Issue 62174

Summary: VCL.xcu still needs correction of misspellings on real font names.
Product: gsl Reporter: tora3
Component: codeAssignee: tora3
Status: CLOSED FIXED QA Contact: issues@gsl <issues>
Severity: Trivial    
Priority: P3 CC: andre.schnabel, issues, y-catch, yossy_takeuchi
Version: OOo 2.0.2Keywords: release_blocker
Target Milestone: OOo 2.1   
Hardware: All   
OS: All   
Issue Type: PATCH Latest Confirmation in: ---
Developer Difficulty: ---
Issue Depends on:    
Issue Blocks: 55680    
Attachments:
Description Flags
Investigations on VCL.xcu of the revision 1.50.
none
A Perl script running as "perl convert_VCL_xcu.pl VCL.xcu" will produce VCL_xcu.ods for investigation.
none
A patch against the revision 1.54
none
A patch-applied file none

Description tora3 2006-02-16 18:44:10 UTC
Since the issue 58535 has no practical progress, misspells in the 
file VCL.xcu have to be corrected one by one following suggestions 
in the issue 31813.

Before making corrections, Japanese community would like to ask 
some information on the item 2 and 6 listed below.

  1. Font names in the file VCL.xcu have to be exactly identical to 
     the real font names defined in the corresponding font files.

      e.g. Correction : ï¼­ï¼³ P明æœ
           Misspelling: MS P 明æœ

  2. What entries have what effects? There seems 23 types of entry.
     "en" has 22 entries and "ja" has only 10 entries.

       en                  ja
       ==================  ==================
       CJK_DISPLAY
       CJK_HEADING         CJK_HEADING
       CJK_PRESENTATION    CJK_PRESENTATION
       CJK_SPREADSHEET     CJK_SPREADSHEET
                           CJK_TEXT
       CTL_DISPLAY
       CTL_HEADING
       CTL_PRESENTATION
       CTL_SPREADSHEET
       CTL_TEXT
       FIXED               FIXED
       LATIN_DISPLAY
       LATIN_FIXED         LATIN_FIXED
       LATIN_HEADING
       LATIN_PRESENTATION  LATIN_PRESENTATION
       LATIN_SPREADSHEET   LATIN_SPREADSHEET
       LATIN_TEXT
       SANS
       SANS_UNICODE
       SERIF
       SYMBOL
       UI_FIXED            UI_FIXED
       UI_SANS             UI_SANS

     For details, see the sheet Summary of VCL_xcu_revsion_1.50.ods.

  3. Need to consider similarity of font families. Gothic related fonts
     are suitable for presentation and not good for default fonts of word 
     processor while Micho related fonts are suitable for word processor 
     but and good for presentation. The stroke of Gothic variant fonts is 
     too thick to read for long time and the stroke of Micho is too thin 
     for audience to look from distance. Current settings of "ja" seem not 
     to consider this preferences.

     For details, see the sheet ja of VCL_xcu_revsion_1.50.ods.

  4. Improper font names have to be removed.
     e.g. Mincho and Gothic are the font names used by Korean. 
     No Japanese fonts use Mincho or Gothic. Japanese ones use "Something 
     Mincho" and "Something Gothic."

  5. The order of font names has to be reconsidered.

  6. Are both English and localized font names needed to be listed?
     e.g. msgothic.ttc includes three fonts. One of them is named MS PGothic 
     for English name and MS Pゴシック for Japanese name.
Comment 1 tora3 2006-02-16 18:48:39 UTC
Created attachment 34227 [details]
Investigations on VCL.xcu of the revision 1.50.
Comment 2 tora3 2006-02-16 18:53:39 UTC
Created attachment 34228 [details]
A Perl script running as "perl convert_VCL_xcu.pl VCL.xcu" will produce VCL_xcu.ods for investigation.
Comment 3 tora3 2006-02-16 18:58:55 UTC
For investigation of the file VCL.xcu, the following command  

  perl convert_VCL_xcu.pl officecfg/registry/data/org/openoffice/VCL.xcu

will produce VCL_xcu.ods including several sheets: summary and individual 
sheets of each language. 

Comment 4 tora3 2006-02-16 19:04:12 UTC
Just information. 

The Perl script has found that:
  the entry "CJK_HEADING" of "zh-hk" has duplicated elements of <value>.
  the entry "CJK_HEADING" of "zh-sq" has duplicated elements of <value>.

Comment 5 philipp.lohmann 2006-02-17 10:27:01 UTC
tora: the entries that are not present in a locale are inherited from english
(which is our last fallback for searching). en locale has all default font
entries used in vcl.

As to whether the localized fontnames precede the english ones in font
selection, i think hdu knows better than me. Any spelling corrections for these
entries are very welcome, thank you.
Comment 6 hdu@apache.org 2006-03-10 12:31:11 UTC
> 6. Are both English and localized font names needed to be listed?
>     e.g. msgothic.ttc includes three fonts. One of them is named MS PGothic 
>     for English name and MS Pゴシック for Japanese name.

One of the font name alternatives should suffice. Having the english names
doesn't make much difference. It just makes it easier for non-native speakers to
maintaining the list.
Comment 7 hdu@apache.org 2006-03-15 09:39:04 UTC
The spelling fixes from Latin "MS P*" font names to their full width
counterparts were applied in CWS vclxcu01. Please check revision 1.50.22.3 of
VCL.xcu.
Comment 8 hdu@apache.org 2006-04-13 16:47:56 UTC
HDU->ES: please verify in CWS vclxcu01
Comment 9 hdu@apache.org 2006-04-13 16:48:54 UTC
.
Comment 10 hdu@apache.org 2006-04-13 16:49:18 UTC
.
Comment 11 eric.savary 2006-05-05 11:49:12 UTC
Quick'n dirty notes 4 me:
iconv -f UTF8 -t UTF16 -o VCLutf16 VCL.xcu
hexdump VCLutf16 > VCLhex
cat VCLhex | cut -d " " -f 2-8 > VCLhex.noadd

Postive testing:
cat VCLhex.noadd | grep "ff33 3000"
cat VCLhex.noadd | grep "ff2d 3000"
->matches

Negative testing:
cat VCLhex.noadd | grep "ff33 0020"
cat VCLhex.noadd | grep "ff2d 0020"
->no match
Comment 12 eric.savary 2006-05-05 11:51:13 UTC
Verified in CWS vclxcu01
Comment 13 eric.savary 2006-05-11 11:05:14 UTC
Ok in src680m168
Comment 14 tora3 2006-07-03 17:15:43 UTC
In reality, fullwidth characters of MS are followed by a halfwidth space letter.

Correct font names begin with:
 FF2D;FULLWIDTH LATIN CAPITAL LETTER M;Lu;0;L;<wide> 004D;;;;N;;;;FF4D;
 FF33;FULLWIDTH LATIN CAPITAL LETTER S;Lu;0;L;<wide> 0053;;;;N;;;;FF53;
 0020;SPACE;Zs;0;WS;;;;;N;;;;;

I apologize for not confirming the fix at the time.
Comment 15 eric.savary 2006-08-15 11:05:35 UTC
ES->HDU: PLease, have a look.
Comment 16 Martin Hollmichel 2006-09-08 01:43:45 UTC
set new target.
Comment 17 tora3 2006-09-10 10:20:43 UTC
Problem:
  There are some misspellings in the file VCL.xcu regarding a set of 4 Japanese 
  primary MS fonts whose name begins with a sequence of fullwidth letter M and S.

Impact:
  Leaving the misspellings untouched results in failure of choosing a default 
  font name of the applications - Writer, Calc, Impress, Draw - on Windows.

  Each application mistakenly chooses the first one - default font name for 
  Solaris - from the list of possible default font names defined in the file 
  VCL.xcu since the application on Windows cannot find an appropriate font name 
  due to the misspellings.

  Consequently, the interoperability with other applications on Windows is broken.
  Creating a document from the scratch with OpenOffice.org on Windows and saving 
  it in Microsoft Office file format, and opening it with Microsoft Office results 
  in different text layout. Microsoft Office cannot find the font name for Solaris 
  specified in the original document and then uses its own default font, instead.

Patch:
  Here is a procedure to correct misspellings regarding a set of 4 Japanese MS
fonts 
  in the file VCL.xcu revision 1.54.

  To correct them, substitute 
   a sequence of fullwidth letter M, S, and ideograph space 
  with 
   a sequence of fullwidth letter M, S, and normal space
  by the following command:

  perl -i.bak -pe 's/(\xef\xbc\xad)(\xef\xbc\xb3)(\xe3\x80\x80)/$1$2 /g' VCL.xcu

Information 1:
  For the previous release, 2.0.3, Japanese community has provided Japanese users
  with a corrected file VCL,xcu and an installer that easily installs the file 
  into user's installation set of OpenOffice.org 2.0.3.

Information 2:
  Microsoft Windows variants Japanese comes with a font "MS UI Gothic" as a UI 
  font, which is one of the three fonts collected in the font file MSGOTHIC.TTC. 

  Our file VCL.xcu, however, currently does not use the font for UI. It is not a 
  big deal, but the font MS UI Gothic is used by other Windows applications as 
  a default UI font and is slightly thiner than MS PGothic, our current one, and 
  saves room for smaller resolution of display device. 

  That could be taken of in the near future.
Comment 18 tora3 2006-09-10 10:22:08 UTC
Created attachment 39065 [details]
A patch against the revision 1.54
Comment 19 tora3 2006-09-10 10:32:25 UTC
Created attachment 39066 [details]
A patch-applied file
Comment 20 hdu@apache.org 2006-09-13 11:49:50 UTC
Thanks a lot for updating the patch! Applied in CWS vclxcu02.
Comment 21 tora3 2006-09-13 12:01:27 UTC
Thanks a lot for the integration.
Comment 22 tora3 2006-09-18 15:47:55 UTC
HDU: I was wondering if the fix you had already made would be incorporated 
     into 2.0.4rcX in time for the final release.
Comment 23 hdu@apache.org 2006-09-19 11:44:13 UTC
I'm afraid the 2.0.4RC are closed for anything but the most serious regressions.
Comment 24 andreschnabel 2006-09-28 10:50:24 UTC
set keyword, to get more attention on the issue for 2.1
Comment 25 hdu@apache.org 2006-10-17 11:49:05 UTC
@es: please verify in CWS vclxcu02 using a japanese locale

@tora: please check the latest VCL.xcu in
http://util.openoffice.org/source/browse/util/officecfg/registry/data/org/openoffice/VCL.xcu?only_with_tag=cws_src680_vclxcu02
Comment 26 tora3 2006-10-17 14:10:51 UTC
hdu: Checked with some command line tools, it looks fine.
     Additionally I am going to check it with Windows, Solaris, and Linux.
Comment 27 tora3 2006-10-18 10:48:02 UTC
The fix has been confirmed with:
 - Windows 2000, OOo 2.0.4rc3 Japanese
   The fixed VCL.xcu makes good effects, as expected.

 - Solaris 10, OOo 2.0.4rc3 Japanese
   No effect is found, which is an expected behavior.

 - Fedora Core 5, OOo 2.0.1 Japanese
   The fixed VCL.xcu changes the default font name of Calc, Impress, and 
   Draw from font A- to font A+. The fix has to have no effect on any platform 
   but Windows. The phenomenon, however, has no problem on this issue. 

   I will file another issue on that. 
   In the list of VCL.xcu, the font A+ is listed prior to font X.
   Their spells encoded in UTF-8 are correct.
   A font file for font A+ is not installed on the system while 
   a font file for font X is installed here:
   /usr/share/fonts/japanese/TrueType/kochi-gothic-subst.ttf
   and listed in the font list of OOo's user interface.
   The default font should be font X instead of font A+ under the 
   circumstances. 

 A-: "MS P ゴシック" - misspell made long long ago.
 A+: "MS Pゴシック" aka "MS PGothic"
  X: "æ±é¢¨ã‚´ã‚·ãƒƒã‚¯" aka "Kochi Gothic"

In conclusion, the fixed VCL.xcu has been verified with Japanese environment.
Thanks a lot.
Comment 28 eric.savary 2006-10-18 16:22:23 UTC
Hi Tora!
Thanx a lot for submitteing an verifying this issue.
Let's bring this work to the end ;)
I reassign to you. Please flag as VERIFIED and be kind to reverify (and close)
it when it will be integrated in the master.

Thank you!
Comment 29 tora3 2006-10-18 17:12:31 UTC
Verified as described above.

i will check it again with an integrated build.
Vielen Dank!
Comment 30 tora3 2006-12-15 03:02:21 UTC
Verified with OOo_2.1.0rc2_20061130_Win32Intel_install_ja_wJRE.exe

Closing...