using gtk3-3.24.31-2.el9.x86_64 on RHEL 9.2 / KDE 5 / X11. Not sure if you are using Ubuntu or Wayland, both of them known to have troubles with SWT and together probably unusable for Eclipse anyway.

What I clearly see is that 57666c4 commit breaks status quo and I see no problems with it reverted or in SDK build I20250119-1800.

[HeikoKlare]

Good point. I am running Ubuntu with Wayland. I will test with X11 to hopefully have a state that worked before the commit introducing the regression as a baseline for comparison.

Just as a "demonstration", this is how it looks with 57666c4 reverted for me:

[iloveeclipse]

OK, let take Wayland out of equation. On X11 Eclipse must work.