Matthieu Moy <[EMAIL PROTECTED]> writes: > Masatake YAMATO <[EMAIL PROTECTED]> writes: > >>> This is perfectly suited if the buffer is the output of "tla changes", >>> because the beginning of the buffer is the list of files, and the end >>> the concatenation of patches. >>> >>> It's still buggy in some cases, in particular, parsing the output of >>> "tla show-changeset" is more difficult. >> >> I see. So the temporally solution is >> >> if "tla changes" is invoked (point) should be used. > > Yes, > >> else if "tla changesets" is invoked (point-min) should be used. > > No, because higher in the code, you can see > > (setq header (buffer-substring-no-properties > (goto-char (point-min)) > ^^^^^^^^^^^^^^^^^^^^^^^ > (progn (re-search-forward "^[^*]" nil t) > ^^^^^^^^^^^^^^^^^^ > (beginning-of-line) > (point))))
OK, was my mistake: This piece of code is not executed for tla show-changeset output. Bug is fixed in [EMAIL PROTECTED] -- Matthieu
