Files uploading into Virtual File System with -s CASE_INSENSITIVE_FS=1

[shinji-TH]

Hi, I did some testing experiment with the flag "-s CASE_INSENSITIVE_FS=1" and notice that the virtual file system allows loading of different files of same name but different caps letters.
i.e. ABC.txt, AbC.txt."

During opening of both files, regardless of "ABC.txt" or "AbC.txt", the open file would always be the last file of the same name that is loaded. In this case is AbC.txt.
I understand this is due to all filenames are converted into lower case names during the lookup of files.

(*edited to make questions more explicit)
I have a few questions with regards to the uploading of files behavior:

1. Why is the case_insensitive flag not applied to uploading file function (function used: "FS_createdDataFile")
2. Is the case sensitive behavior of uploading file function intended within the case_insensitive virtual file system?

[sbc100]

That behaviors makes sense to me. If you have a case insensitive filesystem there is no way to access "ABC.txt" and "AbC.txt" independently, right? Isn't that the definitions of a case insensitive filesystem? What alternative behavior were you imagining/expecting?

[shinji-TH]

Sorry maybe i am not clear on the question of the behavior. The question I have is more on the behaviors of file uploading than accessing the files. i.e. why allowing uploading of multiple files with the same name?

[sbc100]

It seems reasonable to me that is you load two files that map to the same name, the file contents should match the second file that was loaded. The alternative would be perhaps to assert in this case? What behavior would you expect/prefer?

[shinji-TH]

The original expectation was that when the file would be loaded into the file system, it should rename into lowercase or store as original case. However upon loading of second file. It should overwrite the first file instead of writing a new file. It seems weird to me that the case insensitive is applied to the reading/accessing of files in the fs but when uploading of files, we are using case sensitive instead.

[sbc100]

But its sounds like the current behaviour is what you expect. When you open the file under different names "the open file would always be the last file of the same name that is loaded"... it sounds lie the last file essentially clobbers the first file, as expected. I must be misunderstanding the issue :-(

[shinji-TH]

Yea... let me edit the last part to make questions more clearer

[shinji-TH]

Hi, any one happen to know the answers? Please let me know thanks!!

[sbc100]

I'm afraid I still don't really understand what the problem is. Perhaps @juj who added this option will understand.

[shinji-TH]

@juj any comments?
I just wish to know is there any plans for improvement on the file uploading function to prevent the any files with same names being uploaded when the case-insensitive flag is turn on.
The current system allows multiple of the same name to be uploaded but only recognize the latest file that is uploaded. Will this not causes some form of memory bloated/leak? From a users perspective, I could write a document that is 1gb and store into the file system as Abc.txt. Then I made some adjustments into the document, however this time I forgot the exact name and decided to name it as aBc.txt(since its a case-insensitive system). Then the file system will now stored 2 documents with 1gb each. Will this cause any issue to the filesystem if I continue to do multiple such edits?

[juj]

• Why is the case_insensitive flag not applied to uploading file function (function used: "FS_createdDataFile")
• Is the case sensitive behavior of uploading file function intended within the case_insensitive virtual file system?

Indeed that does sound like a bug. The file creation should also be case insensitive, so that there won't ever exist duplicate files of the same case insensitive name. There has never been an intentional feature/reason to do it the way it currently behaves.

[shinji-TH]

@juj, Thanks you for the clarification.
Just to close up the loop.
Is there any intentions on confirming this as bug and plans to implement the fix

[juj]

Yes, that is certainly a bug. Unfortunately I am not able to work on this, but patches are always most welcome!

[AlexWalsh2]

Hello, I would like to start working on this issue if it's still available, could it please be assigned to me?