Bug 457828 - GIT History not visible with GTK3
Summary: GIT History not visible with GTK3
Status: CLOSED DUPLICATE of bug 446075
Alias: None
Product: Platform
Classification: Eclipse Project
Component: SWT (show other bugs)
Version: 4.5   Edit
Hardware: PC Linux
: P3 normal with 1 vote (vote)
Target Milestone: ---   Edit
Assignee: Platform-SWT-Inbox CLA
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: 441566
  Show dependency tree
 
Reported: 2015-01-19 08:17 EST by Robert Roth CLA
Modified: 2015-08-10 13:10 EDT (History)
5 users (show)

See Also:


Attachments
Screenshot showing the invisible GIT history (92.13 KB, image/png)
2015-01-19 08:17 EST, Robert Roth CLA
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Robert Roth CLA 2015-01-19 08:17:21 EST
Created attachment 250040 [details]
Screenshot showing the invisible GIT history

GIT History doesn't show anything in the history table on eclipse mars M4.

Steps to reproduce:
1. clone a git repository
2. import a project from the cloned git repository into the workspace
3. show the history of the project (context menu->team->show in history)

Actual results:
history table in the history view does not show any text, but you can select any rows, and the details for that commit appear below the history list, in the commit details panels (commit info + affected files panels)

Expected results:
history table in the history view shows the commit IDs, commit messages, and other commit info in the table, and the commit information below that in the commit details panels

Build and platform information:

Eclipse for RCP and RAP Developers
Version: Mars Milestone 4 (4.5.0M4)
Build id: 20141217-1924
Reproducible with EGit 3.6.0 and EGit 3.7.0.201501171955 (from the nightlies update site)
Running on Fedora 21, eclipse is using SWT+GTK3

Additional information:
* SVN history works fine (via Subclipse), thus it looks like related to EGit history only
* A Gtk Treeview seems to be overlayed over the git history, so this might be an SWT problem. This can be checked using GTK+ inspector (select a row in the history, Ctrl+Shift+I to open inspector) by adding ".view { background-color:alpha (#FFF, 0.2); }" custom CSS to the GtkTreeView.
Comment 1 Browser Ia CLA 2015-04-02 12:48:08 EDT
Also using Fedora 21, with the exact same problem. If you switch to GTK2 rendering, the list is visible as always, only GTK3 exhibits this behavior.

Under M5, the commit list seemed to be drawn in "invisible ink" most of the time, but there were some exceptions where I could get it to render under GTK3: when there were a small number of commits (browsing a repo that had only 4 total commits), or if you drilled down to an individual's file history where there only a small number of changes, and I think on one occasion when I resized the window just right. None of these are now possible on M6. On M5, when you could get the commit list to render, sometimes the text/labels seemed to be drawn twice, on slightly offset from the other.

Please elevate to critical. The history view is unusable under GTK3 on Fedora 21 as it exists in M6/M5. Falling back to GTK2 is not an option when using the Dark theme (also unusable, but for different reasons).

Note: Might be related to Bug 421127 - [GTK3] All view empty with GTK 3.10
Comment 2 Andrey Loskutov CLA 2015-04-02 13:13:52 EDT
Please check if this is a duplicate of bug 458629 or bug 458633?
Comment 3 Browser Ia CLA 2015-04-02 14:20:29 EDT
(In reply to Andrey Loskutov from comment #2)
> Please check if this is a duplicate of bug 458629 or bug 458633?

Yes this is a duplicate of both of these. The thing that bothers me is Comment 42 for Bug 446075 (which has been marked as a duplicate of Bug 458269) says that this has been "Verified" (I presume that means, verified _fixed_) in 4.5.0.I20150318-2000. But isn't M6 4.5.0.I20150326-0704? Which means this fix is present in M6?

M6 seems to exhibit a lot more UI issues with missing/invisible text on more dialogs (not just the Git History View anymore), if anything the problem is spreading, not resolved.

Finally, shouldn't this bug, as well as Bug 458633, be marked as "Depends on" Bug 441566?
Comment 4 Snjezana Peco CLA 2015-04-02 18:35:35 EDT

*** This bug has been marked as a duplicate of bug 446075 ***
Comment 5 Arun Thondapu CLA 2015-04-06 04:27:49 EDT
(In reply to Browser Ia from comment #3)
> (In reply to Andrey Loskutov from comment #2)
> > Please check if this is a duplicate of bug 458629 or bug 458633?
> 
> Yes this is a duplicate of both of these. The thing that bothers me is
> Comment 42 for Bug 446075 (which has been marked as a duplicate of Bug
> 458269) says that this has been "Verified" (I presume that means, verified
> _fixed_) in 4.5.0.I20150318-2000. But isn't M6 4.5.0.I20150326-0704? Which
> means this fix is present in M6?

FYI, M6 is 4.5.0.I20150320-0800 and does not include the fix for bug 446075, the earliest I-build which includes this fix is I20150331-0800, please verify using the same, thanks!