X-Authentication-Warning: delorie.com: mailnull set sender to djgpp-bounces using -f From: "Thomas Mueller" Newsgroups: comp.os.msdos.djgpp Subject: Re: GNU Emacs DOS (DJGPP) port converts upper-ASCII characters to ASCII 127 Date: 7 Feb 2002 11:38:51 GMT Lines: 60 Message-ID: References: NNTP-Posting-Host: dial3-124.bluegrass.net (208.147.34.124) Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Trace: fu-berlin.de 1013081931 45106035 208.147.34.124 (16 [49635]) X-Mailer: NOS-BOX 2.05 To: djgpp AT delorie DOT com DJ-Gateway: from newsgroup comp.os.msdos.djgpp Reply-To: djgpp AT delorie DOT com from Eli Zaretskii: When you say ``which is in German'', what do you mean, exactly? That is, how are the German characters encoded? Are they encoded in the same codepage that is installed as the default on your system, or are they encoded in some other encoding, such as Latin-1? The newsletter is actually in the German language, as opposed to English, and specifies ISO-8859-1 charset. From the headers: Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit MIME-Version: 1.0 > The DJGPP port of Emacs assumes codepage-encoded non-ASCII characters by default, unless told otherwise. To see what codepage does your system report to Emacs, evaluate the variable dos-codepage (inside Emacs, type "M-: dos-codepage RET"). To tell Emacs to read a file encoded in latin-1, type "C-x RET c latin-1 RET" immediately before you type "C-x C-f" (or immediately before you click Files->Open). > All this is explained in the Emacs manual, of course. I did what you said (the latter, maybe not at the right time. But I had the same problem even after adding to _EMACS (set-terminal-coding-systsm 'iso-latin-1) I think I read that in a book, not referring to DOS version. But a smaller file, heise online newsletter by itself, about 10 KB, displayed correctly in ISO-8859-1. Also, with another file slightly exceeding 1.5 MB, DOS Emacs couldn't determine line numbers (not enough memory allocated by cwsdpmi?), but vim 6 had no such problem. DOS Emacs didn't harm that file because I didn't modify or save. Linux version of Emacs, in text mode, non-X-Windows, showed the latin-1 characters correctly, and saved correctly after I deleted parts that were spam. > No, not weird. Emacs thought that the file was encoded in some codepage, probably cp437 or cp850, but that wasn't true. So it saw some codes that are undefined in the codepage, and replaced them with 127, which is the glyph used for undefined characters (the Windows and X versions display an empty box instead). > It's not a bug, it's a usage error. A usage error, like I should have used vim 6 or elvis 2.1_4, or maybe setedit, instead of Emacs? Vim never messed up the upper-ASCII characters, nor did EPM in OS/2. What other text editor converts upper-ASCII characters to 7-bit and saves the file that way? Actually codepages 437 and 850 define all 256 characters, 128 = capital C-cedilla, 129 = lower-case u-umlaut, 130 = e-acute-accent, etc. I am accustomed to reading and mentally transliterating messages from the IBM charset I saw to ISO-8859-1 (not entirely). Even if Emacs displays the upper-ASCII characters as ASCII 127, Emacs shouldn't actually write these changes to the file. Apparently a lot of people prefer vi and relatives over Emacs.