Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!husc6!sri-unix!sri-spam!ames!ucbcad!ucbvax!decvax!minow From: minow@decvax.UUCP (Martin Minow) Newsgroups: comp.lang.c Subject: Re: Compatibility with EBCDIC Message-ID: <136@decvax.UUCP> Date: Sun, 23-Aug-87 19:50:09 EDT Article-I.D.: decvax.136 Posted: Sun Aug 23 19:50:09 1987 Date-Received: Mon, 24-Aug-87 03:37:16 EDT References: <855@tjalk.cs.vu.nl> <2683@hoptoad.uucp> <1519@sol.ARPA> <8452@utzoo.UUCP> Reply-To: minow@decvax.UUCP (Martin Minow) Organization: Digital Eq. Corp. - Merrimack NH. Lines: 31 Several collegues comment on EBCDIC. Here is my contribution: 1. The main problem with EBCDIC is the lack of a common standard for the character semantics. Several years ago, the ANSI C committee distributed at least three incompatible EBCDIC "standards." 2. I don't believe that EBCDIC has been standardized by ANSI -- i.e., its contents are under the control of a single manufacturer. Of course, this might be an advantage. Unfortunately (as noted above), this has resulted in unspecified characters being added in different places. 3. I used to think that the fact that the alphabet was discontiguous was a problem. Unfortunately, ISO Latin 1 added two graphic characters (multiply and divide) in the middle of the "right-hand" alphabetic portion. 4. Some of the strangeness of EBCDIC resulted from its being an evolution of an earlier (BCD) standard, and the close connection of both standards to punch card codes. 5. Before berating EBCDIC for problems in sorting, it should be noted that there is no commonly acceptable standard for character order. For example, In Swedish, the letters run a..z, a-ring, a-two_dots, o-two_dots; while in Danish, a-ring follows o-two_dots. Any implementation that needs to order text alphabetically must apply fairly sophisticated procedures that are both language and country- specific. ANSI C attempts to solve this by adding localization macros and the strcoll() function. Martin Minow decvax!minow