IBM codepage 858

Contents	Tables	External
Introduction	IBM00858	Home
	CP1004	Contact
History	References	XHTML

  Introduction

The official name of IBM codepage 850 after adding the EUR (euro) currency symbol as hex. D5 (dec. 213) is IBM00858 (858) aka PC-Multilingual-850+euro. Unfortunately old OS/2 browsers like WebExplorer v1.1, Netscape 2.02, and Netscape 4.61 don't support the complete glyph set in these codepages. Lynx 2.8.5 has almost no problems. Sorry, please use a display mode with at least 100 columns to view this document with Lynx.

Only WebExplorer knows ' (apostrophe), and this symbolic character reference entity was omitted in the first edition of XHTML 1.0. The second edition proposes to use the decimal reference ', and this works even with old Netscape versions. Old Netscape versions unlike WebExplorer also don't support ← (left arrow), → (right arrow), and ↑ (up arrow). Some symbolic references like ↕ (vertical arrow) are not yet defined for XHTML.

A similar problem is  , WebExplorer and Netscape 2.02 often handle this reference incorrectly, for more predictable results use  . WebExplorer displays ™ as (TM), but Netscape 4.61 supports only the decimal ™. This is also true for several other references incl. • (bullet •) and ƒ (florin resp. function of ƒ), but at least € is supported by Netscape 4.61.

Please note that old browsers don't support hex. references like €. Decimal references like € are much better, because even if the browser has no idea what to do it displays only ? (question mark) and not the complete undecoded hex. reference.

Netscape 2.02 displays &ltrif; (left-pointing triangle, filled) as &lt;rif; (<rif;). This and some other symbolic character references shown below are stolen from various ISO subsets of MathML. Lynx 2.8.5 supports some "old" Unicode characters like &#x25A0; (black square) instead of the similar &#x25AA; (black small square) to represent PC-Multilingual-850+euro character 254 (FEh).

Many programs support only subsets of Unicode, and Lynx 2.8.5 handles this problem rather well. However MathML introduced a symbolic character reference entity &squf; for the black small square &#x25AA;, therefore new programs forced to support only subsets of Unicode will probably include &squf; instead of the "old" black square &#x25A0;. Seven ambiguous "old" vs. MathML cases are highlighted as links in the following table (0Fh, 10h, 11h, 16h, 1Eh, 1Fh, and FEh).

An old version of this page included &#x220E; (end of proof) as third potential Unicode character to represent PC-Multilingual-850+euro character 254 (FEh). Unfortunately I don't find the ISO reference mentioning this character as &qed; anymore, maybe this was only in an expired draft. MathML 2.0 has no symbolic character reference entity for &#x220E; (end of proof), I only found a TeX character name qed in one of the MathML tables, which is &#x25AE; (&marker;, black vertical rectangle). In the final Unicode 4.0 names list &#x25AE; and &#x220E; are cross-referenced.

MicroSoft                  IBM
1A 0x001A SUBstitute       1A 0x001C File Separator
1C 0x001C File Separator   1C 0x007F DELete
7F 0x007F DELete           7F 0x001A SUBstitute

ftp://ftp.unicode.org/Public/MAPPINGS/VENDORS/MISC/IBMGRAPH.TXT
ftp://ftp.unicode.org/Public/MAPPINGS/VENDORS/IBM/README.TXT

   Entity names shown as links are not (yet) defined for XHTML. CP858 codes shown as links are approximations..

legend	°	not visible with any browser	?	undecoded decimal entity
	¹	visible with WebExplorer 1.1	²	visible with OS/2 Netscape 2.02
	³	visible with all tested browsers	ª	visible with OS/2 Netscape 4.61

  History

Men at work.. This is at least the fourth version of this page, but now I finally understood the huge difference between a decimal character reference entity like &#128; and the character 128 in documents with a charset which is not Unicode or one of its encodings and proper subsets. To put it simply, a numeric character reference entity is always Unicode (or Latin-1 in HTML 2, but Latin-1 is a proper subset of Unicode). Therefore &#128; is always a control character, and whatever some browsers do in this case is undefined. However the character € (byte 80h, decimal 128) is the character at position 128 of the declared document charset, if this charset really has a single byte encoding. And because this page is now finally declared as windows-1252 instead of us-ascii the byte 128 resp. encoded character € is an Euro, even if some browsers don't know how to handle this charset.

BTW, I've also found a weird way to use symbolic character reference entities defined for MathML 2.0 in a XHTML 1.0 transitional document, see the header of the source of this file. Valid XHTML, apparently.

  References

• W3C I18N: character sets  /  character model  /  resources.
• W3C MathML: Characters by Name (180 KB)  /  Unicode blocks.
• Extensible Markup Language XML 1.1 (latest W3C version).
• IANA list of character sets and registrations not defined in RfCs.
• Unicode in XML and other Markup Languages: TR #20 (revision 7).
• Unicode 4.0 final names list (text, 725 KB), index [A-D] [E-N] [O-Z].
• IETF Policy on Character Sets and Languages: RfC 2277, BCP 18.
• IANA Charset Registration Procedures: RfC 2278, BCP 19.
• UTF-8, a transformation format of ISO 10646: RfC 3629, STD 63.
• Using International Characters in Internet Mail.
• Jukka Korpela: Characters and encodings, shy, Latin-9 ("Latin-0").
• A.J. Flavell: Checklist for HTML character coding.
• Alan Wood: Using special characters from Windows Glyph List 4.
• Ken Borgendale: OS/2 Warp Internationalization, OS/2 Glyph List .
• Old codepages 1004 OS2LATIN1 and 1252 MS-ANSI without Euro.
• Windows codepages OEM 850 and 858, SBCS 1252, and more.
• IBM ICU converter explorer: UTF-8, 437, 850, 858, 1252, and more.
• Collection of character sets by Kosta Kostis with Unicode names.
• HTML character set by W.I.Kaye with GIFs for critical characters.
• I.S.Graham: Greek, math., and miscellaneous HTML entity tables.
• I.S.Graham: Historical test document with HTML 3.2 entities.
• W3C HTML 4.01 test suite: Latin 1 characters.
• Letter database: codepage 850 (note hex. D5 resp. dec. 213).
• Letter database: codepage 850 to Latin 1 translations.
• Letter database: windows-1252 to Latin 0 translations.
• Unicode translation tables: CP850.TXT, IBMGRAPH.TXT.
• KEDIT IBM850 conversion: xhtml.kex (applied on itself ).
• UTF-8 translations for REXX: utf-8.cmd (437, 850, 1004).
• Incompatible OS/2 file name characters (437, 850, 858, 1004).

Last update: 8 Mar 2007 20:00 by F.Ellermann