I want Japanese mark of a voiced sound in speech characters to be unified with NFC.
I want Japanese mark of a voiced sound in speech characters to be unified with NFC.
-Overview-
example: the 'ga' syllabary series in the Japanese alphabet
'ガ'is U+30AC on windows or unix/linux
-> NFC(Normalization Form Canonical Compression)
'ガ'is U+30AB U+3099U on mac OSX
-> NFD(Normalization Form Canonical Decompression)
As you know, mac OSX use only NFD.
Windows use NFC usually, but NFD can be used at the same time.
When trying to share BOX folder between Win and mac, Inconvenience occurs.
-problem-
First
In Windows explorer, NFC and NFD are displayed as the same character on the screen, so they are indistinguishable.
The second
In Adobe InDesign, image files are managed as "link files", but file names are managed by NFC, so upload InDesign data created on mac OS to the shared BOX folder and download it on Windows. Cutting file link will occur.
The third
Where NFC uses one character in UTF-8, NFD uses two characters in UTF-8.
The file "がぎぐげござじずぜぞだぢづでどばびぶべぼぱぴぷぺぽガギグゲゴザジズゼゾダヂヅデドバビブベボパピプペポヴ.txt" can be used on macOSX, but this file cannot be saved on the BOX server by Finder using BOX Drive.
Even if two characters are used, "xx.txt" uses only 106 characters.
And from Windows PC,
The file "がぎぐげござじずぜぞだぢづでどばびぶべぼぱぴぷぺぽガギグゲゴザジズゼゾダヂヅデドバビブベボパピプペポヴ.txt" (By NFC) there will be no problem with BOX Drive.
But, with macOS X / Box Drive, the inside of the BOX server is not displayed in the Finder.
"がぎぐげござじずぜぞだぢづでどばびぶべぼぱぴぷぺぽガギグゲゴザジズゼゾダヂヅデドバビブベボパピプペポヴ.txt"(By NFC) uses only 55 characters in NFC.
Are you having another problem?
-Operation on mac OSX-
When writing to a FAT32 formatted USB memory device on mac OSX, write the file name using NFC.
When uploading data to a shared folder using the cifs protocol on mac OSX, write the file name in NFC format.
When uploading data to a shared folder using the AFP protocol on mac OSX, write the file name in NFC format.
On mac OSX, write file name in NFD format only when writing data to HFS+ format HDD.
mac OSX does not accept any NFC on HFS+, and NFC automatically converts to NFD when downloading data from the BOX server.
-Proposal-
When writing data to the BOX server using BOX Drive for MacOS, please convert it to NFC instead of NFD and write the data.
When uploading to the BOX server, no one will be troubled even if it is converted to NFC instead of NFD and saved. Inconvenience has occurred rather than being saved as NFD.
-Additional explanation-
mark of a voiced sound in speech characters is below.
がぎぐげご
ざじずぜぞ
だぢづでど
ばびぶべぼ
ぱぴぷぺぽ
ガギグゲゴ
ザジズゼゾ
ダヂヅデド
バビブベボ
パピプペポ
ヴ
by NFC
がぎぐげご
ざじずぜぞ
だぢづでど
ばびぶべぼ
ぱぴぷぺぽ
ガギグゲゴ
ザジズゼゾ
ダヂヅデド
バビブベボ
パピプペポ
ヴ
by NFD
-
KIPR
commented
I strongly support this request.
However, I do not think the current thread title fully represents the nature of the issue. In my view, the underlying problem is not specific to Japanese voiced sound marks, but rather Unicode filename normalization interoperability between macOS and Windows.
This issue still exists in current versions of macOS and Box Drive in 2026, and it causes interoperability problems when Box is used collaboratively between macOS and Windows.
I investigated the behavior and found that the issue can be reproduced as follows:
* Creating a file or folder in Finder on the macOS Desktop results in a decomposed (NFD) filename.
* Creating a file or folder in Finder inside Box Drive also results in an NFD filename.
* Creating the same filename from a Unix command-line tool in Terminal on the macOS Desktop preserves the NFC filename.
* Creating the same filename from Terminal directly inside Box Drive also preserves the NFC filename.
* Renaming an existing NFC filename in Finder changes the filename to NFD.These results indicate that Box Drive itself is not converting NFC filenames to NFD. Rather, Finder generates decomposed filenames, and Box Drive uploads those filenames to Box without normalizing them.
Modern APFS does not normalize filenames when they are created; it can preserve either NFC or NFD filenames. Therefore, this is not simply an HFS+ filesystem issue. Finder still generates decomposed filenames, while many Unix-derived command-line tools preserve the supplied Unicode representation.
This creates a cross-platform interoperability problem in Box. macOS itself can generally handle canonically equivalent filenames transparently, but once an NFD filename created by Finder is uploaded to Box, the decomposed filename is exposed to Windows and other platforms.
For example, the Japanese katakana character:
ダ
may be uploaded as two Unicode code points:
U+30BF (タ) + U+3099 (゛; COMBINING KATAKANA-HIRAGANA VOICED SOUND MARK)
instead of the NFC representation:
U+30C0 (ダ)
This is not limited in principle to Japanese voiced and semi-voiced characters; Unicode normalization differences can affect other characters and languages as well.
Because Box is a cross-platform content management and collaboration service, I believe Box Drive for macOS is an appropriate layer at which to resolve this incompatibility.
I therefore request that Box Drive for macOS provide an option to normalize newly created or renamed file and folder names to Unicode NFC before sending their names to the Box service.
Ideally, this should apply when a new Box item name is created through Box Drive, including:
* new files and folders;
* renamed files and folders; and
* new items created by copying files or folders in Finder.Existing Box items should not need to be renamed merely because their contents are edited.
This request was originally submitted in 2020, and the interoperability issue is still reproducible more than six years later. I would appreciate it if Box could reconsider adding NFC normalization support to Box Drive for macOS.