[PATCH] Last assault on binary-process-output

This might break packages which expect NTEmacs variable binary-process-output to do things. Packages in XEmacs sumo tarball are clear, please test 3rd party packages which you are using. The correct fix is: Whenever a construct (let ((binary-process-output t)) ... ) is used, add one more binding: (let ((binary-process-output t) (coding-system-for-read 'binary)) ... ) This way, old compatibility is not broken and XEmacs compatibility is restored. I'll delete the code and definitions unless it is going to break an overwhelming number of packages (I do not believe but I made this change in a revertable way). NT testers please give your feedback soon if possible. Big K 2000-01-26 Kirill 'Big K' Katsnelson <kkm(a)dtmx.com&gt; * callproc.c (Fcall_process_internal): Ignore Vbinary-process_output. Index: callproc.c =================================================================== RCS file: /usr/CVSroot/XEmacs/xemacs/src/callproc.c,v retrieving revision 1.29.2.11 diff -u -r1.29.2.11 callproc.c --- callproc.c 2000/01/24 07:36:19 1.29.2.11 +++ callproc.c 2000/01/26 21:33:08 ＠＠ -501,16 +501,19 ＠＠ if (nread == 0) break; +#if 0 #ifdef DOS_NT /* Until we pull out of MULE things like make_decoding_input_stream(), we do the following which is less elegant. --marcpa */ + /* We have pulled them out. -- kkm */ { int lf_count = 0; if (NILP (Vbinary_process_output)) { nread = crlf_to_lf(nread, bufptr, &lf_count); } } +#endif #endif total_read += nread;