# Ruby on win32 cannot handle certain filenames

**URL:** <https://www.ruby-forum.com/t/ruby-on-win32-cannot-handle-certain-filenames/75888>\
**Category:** Ruby\
**Created:** [November 14, 2006, 1:28am UTC](https://www.ruby-forum.com/t/ruby-on-win32-cannot-handle-certain-filenames/75888 "2006-11-14T01:28:56Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![david](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/aeb1de/32.png) [@david](https://www.ruby-forum.com/u/david)\
**Post date:** [November 14, 2006, 1:28am UTC](https://www.ruby-forum.com/t/ruby-on-win32-cannot-handle-certain-filenames/75888/1 "2006-11-14T01:28:56Z")

</div>

Hello all!

This is my first post in any one the ruby forums 🙂

I have a serious ruby windows problem. When I use IO related calls (such  
as Dir.glob, File.open, etc) on my machine (WinXP), filenames are always  
returned in the shift\_jis charset. I’ve been using iconv to convert to  
utf8 but then I came across a much more serious problem: because  
File.open is using the shift\_jis charset for filenames, it is NOT  
POSSIBLE (!) for Ruby to open files that have say European chars in the  
filename!! In this day and age SURELY it cannot be the case that it’s  
not possible in Ruby! It must be my inexperience 🙂 There must be some  
way that I don’t know about. Any ideas/opinions/suggestions?

Also, I’ve tried changing the KCODE but it has absolutely no effect on  
Dir.glob or Flie.open.

Also, when I used Dir.glob, Japanese filenames worked fine but one file  
that had an ë in it (e with umlat) had been converted somewhere in  
ruby’s internals to just a plain ASCII e. There must be some way to  
disable this internal charset conversion.

Golly

---

<div class="post-metadata">

**Author:** ![david](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/aeb1de/32.png) [@david](https://www.ruby-forum.com/u/david)\
**Post date:** [November 14, 2006, 4:14pm UTC](https://www.ruby-forum.com/t/ruby-on-win32-cannot-handle-certain-filenames/75888/2 "2006-11-14T16:14:47Z")

</div>

On 11/13/06, David B. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> way that I don’t know about. Any ideas/opinions/suggestions?
> 
> Also, I’ve tried changing the KCODE but it has absolutely no effect on  
> Dir.glob or Flie.open.
> 
> Also, when I used Dir.glob, Japanese filenames worked fine but one file  
> that had an ë in it (e with umlat) had been converted somewhere in  
> ruby’s internals to just a plain ASCII e. There must be some way to  
> disable this internal charset conversion.

You’re correct.

This is a limitation of Ruby as it is currently built. I haven’t had  
time to check whether Matz has m17n Strings in, but this will not be  
fixable in Ruby 1.8 without using a Unicode string extension and a  
significant overhaul to the Windows code.

-austin

---

<div class="post-metadata">

**Author:** ![david](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/aeb1de/32.png) [@david](https://www.ruby-forum.com/u/david)\
**Post date:** [November 14, 2006, 5:45pm UTC](https://www.ruby-forum.com/t/ruby-on-win32-cannot-handle-certain-filenames/75888/3 "2006-11-14T17:45:01Z")

</div>

Hi,

At Tue, 14 Nov 2006 09:28:58 +0900,  
David B. wrote in [ruby-talk:224879]:

> I have a serious ruby windows problem. When I use IO related calls (such  
> as Dir.glob, File.open, etc) on my machine (WinXP), filenames are always  
> returned in the shift\_jis charset.

Do you use Japanese version Windows? Those methods use “OEM  
string” but have no shift\_jis specific code.

> File.open is using the shift\_jis charset for filenames, it is NOT  
> POSSIBLE (!) for Ruby to open files that have say European chars in the  
> filename!! In this day and age SURELY it cannot be the case that it’s  
> not possible in Ruby! It must be my inexperience 🙂 There must be some  
> way that I don’t know about. Any ideas/opinions/suggestions?

If your system runs with European 8-bit charset, it should work  
by setting $KCODE to “N”.

> Also, I’ve tried changing the KCODE but it has absolutely no effect on  
> Dir.glob or Flie.open.

$KCODE is for internal use, typically Regexp.

> Also, when I used Dir.glob, Japanese filenames worked fine but one file  
> that had an ë in it (e with umlat) had been converted somewhere in  
> ruby’s internals to just a plain ASCII e. There must be some way to  
> disable this internal charset conversion.

Not ruby’s internals, it’s done in Windows kernel.

---

<div class="post-metadata">

**Author:** ![david](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/aeb1de/32.png) [@david](https://www.ruby-forum.com/u/david)\
**Post date:** [November 14, 2006, 6:27pm UTC](https://www.ruby-forum.com/t/ruby-on-win32-cannot-handle-certain-filenames/75888/4 "2006-11-14T18:27:18Z")

</div>

On 11/14/06, Nobuyoshi N. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> > Also, when I used Dir.glob, Japanese filenames worked fine but one file  
> > that had an ë in it (e with umlat) had been converted somewhere in  
> > ruby’s internals to just a plain ASCII e. There must be some way to  
> > disable this internal charset conversion.  
> > Not ruby’s internals, it’s done in Windows kernel.

Sort of. FindFirstFile is used, not FindFirstFileW (which returns a  
UTF-16 string). The Windows kernel does the translation “safely”, but  
if Ruby used FindFirstFileW (explicitly!), it would work. Sadly, this  
is very difficult without an encoding-aware String.

-austin