Re: MD5 comparison fails in package-get

Jan> "Stephen J. Turnbull" <turnbull(a)sk.tsukuba.ac.jp&gt; writes: Jan> Yes. But that is wrong. If you apply my patch for the new Jan> version of i-f-c-i to XEmacs and EFS, it will work with EFS. OK, but somebody needs to make sure i-c-f-i gets inserted in all the relevant places. Jan> Isn't MD5 supposed to be just working regardless of the Jan> coding system the buffer is in? It goes to all kinds of pain Jan> to make sure it does something like that. md5 cannot "just work" regardless of the coding system the buffer is in. For example, mailers must be ready to handle inconsistent spool files containing say S-JIS and ISO-8859-1 simultaneously. So they will presumably read in the buffer in binary mode. But the presentation buffer may be in yet another coding system (here ISO-2022 makes sense) in order to save the message in a sensible encoding. So doing an md5 on that buffer will depend on whether you want the checksum to make sense for S-JIS (which none of the buffers are in!) or for ISO-2022-JP, which will depend on whether you are confirming the incoming checksum, or creating the outgoing one. How a sensible system should handle this is another question. As Darryl Okahata is finding out with crypt.el. -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 __________________________________________________________________________ __________________________________________________________________________ What are those two straight lines for? "Free software rules."