Godot 4.4 get_modified_time fails on Windows system

[jospic]

Tested versions

• Reproducible in Godot 4.4 stable version

System information

Windows 10 - v4.4.stable.official [4c311cb]

Issue description

FileAccess.get_modified_time(file_path) returns "0" on Windows system.

Steps to reproduce

func get_file_modification_time(file_path: String) -> String:
	
	if FileAccess.file_exists(file_path):
		var timestamp = FileAccess.get_modified_time(file_path)    # seems to fail on Windows platform
		var date_time = Time.get_datetime_dict_from_unix_time(timestamp)
		
		var formatted_time = "%02d/%02d/%04d %02d:%02d:%02d" % [
			date_time.day, date_time.month, date_time.year,
			date_time.hour, date_time.minute, date_time.second
		]

		return formatted_time
	else:
		return "File not exists!"

Minimal reproduction project (MRP)

See "Steps to reproduce"

[matheusmdx]

Can't reproduce on my end. When you did the first test was checking if file exists? Any file returns 0? When returns 0 means an error occurred (most of the cases an invalid path).

[bruvzg]

What kind of files are you checking? If you are trying to get modification time of the file with res:// path in the exported project, it won't work since it's part of PCK, not a real file. And PCK does not store modification time (it's read-only).

It should return 0 if:

• path is invalid.
• path is directory, not file.
• you do not have read permissions for the file.
• time is set to value before beginning of Unix epoch (Windows is counting time since January 1, 1601).
• it's not supported by the file system.

[jospic]

The same script works without issues on Godot 4.3 official version.

File exists and path Is an absolute path with a file at the end (in that case a SQLite file)

[jospic]

Can't reproduce on my end. When you did the first test was checking if file exists? Any file returns 0? When returns 0 means an error occurred (most of the cases an invalid path).

The first check, file exists, pass without problem. Path Is valid and real.

[Calinou]

@jospic Can you try following the same steps on 4.4 dev/beta/RCs to determine when the regression started? You can download them here.

[jospic]

@jospic Can you try following the same steps on 4.4 dev/beta/RCs to determine when the regression started? You can download them here.

Hi @Calinou,
ok, I did various tests starting from dev versions up to all the release candidates. I verified that dev 1 does not show the issue, while it is present from dev 2 onwards.

It might be related to the pull "Reduce code duplication in FileAccess ([GH-92167]" from DEV 2 Changelog (#92167)) ?

Thanks in advance.
-j

[bruvzg]

Most likely it is related to #74830, but not sure why it's not working in your case. The only reason I can think of is, the file is already opened for writing with exclusive access, which might be interfering with the CreateFileW.

[jospic]

Most likely it is related to #74830, but not sure why it's not working in your case. The only reason I can think of is, the file is already opened for writing with exclusive access, which might be interfering with the CreateFileW.

I don't know, but I don't think it's due to some blocking problem on the specific file. I have tested on many different files and I confirm that the dev 1 build does not show any problems, unlike the dev 2

[bruvzg]

I don't know, but I don't think it's due to some blocking problem on the specific file.

Blocking definite is an issue, and easily reproducible one. It's possible that there's something else as well, but I can't reproduce it with any other file.