As noted in https://jira.transvar.org/browse/IGBF-2608, This behavior is also found in version 9.1.4.

Two problems on Windows 10:

Saving a track with the same name and in the same place as an existing saved track will not result in a pop-up error message like it should, but will overwrite the original file.

Saving a track with the same name and in the same place as an existing saved track but with the file type extension included in the name, like FileName.bed, will result in the appropriate pop-up error message. But clicking the "Yes, replace the file" option will result in a new file being created with two file type extensions in the name, like FileName.bed.bed, and will not overwrite the original file.

Release Testing on Mac M1.
This problem persists in 9.1.8 and 9.1.10 during release testing. Though the "second" problem from my previous comment: "Yes, replace file" is fixed now when the pop-up does appear.

A pop-up asks to replace the file in the case of bookmarks. For example a bookmark.html file, but not for a track.bed file.

To reproduce:
1. Reset preferences by going to File > Preferences > Other Options > Reset Preferences to Defaults
2. Open genome A_Thaliana Jun_2009
3. Choose Araport track.
4. Right click, then “Save track as…”
5. Save as name: “test” with file type: bed to the desktop
6. Repeat:
7. Choose Araport track.
8. Right click, then “Save track as…”
9. Save as: “test” with file type: bed to the desktop
10. Notice no request to replace file.
11. On the desktop, right click test.bed
12. Click Get Info
13. Notice it is Modified today at this time. The file was replaced without asking.

Release testing protocol found at: FileChooser, General Function Checklist. https://wiki.bioviz.org/confluence/display/ITD/FileChooser

8.5.4 - Uses the non-native file chooser. If file name "test" entered then it adds the extension then makes the user click save again, then it checks if other file exists

9.0.0 - Native file chooser used. Entering "test" or "test.bed" both ask if i want to replace file.

9.0.2 - Native file chooser used. Entering "test" the file is overwritten without asking if I want to replace. Entering "test.bed" asks if I want to replace the file. This indicates that the regression took place between 9.0.0 and 9.0.2.

This is a bug that was introduced in IGB 9.0.0 when we refactored to use the native file chooser. In 9.0.0 the behavior is correct when it asks the user if they would like to overwrite the file, but the filename is saved incorrectly as filename.bed.*.bed. This problem was attempted to be fixed but unfortunately the fix broke the logic for checking if the file already exists.

Issue seems to be in the following files:

AbstractExportFileAction - saveAsFile ->Look at line 96
FileChooserUtil

Other files that use FileChooserUtil.java -> saveFilesFromFxChooser:

AbstractExportFileAction.java
WebLInksView.java
BookmarkManagerView.java
ExportDialog.java
SaveSessionAction.java

The issue around saving files is related to many other tickets (see IGBF-1128) of which this ticket is effectively a clone of IGBF-1410.

In AbstractExportFileAction.java in the saveAsFile method there is a call to make a new FileChooser.ExtensionFilter. According to the official Java documentation the way the filter works is by including a wildcard "*" in front of the desired filetypes (i.e. *.png). However, this wildcard was removed as part of IGBF-1128. This was done because the wildcard appeared to be added to the file extension when saved, causing issues. However, it appears to have had the unintended consequence of causing the Save File dialog box to behave incorrectly and fail to check if a file already exists.

To fix the issue the wildcard needed to be included in AbstractExportFileAction.java. To fix the issue with the file extension, I removed what appeared to be an unneeded addition of the selectedExtension to the file path. There could be an edge case for why this logic was included, but after testing I have been unable to determine why it was there.

See my previous comment regarding other classes that use FileChooserUtil.java. We will need to test saving both annotation and graph tracks, sessions, bookmarks, and weblinks. In addition, there was an issue with the file extensions appearing twice on windows that may also be related (IGBF-1162).

Commit: https://bitbucket.org/nfreese/nowlanfork-igb/commits/e9392d022999e03423d450a018d97e136877ba8e

In addition, I also found a new bug where the selection of the output file type (bed or bed detail) does not appear to be consistent. Bed detail is sometimes saved when bed is selected and vice versa. I have created IGBF-3203 to fix this issue.

Testing with Logan Weidenhammer we identified an existing issue (see IGBF-1184) with weblinks that is related to this issue. I was able to fix the issue in the following commit. Because the issues are closely related I think we can include it under 2654 as its own commit (i.e. we should not squash them together).

https://bitbucket.org/nfreese/nowlanfork-igb/commits/8653e64c435b4f21e875a8713844871e0e6b8de5

The two commits have been reviewed.
Tested file saving for annotation tracks, graph tracks, sessions, bookmarks, weblinks, and image saving on Windows and Mac, 9.1.8 and 9.1.10.

Dr. Freese fixed the existing weblinks issue where not choosing a file type would result in the file being replaced without asking.

We wanted to note a previous issue on 9.1.8 where Windows would save double file extensions is resolved in 9.1.10, with the caveat that Windows' file saver still shows a double extension when in the process of saving a track. This does not actually cause the file to be saved with multiple extensions. This is not a regression in this new release.

Recommending Pull Request.

Pull request: https://bitbucket.org/lorainelab/integrated-genome-browser/pull-requests/916/igbf-2654

Merged. Installers built for "master" branch. I have notarized the MacOS installer master.dmg and uploaded it to repository downloads section. I have also deployed the master branch installers to the IGB "early access" section of the bioviz.org web site.

Ready for final testing.

I'm sorry I didn't catch this on the pre-merge testing.
Windows allows saving sessions without a visible file extension, but IGB still understands what to do with it.

I can choose to save a session, backspace over the ".xml" in the suggested file name (with XML indicated as the 'type' in the windows native file chooser, and then save it. It is saved without an extension as a generic file.
Then when loading a session, the file does not show up in IGB Load-Session file chooser (issue present in 9.1.8 also), but you CAN load the saved session normally if you type in the file name.

For example: files named savedSession and savedSession.xml are treated as unique files. Though IGB can still load the savedSesson (no extension) file normally.

On Windows IGB 9.1.8 this particular issue is not present.

All other types of saving worked normally.

List of files that call FileChooser.ExtensionFilter:

AbstractExportFileAction.java - fixed	
RunScriptAction.java	- fixed
SynonymsControlPanel.java	- fixed
CustomGenomeDialogPanel.java - fixed
WebLinksView.java	-fixed
BookmarkManagerView.java	-fixed
LoadSessionAction.java	-fixed
SaveSessionAction.java	-fixed

List of files that call saveFilesFromFxChooser

AbstractExportFileAction.java	-fixed
WebLinksView.java	-fixed
BookmarkManagerView.java	-fixed
ExportDialog.java	
SaveSessionAction.java	-fixed

Need to work on ExportDialog.java

I've pushed a new commit to my 2654 branch that addresses an issue with sessions: https://bitbucket.org/nfreese/nowlanfork-igb/commits/4b2ae9ca13c4116b981bafb4a2eb72fa14929064

I have fixed the ExportDialog.java file extension.

Commit: https://bitbucket.org/nfreese/nowlanfork-igb/commits/7a6a8d907c3cc0be92114b93f51baf88d0c64462

Commits reviewed.

I tested on Windows saving images of all 3 file types as all 4 different views both with an without file extensions. All of them saved with the appropriate extensions, no files were created without extensions, and no files were created with double extensions.

Recommending Pull Request.

Pull request: https://bitbucket.org/lorainelab/integrated-genome-browser/pull-requests/917/igbf-2654

PR is merged and new master branch installers are built. Apple installer is notarized and uploaded to Downloads section of the IGB team repository. New installers are deployed to the "early access" section of the BioViz.org Web site.

Ready for testing.

Post-merge testing is complete. I tested the same functions as my previous comment as well as ran through the Image Export subsystem tests. Everything is working as intended.
Moving to done.