Loading ...
Sorry, an error occurred while loading the content.

Re: Unicode conversion bug?

Expand Messages
  • Tony Mechelynck
    ... The BOM is a valid codepoint, ZERO-WIDTH NO-BREAK SPACE; however its use in that capacity is now deprecated (ZERO-WIDTH NON-JOINER is preferred IIUC).
    Message 1 of 25 , May 1, 2008
    • 0 Attachment
      On 02/05/08 03:29, T.P.S.Nakagawa wrote:
      > This is Nakagawa, Japanese user on unix and Windows XP SP2.
      >
      > ( Windows Japanese version's native character set is cp932 )
      >
      >
      > I found same problem 30 April, another way.
      > ( I did'n t read this ML usually )
      >
      >
      > I found.
      >
      > Windows notepad can read no-bomb UTF-8 file, but if edited or created
      > file is saved with bomb.
      >
      > By unicode consocium standard, UTF-8 file can added bomb.
      > but UTF-8 not need bomb.
      >
      >
      > Vim use libiconv by FSF.
      > I search on unix platform, `man iconv_open`, libiconv don't know
      > UTF-8 with bomb. Then, libiconv can't detect UTF-8 with bomb.
      > this is problem of libiconv, not Vim, by strict logic.
      >
      >
      > In this reason, In my vimrc set filencodings sorted earlier UTF-8 than cp932
      > but notepad saved file detected cp932.
      >
      > On justice way, We have to patch to libiconv, add UTF-8 bomb.
      > But Can't save this weak point by Vim side?
      >
      >

      The BOM is a valid codepoint, ZERO-WIDTH NO-BREAK SPACE; however its use
      in that capacity is now deprecated (ZERO-WIDTH NON-JOINER is preferred
      IIUC). Still, if any software accepts UTF-8 files "only" if there is no
      BOM in them, that software cannot be called UTF-8-compliant.

      I don't know which libiconv you are using, but my version of Vim (which
      has +iconv and a GTK2/Gnome2 GUI) accepts the BOM with no problem, not
      only in UTF-16/UTF-32 but also in UTF-8.


      Best regards,
      Tony.
      --
      The church is near but the road is icy; the bar is far away but I will
      walk carefully.
      -- Russian Proverb

      --~--~---------~--~----~------------~-------~--~----~
      You received this message from the "vim_multibyte" maillist.
      For more information, visit http://www.vim.org/maillist.php
      -~----------~----~----~----~------~----~------~--~---
    • T.P.S.Nakagawa
      Thanks for reading my broken english, and retry and reply. ... ... Sorry , I did t try by gvim on Windows ( downloaded, version 7.1 / 2007 May 12 compiled
      Message 2 of 25 , May 2, 2008
      • 0 Attachment
        Thanks for reading my broken english, and retry and reply.


        2008-05-02 11:36 (JST) , Tony Mechelynck sent follow message:
        >> Vim use libiconv by FSF.
        >> I search on unix platform, `man iconv_open`, libiconv don't know
        >> UTF-8 with bomb. Then, libiconv can't detect UTF-8 with bomb.
        >> this is problem of libiconv, not Vim, by strict logic.
        >>
        >>
        >> In this reason, In my vimrc set filencodings sorted earlier UTF-8 than cp932
        >> but notepad saved file detected cp932.
        >>
        >> On justice way, We have to patch to libiconv, add UTF-8 bomb.
        >> But Can't save this weak point by Vim side?
        >>
        >>
        <>
        > I don't know which libiconv you are using, but my version of Vim (which
        > has +iconv and a GTK2/Gnome2 GUI) accepts the BOM with no problem, not
        > only in UTF-16/UTF-32 but also in UTF-8.
        >

        Sorry , I did't try by gvim on Windows ( downloaded, version 7.1 / 2007 May 12
        compiled by Mr. Bram ) + related by http://www.vim.org/download.php#pc
        old version sourceforge libiconv for Win32
        http://sourceforge.net/project/showfiles.php?group_id=25167&package_id=51458
        ( Release 1.9.1 / January 14 2004 )

        on unix ( FreeBSD 5-stable ), with libiconv version 1.11 (+ OS patch 1 )
        same (.|_)vimrc file, console vim can detect UTF-8 + BOM by notepad.
        ( vim version is self compile by cvs source 2007 July 31, he say version 7.1.147 )


        Someone compile libiconv ver 1.11 or later for Win 32 platform?
        ( I tried 6 month ago, and I fail and busy )

        --
        Nakagawa Tsuneo mailto:yaemon@...
        Web site ( Jpanese only ) http://www.kikansha.jp/~yaemon/


        --~--~---------~--~----~------------~-------~--~----~
        You received this message from the "vim_multibyte" maillist.
        For more information, visit http://www.vim.org/maillist.php
        -~----------~----~----~----~------~----~------~--~---
      • Tony Mechelynck
        On 02/05/08 10:17, T.P.S.Nakagawa wrote: [...] ... This is 7.1.000. I recommend the updated Vim and gvim compiled by Steve Hall (currently 7.1.293), obtainable
        Message 3 of 25 , May 2, 2008
        • 0 Attachment
          On 02/05/08 10:17, T.P.S.Nakagawa wrote:
          [...]
          > Sorry , I did't try by gvim on Windows ( downloaded, version 7.1 / 2007 May 12
          > compiled by Mr. Bram ) + related by http://www.vim.org/download.php#pc
          > old version sourceforge libiconv for Win32
          > http://sourceforge.net/project/showfiles.php?group_id=25167&package_id=51458
          > ( Release 1.9.1 / January 14 2004 )

          This is 7.1.000. I recommend the updated Vim and gvim compiled by Steve
          Hall (currently 7.1.293), obtainable from
          https://sourceforge.net/project/showfiles.php?group_id=43866&package_id=39721
          . It has +iconv/dyn, its full ":version" text can be seen by clicking on
          the word "Notes" or the clipboard-like icon next to the version number
          on that same page.

          >
          > on unix ( FreeBSD 5-stable ), with libiconv version 1.11 (+ OS patch 1 )
          > same (.|_)vimrc file, console vim can detect UTF-8 + BOM by notepad.
          > ( vim version is self compile by cvs source 2007 July 31, he say version 7.1.147 )
          >
          >
          > Someone compile libiconv ver 1.11 or later for Win 32 platform?
          > ( I tried 6 month ago, and I fail and busy )
          >

          Libiconv 1.9.2 is available precompiled from GnuWin32 from the page
          http://gnuwin32.sourceforge.net/packages/libiconv.htm . This is one
          sub-sub-level above what you already have -- you might try using this
          until or unless you succeed to get (or make) a later version. Since
          Steve Hall's builds include +iconv/dyn and the "Compilation" and
          "Linking" parts of its ":version" text mention no iconv version, I
          suppose upgrading iconv (if and when) means simply dropping the DLL over
          the former version (in the PATH or in $VIMRUNTIME). I expect, though,
          that upgrading Vim (see above) will be more important than upgrading iconv.

          Libiconv 1.12 is distributed in source form at
          http://ftp.gnu.org/pub/gnu/libiconv/libiconv-1.12.tar.gz but I suppose
          these sources are meant primarily for Unix -- I don't know what
          adaptations (if any) might be necessary to build a Windows DLL from them.


          Best regards,
          Tony.
          --
          The human mind ordinarily operates at only ten percent of its capacity
          -- the rest is overhead for the operating system.

          --~--~---------~--~----~------------~-------~--~----~
          You received this message from the "vim_multibyte" maillist.
          For more information, visit http://www.vim.org/maillist.php
          -~----------~----~----~----~------~----~------~--~---
        • T.P.S.Nakagawa
          Thank you Tony, and excuse me too late reply. ... O.K. I install all GnuWin32 and add path to it s bin. I erase old iconv.dll, result of all test pattern is
          Message 4 of 25 , May 3, 2008
          • 0 Attachment
            Thank you Tony, and excuse me too late reply.


            2008-05-02 18:46 (JST) , Tony Mechelynck sent follow message:

            >> Someone compile libiconv ver 1.11 or later for Win 32 platform?
            >> ( I tried 6 month ago, and I fail and busy )
            >>
            >
            > Libiconv 1.9.2 is available precompiled from GnuWin32 from the page

            O.K. I install all GnuWin32 and add path to it's bin.
            I erase old iconv.dll, result of all test pattern is same to old iconv.dll.

            # And Thank you webmaster of http://www.vim.org/ (is Mr. Bram? ) to link change.


            > This is 7.1.000. I recommend the updated Vim and gvim compiled by Steve
            > Hall (currently 7.1.293), obtainable from
            > https://sourceforge.net/project/showfiles.php?group_id=43866&package_id=39721

            Next, I upgrade vim to 7.1.293.
            result of all test pattern is same :-(


            I think, in this time, return to need install new version of iconv.dll for
            Windows.

            My Win box is too poor.
            If I try, need cross compile on Free BSD box. But I didn't try cross compile, too.


            Best regards,
            Nakagawa, a.k.a. yaemon

            --
            Nakagawa Tsuneo mailto:yaemon@...
            Web site ( Jpanese only ) http://www.kikansha.jp/~yaemon/

            --~--~---------~--~----~------------~-------~--~----~
            You received this message from the "vim_multibyte" maillist.
            For more information, visit http://www.vim.org/maillist.php
            -~----------~----~----~----~------~----~------~--~---
          • Tony Mechelynck
            ... No problem. [...] ... Make sure to have iconv.dll (and possibly all GnuWin32 executables) in the PATH (which, on Windows, is a semicolon-separated list).
            Message 5 of 25 , May 3, 2008
            • 0 Attachment
              On 03/05/08 20:30, T.P.S.Nakagawa wrote:
              > Thank you Tony, and excuse me too late reply.

              No problem.

              [...]
              > > Libiconv 1.9.2 is available precompiled from GnuWin32 from the page
              >
              > O.K. I install all GnuWin32 and add path to it's bin.
              > I erase old iconv.dll, result of all test pattern is same to old iconv.dll.

              Make sure to have iconv.dll (and possibly all GnuWin32 executables) in
              the PATH (which, on Windows, is a semicolon-separated list). How to set
              that depends on your Windows version. IIRC, in XP it is at "Control
              Panel -> System -> Advanced -> Environment Variables" or something similar.

              >
              > # And Thank you webmaster of http://www.vim.org/ (is Mr. Bram? ) to link change.
              [...]

              I think it is Bram, yes.

              Best regards,
              Tony.
              --
              "I am ready to meet my Maker. Whether my Maker is prepared for the
              great ordeal of meeting me is another matter."
              -- Winston Churchill

              --~--~---------~--~----~------------~-------~--~----~
              You received this message from the "vim_multibyte" maillist.
              For more information, visit http://www.vim.org/maillist.php
              -~----------~----~----~----~------~----~------~--~---
            • T.P.S.Nakagawa
              Thank you Tony. ... Oh , Yes. ( My english is too poor, but not begenner of PC ) I see :fileencodings on gvim , and my _vimrc set ...
              Message 6 of 25 , May 3, 2008
              • 0 Attachment
                Thank you Tony.


                2008-05-04 8:08 (JST) , Tony Mechelynck sent follow message:
                >> > Libiconv 1.9.2 is available precompiled from GnuWin32 from the page
                >>
                >> O.K. I install all GnuWin32 and add path to it's bin.
                >> I erase old iconv.dll, result of all test pattern is same to old iconv.dll.
                >
                > Make sure to have iconv.dll (and possibly all GnuWin32 executables) in
                > the PATH (which, on Windows, is a semicolon-separated list). How to set
                > that depends on your Windows version. IIRC, in XP it is at "Control
                > Panel -> System -> Advanced -> Environment Variables" or something similar.

                Oh , Yes. ( My english is too poor, but not begenner of PC )

                I see :fileencodings on gvim , and my _vimrc set

                > if has('iconv')
                > set
                fileencodings=ascii,iso-2022-jp,utf-8,euc-jp,utf-16,cp932,java,ucs-2-internal,euc-jis0213,utf-16,ISO-8859-1
                > endif

                if not path success, `:fileencodings?` will return another value, is'not it?


                best regards,
                Nakagawa, a.k.a yaemon


                --
                NAKAGAWA Tsuneo mailto:yaemon@...
                Web site ( Japanese ony ) http://www.kikansha.jp/~yaemon/

                --~--~---------~--~----~------------~-------~--~----~
                You received this message from the "vim_multibyte" maillist.
                For more information, visit http://www.vim.org/maillist.php
                -~----------~----~----~----~------~----~------~--~---
              • Tony Mechelynck
                ... Yes, a Vim build compiled with +iconv/dyn will act as -iconv (has( iconv ) == 0) if it cannot establish contact with _any_ iconv.dll so the other :if
                Message 7 of 25 , May 3, 2008
                • 0 Attachment
                  On 04/05/08 03:06, T.P.S.Nakagawa wrote:
                  > Thank you Tony.
                  >
                  >
                  > 2008-05-04 8:08 (JST) , Tony Mechelynck sent follow message:
                  > >> > Libiconv 1.9.2 is available precompiled from GnuWin32 from the page
                  > >>
                  > >> O.K. I install all GnuWin32 and add path to it's bin.
                  > >> I erase old iconv.dll, result of all test pattern is same to old iconv.dll.
                  > >
                  > > Make sure to have iconv.dll (and possibly all GnuWin32 executables) in
                  > > the PATH (which, on Windows, is a semicolon-separated list). How to set
                  > > that depends on your Windows version. IIRC, in XP it is at "Control
                  > > Panel -> System -> Advanced -> Environment Variables" or something similar.
                  >
                  > Oh , Yes. ( My english is too poor, but not begenner of PC )
                  >
                  > I see :fileencodings on gvim , and my _vimrc set
                  >
                  > > if has('iconv')
                  > > set
                  > fileencodings=ascii,iso-2022-jp,utf-8,euc-jp,utf-16,cp932,java,ucs-2-internal,euc-jis0213,utf-16,ISO-8859-1
                  > > endif
                  >
                  > if not path success, `:fileencodings?` will return another value, is'not it?
                  >
                  >
                  > best regards,
                  > Nakagawa, a.k.a yaemon
                  >
                  >

                  Yes, a Vim build compiled with +iconv/dyn will act as -iconv
                  (has("iconv") == 0) if it cannot establish contact with _any_ iconv.dll
                  so the other ":if" branch will be followed, and ":set fileencodings?"
                  (not ":fileencodings" which returns an error) will display a different
                  value.

                  That 'fileencodings' setting makes me wonder:
                  - Shouldn't it start with "ucs-bom" to detect those Unicode files which
                  have a BOM?
                  - Can "ascii" give a fail signal (doesn't Vim treat it as an alias for
                  "latin1")? -- If it can, then it's OK there, but otherwise not.
                  - What is "ucs-2-internal"? Won't it be detected as "utf-16" (which is a
                  superset of UCS-2) by the "utf-16" entry three steps earlier?
                  - Why is "utf-16" mentioned twice? (The second entry does no harm, but
                  will never be used.)
                  - Won't you ever receive UCS-2/UTF-16 files in little-endian ordering
                  (which is standard on Intel ix86 and therefore on Windows)?


                  Best regards,
                  Tony.
                  --
                  If everybody minded their own business, the world would go
                  around a deal faster.
                  -- The Duchess, "Through the Looking Glass"

                  --~--~---------~--~----~------------~-------~--~----~
                  You received this message from the "vim_multibyte" maillist.
                  For more information, visit http://www.vim.org/maillist.php
                  -~----------~----~----~----~------~----~------~--~---
                • T.P.S.Nakagawa
                  Sorry every typo and send my dirty setting by historical reason. ... Thank you. I correct this sort. ...
                  Message 8 of 25 , May 3, 2008
                  • 0 Attachment
                    Sorry every typo and send my dirty setting by historical reason.


                    2008-05-04 11:24 (JST) , Tony Mechelynck sent follow message:
                    > That 'fileencodings' setting makes me wonder:
                    > - Shouldn't it start with "ucs-bom" to detect those Unicode files which
                    > have a BOM?
                    > - Can "ascii" give a fail signal (doesn't Vim treat it as an alias for
                    > "latin1")? -- If it can, then it's OK there, but otherwise not.
                    > - What is "ucs-2-internal"? Won't it be detected as "utf-16" (which is a
                    > superset of UCS-2) by the "utf-16" entry three steps earlier?
                    > - Why is "utf-16" mentioned twice? (The second entry does no harm, but
                    > will never be used.)
                    > - Won't you ever receive UCS-2/UTF-16 files in little-endian ordering
                    > (which is standard on Intel ix86 and therefore on Windows)?

                    Thank you. I correct this sort.

                    > set
                    fileencodings=ascii,iso-2022-jp,utf-8,euc-jp,java,utf-16,ucs-bomb,cp932,utf-16LE,euc-jis0213,ISO-8859-1


                    First is "ascii" , that's my intention.
                    ( Is it really alias of latin1? latin1 have another code of ascii ).

                    If not exist this first, all ascii code ( ex. programming source ) detected
                    iso-2022-jp and Can't add multibyte comment by UTF-8.

                    I have a trick in this.


                    ----- last of vimrc -------
                    if has( 'autocmd' )
                    source $HOME/.vim/mine/filetype.vim " about filetype
                    source $HOME/.vim/mine/encode.vim " about encoding to save
                    endif

                    ------ cat encode.vim -----
                    " $Id: encode.vim,v 1.6 2007/11/08 05:03:18 yaemon Exp $
                    " set fileencode to save
                    "

                    :autocmd BufNewFile,BufRead *.jis set fileencoding=iso-2022-jp
                    :autocmd BufNewFile,BufRead *.sjis set fileencoding=shift-jis
                    :autocmd BufNewFile,BufRead *.euc-jp set fileencoding=euc-jp
                    :autocmd BufNewFile,BufRead *.mozex set fileencoding=utf-8
                    :autocmd BufNewFile,BufRead *.elm set fileencoding=utf-8
                    :autocmd BufNewFile,BufRead cddbread.* set fileencoding=utf-8

                    :autocmd BufNewFile,BufRead * call s:DefaultSaveCode()

                    function s:DefaultSaveCode()
                    if ( ( &fileencoding == "" ) || ( &fileencoding == "ascii" ) )
                    let &fileencoding = "utf-8"
                    endif
                    endfunction
                    ----------------------------------


                    Isn't is elegant for add UTF-8 to ascii file?



                    Best, Best, regard
                    yaemon

                    --
                    NAKAGAWA Tsuneo (a.k.a. yaemon ) mailto:yaemon@...
                    Web site ( Japanese ony ) http://www.kikansha.jp/~yaemon/

                    --~--~---------~--~----~------------~-------~--~----~
                    You received this message from the "vim_multibyte" maillist.
                    For more information, visit http://www.vim.org/maillist.php
                    -~----------~----~----~----~------~----~------~--~---
                  • T.P.S.Nakagawa
                    Good morning ( it s 8:25 in Japan ) ... Today, I success compile iconv-1.12 for Windows. by cross compile on unix box, mingw32. If you try this, please get
                    Message 9 of 25 , May 4, 2008
                    • 0 Attachment
                      Good morning ( it's 8:25 in Japan )


                      2008-05-04 3:30 (JST) , I sent follow message:
                      > I think, in this time, return to need install new version of iconv.dll for
                      > Windows.
                      >
                      > My Win box is too poor.
                      > If I try, need cross compile on Free BSD box. But I didn't try cross compile, too.

                      Today, I success compile iconv-1.12 for Windows.
                      by cross compile on unix box, mingw32.

                      If you try this, please get here
                      http://www.kikansha.jp/~yaemon/mingw/libiconv-1.12-mingw.zip


                      ...but, I can't correctry open file edited and saved UTF-8 by notepad :(


                      Best regard

                      --
                      NAKAGAWA Tsuneo (a.k.a. yaemon ) mailto:yaemon@...
                      Web site ( Japanese ony ) http://www.kikansha.jp/~yaemon/

                      --~--~---------~--~----~------------~-------~--~----~
                      You received this message from the "vim_multibyte" maillist.
                      For more information, visit http://www.vim.org/maillist.php
                      -~----------~----~----~----~------~----~------~--~---
                    • T.P.S.Nakagawa
                      Hello ... ... I see! On Windows XP, file encoding detect success or not, gvim can t display UTF8 + BOM, if &fileencodings setted. lush-up way is that. ...
                      Message 10 of 25 , May 5, 2008
                      • 0 Attachment
                        Hello


                        2008-05-05 8:25 (JST) , I sent follow message:
                        > Today, I success compile iconv-1.12 for Windows.
                        <>
                        > ...but, I can't correctry open file edited and saved UTF-8 by notepad :(

                        I see!

                        On Windows XP, file encoding detect success or not, gvim can't display
                        UTF8 + BOM, if &fileencodings setted.


                        lush-up way is that.

                        :set binary
                        2x
                        :set fenc=
                        :w
                        :set nobinary
                        :e!


                        cursed notepad!
                        but isn't it bug of gvim?


                        regard
                        --

                        NAKAGAWA Tsuneo (a.k.a. yaemon ) mailto:yaemon@...
                        Web site ( Japanese ony ) http://www.kikansha.jp/~yaemon/

                        --~--~---------~--~----~------------~-------~--~----~
                        You received this message from the "vim_multibyte" maillist.
                        For more information, visit http://www.vim.org/maillist.php
                        -~----------~----~----~----~------~----~------~--~---
                      • Tony Mechelynck
                        ... If what you said above is exact, it s a Notepad bug: a UTF-8 BOM is three bytes, a UTF-16 BOM (also used for UCS-2) is two bytes, a UTF-32 BOM is four
                        Message 11 of 25 , May 5, 2008
                        • 0 Attachment
                          On 05/05/08 09:17, T.P.S.Nakagawa wrote:
                          > Hello
                          >
                          >
                          > 2008-05-05 8:25 (JST) , I sent follow message:
                          > > Today, I success compile iconv-1.12 for Windows.
                          > <>
                          > > ...but, I can't correctry open file edited and saved UTF-8 by notepad :(
                          >
                          > I see!
                          >
                          > On Windows XP, file encoding detect success or not, gvim can't display
                          > UTF8 + BOM, if&fileencodings setted.
                          >
                          >
                          > lush-up way is that.
                          >
                          > :set binary
                          > 2x
                          > :set fenc=
                          > :w
                          > :set nobinary
                          > :e!
                          >
                          >
                          > cursed notepad!
                          > but isn't it bug of gvim?
                          >
                          >
                          > regard

                          If what you said above is exact, it's a Notepad bug: a UTF-8 BOM is
                          three bytes, a UTF-16 BOM (also used for UCS-2) is two bytes, a UTF-32
                          BOM is four bytes. If there was a two-byte BOM on a UTF-8 file, it's a
                          bug in the program which produced that file.

                          When a Windows program (such as WordPad) saves a file as "Unicode text",
                          it's usually UTF-16le with BOM, which means that the first two bytes are
                          FF FE and that after that, even bytes are often null bytes.

                          Best regards,
                          Tony.
                          --
                          Don't take life too seriously -- you'll never get out of it alive.

                          --~--~---------~--~----~------------~-------~--~----~
                          You received this message from the "vim_multibyte" maillist.
                          For more information, visit http://www.vim.org/maillist.php
                          -~----------~----~----~----~------~----~------~--~---
                        • T.P.S.Nakagawa
                          Sorry, Tony. But I pleasure of report next thing of this problem. ... Oh yes. I delete 2 bytes , that displayed in unix UTF-8 console. But by shown od -xc
                          Message 12 of 25 , May 5, 2008
                          • 0 Attachment
                            Sorry, Tony.

                            But I pleasure of report next thing of this problem.

                            2008-05-05 23:48 (JST) , Tony Mechelynck sent follow message:
                            > If what you said above is exact, it's a Notepad bug: a UTF-8 BOM is
                            > three bytes, a UTF-16 BOM (also used for UCS-2) is two bytes, a UTF-32

                            Oh yes. I delete 2 bytes , that displayed in unix UTF-8 console.
                            But by shown "od -xc" command, notepad attach 3 bytes of BOM. sorry.


                            Then, I report more deep for this problem.
                            Vim read UTF-8 + BOM , if fileencodings setted, allways display by UTF-8.
                            so Windows Japanese version ( must display cp932 )
                            so unix console setted ja_JP.eucJP.

                            That's all of reason , bad display.

                            I read 1 hour sources, around *p_fencs setting, but I sleeped.
                            It's hard of read part of big source.


                            Best regard, by yaemon.


                            P.S. now, download page of libiconv is
                            http://www.kikansha.jp/~yaemon/misc/libiconv
                            --
                            NAKAGAWA Tsuneo (a.k.a. yaemon ) mailto:yaemon@...
                            Web site ( Japanese ony ) http://www.kikansha.jp/~yaemon/

                            --~--~---------~--~----~------------~-------~--~----~
                            You received this message from the "vim_multibyte" maillist.
                            For more information, visit http://www.vim.org/maillist.php
                            -~----------~----~----~----~------~----~------~--~---
                          • Tony Mechelynck
                            ... If your fileencodings starts with ucs-bom , Vim ought to detect correctly any Unicode encoding when there is a BOM without interfering with the
                            Message 13 of 25 , May 5, 2008
                            • 0 Attachment
                              On 06/05/08 04:58, T.P.S.Nakagawa wrote:
                              > Sorry, Tony.
                              >
                              > But I pleasure of report next thing of this problem.
                              >
                              > 2008-05-05 23:48 (JST) , Tony Mechelynck sent follow message:
                              > > If what you said above is exact, it's a Notepad bug: a UTF-8 BOM is
                              > > three bytes, a UTF-16 BOM (also used for UCS-2) is two bytes, a UTF-32
                              >
                              > Oh yes. I delete 2 bytes , that displayed in unix UTF-8 console.
                              > But by shown "od -xc" command, notepad attach 3 bytes of BOM. sorry.
                              >
                              >
                              > Then, I report more deep for this problem.
                              > Vim read UTF-8 + BOM , if fileencodings setted, allways display by UTF-8.
                              > so Windows Japanese version ( must display cp932 )
                              > so unix console setted ja_JP.eucJP.

                              If your 'fileencodings' starts with "ucs-bom", Vim ought to detect
                              correctly any Unicode encoding when there is a BOM without interfering
                              with the detection of other encodings, unless they may start with one or
                              more of the following codes and contain not a single invalid byte (or
                              invalid sequence of bytes) for the corresponding Unicode encoding (I
                              know that many combinations of bytes higher than 0x7F are invalid in
                              UTF-8; I'm less sure about the other):

                              EF BB BF UTF-8
                              FE FF UTF-16be
                              FF FE UTF-16le
                              00 00 FE FF UTF-32be
                              FF FE 00 00 UTF-32le

                              Notice that Vim (and any other program with BOM detection) may "guess
                              wrong" if a file in UTF-16le with BOM starts with a NULL; but I suppose
                              that such a case is so rare it may be safely ignored.

                              Notes:
                              - Even if editing cp932 files, you may set 'encoding' to utf-8
                              - In GUI mode, anything that 'encoding' can represent, can be displayed
                              if your 'guifont' has a glyph for it. Characters for which your
                              'guifont' has no glyph may be represented by a "placeholder" question
                              mark or hollow box etc.; but if you use the GTK2 GUI (X11 only, thus not
                              on Windows) it may, in some cases, be clever enough to find an
                              appropriate glyph in a different font.
                              - Even if your terminal display is set to accept cp932 output, you may
                              still set 'encoding' to utf-8 in Console mode if 'termencoding' is set
                              to cp932, but of course in that case if you edit Unicode (or other
                              non-cp932) files containing characters which cannot be represented in
                              cp932, you will get a "placeholder" display (possibly a question mark or
                              a hollow box) at that position even though the actual contents of the
                              file are correct.
                              - The above applies also, of course, with "cp932" replaced everywhere by
                              "euc-jp".

                              >
                              > That's all of reason , bad display.
                              >
                              > I read 1 hour sources, around *p_fencs setting, but I sleeped.
                              > It's hard of read part of big source.

                              Yes, especially when you're lacking sleep. ;-)

                              >
                              >
                              > Best regard, by yaemon.
                              >
                              >
                              > P.S. now, download page of libiconv is
                              > http://www.kikansha.jp/~yaemon/misc/libiconv
                              > --
                              > NAKAGAWA Tsuneo (a.k.a. yaemon ) mailto:yaemon@...
                              > Web site ( Japanese ony ) http://www.kikansha.jp/~yaemon/

                              Best regards,
                              Tony.
                              --
                              "The Good Ship Enterprise" (to the tune of "The Good Ship Lollipop")

                              On the good ship Enterprise
                              Every week there's a new surprise
                              Where the Romulans lurk
                              And the Klingons often go berserk.

                              Yes, the good ship Enterprise
                              There's excitement anywhere it flies
                              Where Tribbles play
                              And Nurse Chapel never gets her way.

                              See Captain Kirk standing on the bridge,
                              Mr. Spock is at his side.
                              The weekly menace, ooh-ooh
                              It gets fried, scattered far and wide.

                              It's the good ship Enterprise
                              Heading out where danger lies
                              And you live in dread
                              If you're wearing a shirt that's red.
                              -- Doris Robin and Karen Trimble of The L.A. Filkharmonics

                              --~--~---------~--~----~------------~-------~--~----~
                              You received this message from the "vim_multibyte" maillist.
                              For more information, visit http://www.vim.org/maillist.php
                              -~----------~----~----~----~------~----~------~--~---
                            • T.P.S.Nakagawa
                              Thanks all. ... Oh!? Did I forget write? After libiconv version up to 1.11 , Vim detect all time correct charset. ( If I confidence set fileencoding? ) And,
                              Message 14 of 25 , May 7, 2008
                              • 0 Attachment
                                Thanks all.


                                2008-05-06 13:30 (JST) , Tony Mechelynck sent follow message:
                                > If your 'fileencodings' starts with "ucs-bom", Vim ought to detect
                                > correctly any Unicode encoding when there is a BOM without interfering
                                > with the detection of other encodings, unless they may start with one or
                                > more of the following codes and contain not a single invalid byte (or
                                > invalid sequence of bytes) for the corresponding Unicode encoding (I
                                > know that many combinations of bytes higher than 0x7F are invalid in
                                > UTF-8; I'm less sure about the other):

                                Oh!? Did I forget write?
                                After libiconv version up to 1.11 , Vim detect all time correct charset.
                                ( If I confidence "set fileencoding?" )

                                And, if I wrote in _vimrc "set fileencodings=" ( set null string ),
                                or started by ucs-bomb,
                                or overwrite by "set fileencodings=ucs-bomb" ( that's default of Win32),
                                at all case gvim try diaplay by UTF-8.

                                Is it only Windows Japanese version's ?
                                By latin-1 charset, if you wrote (c) _copyright by one character_ or accent + e
                                by notepad and save by UTF-8, gvim display as you want?


                                Best regard, by yaemon

                                --
                                NAKAGAWA Tsuneo (a.k.a. yaemon ) mailto:yaemon@...
                                Web site ( Japanese ony ) http://www.kikansha.jp/~yaemon/

                                --~--~---------~--~----~------------~-------~--~----~
                                You received this message from the "vim_multibyte" maillist.
                                For more information, visit http://www.vim.org/maillist.php
                                -~----------~----~----~----~------~----~------~--~---
                              Your message has been successfully submitted and would be delivered to recipients shortly.