Re: [Crash!] It did it again!

Martin> It looks like determine_real_coding_system() should be more Martin> careful about the characters it accepts after the "coding:" Martin> cookie. Martin> Here is a patch, only slightly tested, that should fix the Martin> problems with determine_real_coding_system(). Moriaka-san, Martin> can you please, please, review this patch, since you have been Martin> working on this code? Domo arigato. tomo> Sorry for late answer. From September 16, I went to India and Thai. tomo> That was interesting travel, but now I'm very tired and I have cold. tomo> So I don't have enough power to read detail of the patch. However I tomo> found a trivial problem: Yes. How about this definition? /* Valid characters in a coding system name - same as MIME charset name */ #define CODING_SYSTEM_NAME_CHARS \ "ABCDEFGHIJKLMNOPQRSTUVWXYZ" \ "abcdefghijklmnopqrstuvwxyz" \ "0123456789" \ "!#$%&'*+-.^_`{|}~" tomo> At least, shift_jis should be accepted. I don't know the right tomo> definition of coding-system-name, but it is a symbol so more tomo> characters are required. If we decide to accept every names of tomo> MIME-charset, one criterion is to use MIME syntax. Usually a name of tomo> MIME-charset should be a <token> of MIME syntax. According to RFC tomo> 2045, Since coding system names may be imported in future from MIME charsets, and since the MIME standard had to solve this same problem we have, it is reasonable to adopt the MIME charset valid characters as above. tomo> token := 1*<any (US-ASCII) CHAR except SPACE, CTLs, tomo> or tspecials> tomo> tspecials := "(" / ")" / "<" / ">" / "＠" / tomo> "," / ";" / ":" / "\" / <"> tomo> "/" / "[" / "]" / "?" / "=" tomo> ; Must be in quoted-string, tomo> ; to use within parameter values Perhaps you would like to look at this patch and put it into the source yourself, when it's been adequately tested (which I have not yet done). Martin