"Stephen J. Turnbull" <turnbull(a)sk.tsukuba.ac.jp&gt; writes: Here's the fix. 2000-11-10 Yoshiki Hayashi <yoshiki(a)xemacs.org&gt; * info.el (Info-extract-menu-node-name): Comma is a delimiter. Index: info.el =================================================================== RCS file: /usr/CVSroot/XEmacs/xemacs/lisp/info.el,v retrieving revision 1.15.2.21 diff -u -r1.15.2.21 info.el --- info.el 2000/11/06 05:44:18 1.15.2.21 +++ info.el 2000/11/10 03:51:00 ＠＠ -2008,7 +2008,7 ＠＠ ;; Kludge. ;; Allow dots in node name not followed by whitespace. (re-search-forward - (concat "\\(([^)]+)[^." + (concat "\\(([^)]+)[^.," (if multi-line "" "\n") "]*\\|\\([^.,\t" (if multi-line "" "\n") -- Yoshiki Hayashi

Klaus> This form and some others are also contained in Texinfo Klaus> files at places which aren't intended as cross-refence or Klaus> menu. If we change the definition of a cross-reference, Klaus> this might have an influence on the probability and the Klaus> extent of such false cross-references. Agreed. But I don't really care about that, even if they fool automatic validators. The real problem is (if I understand correctly) node names of the form "Sugar, spice, and everything nice" will be truncated to "Sugar" by Yoshiki's patch, and probably not referenced correctly. Is that true? If so, Yoshiki's patch should not be applied, and we should fix the Texinfo sources instead. (I can see easy workarounds for uses of period in node names, but not for this use of comma.) Klaus> (cond ((string-match "XEmacs" (emacs-version)) Klaus> (defun Info-navigate () (Info-follow-nearest-node (point)))) Klaus> (t Klaus> (defun Info-navigate () (Info-follow-nearest-node)))) Yuck! Do we need this gratuitous difference in this function? Shouldn't we make the argument optional and default to point? -- 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 straight lines for? "XEmacs rules."

Yoshiki> Yes. But IIRC, Texinfo doesn't allow comma and period in Yoshiki> node names so stopping at first dot or comma is the right Yoshiki> thing to do. Yeah, you're absolutely right. Texinfo uses commas as separators in the ＠node line and provides no way of escaping them. (Forgetting that was dumb....) The Texinfo documentation prohibits use of commas (and apostrophes) in node names. Presumably periods are forbidden, too. ;-) OK, I withdraw my objection; Yoshiki's patch is correct. "Fake" cross-references don't bother me enough to veto, only broken ones. (Assuming that cross references are at least smart enough to check for the *note prefix, etc. If not, that's what we should fix.) Klaus? Are we missing something? Info-follow-nearest-node seems very unreliable, unless it is bound to a key used in a map bound to an extent (such as the one used to identify, and highlight, a cross- reference when the mouse is near it). For example, in one run, it terminated in (libc) Program Basics, Program Termination, Termination Internals on ------------------------------------------------------------------------ * A process exit status is saved to be reported back to the parent process via `wait' or `waitpid'; see *Note Process Completion::. ------------------------------------------------------------------------ with messages ------------------------------------------------------------------------ Top node... Top node...done Selecting menu item Termination Internals... Selecting menu item Termination Internals...done Selecting menu item A process exit status is saved to be reported back to the parent process via `wait' or `waitpid'; see *Note Process Completion... No such node: A process exit status is saved to be reported back to the parent ------------------------------------------------------------------------ I don't see that (or any of the other runs I tried) as evidence that Yoshiki's patch is wrong. The correct reference is found by the mouse, and Info-follow-reference (`f') insists on user identification of the node to jump to when outside of the reference itself. I-f-n-n is just too naive to correctly parse the text of the node. Vetoing Yoshiki's patch, thus breaking the legal format "*Note Node Title: (file)Node Name, explanatory text." is not going to save I-f-n-n in any of the cases I saw. Probably fixing I-f-n-n to look for extents instead of regexps would get rid of almost all anomalies of the kind you're talking about. Yoshiki> The only thing we can do is just to pray every Texinfo Yoshiki> file authors write them right We should be able to run a verifier. (nconc queue '("texinfo verifier")) Yoshiki> and to hope we can some day get rid of Info format. ;-) Unfortunately, generic HTML is hardly an improvement. -- 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 straight lines for? "XEmacs rules."