Re: Trouble with nnir.el ("Invalid function ...")

[kai.grossjohann@gmx.net (Kai Großjohann)]

jesseh@cs.kun.nl (Jesse F. Hughes) writes:

>> | Debugger entered--Lisp error: (wrong-type-argument arrayp nil)
>> |   aset(nil 0 1)
>> |   (mail-header-set-number novitem idx)
>> |   (while (not (null artlist)) (setq art (car artlist)) (or (numberp art) (nnheader-report ... "nnir-retrieve-headers doesn't grok message ids: %s" art)) (setq artitem (nnir-artlist-article nnir-artlist art)) (setq artrsv (nnir-artitem-rsv artitem)) (setq artfullgroup (nnir-artitem-group artitem)) (setq artno (nnir-artitem-number artitem)) (setq artgroup (gnus-group-real-name artfullgroup)) (setq server (nnir-group-server artfullgroup)) (set-buffer nntp-server-buffer) (nnir-possibly-change-server server) (case (setq foo ...) (nov ... ... ...) (headers ... ... ...) (t ...)) (mail-header-set-number novitem idx) (mail-header-set-from novitem (mail-header-from novitem)) (mail-header-set-subject novitem (format "[%d: %s/%d] %s" artrsv artgroup artno ...)) (push novitem novdata) (setq artlist (cdr artlist)) (setq idx (1+ idx)))


Really strange.  Have a look at this code:

	(nnir-possibly-change-server server)
        (case (setq foo (gnus-retrieve-headers (list artno) artfullgroup nil))
          (nov
           (goto-char (point-min))
           (setq novitem (nnheader-parse-nov))
           (unless novitem
             (pop-to-buffer nntp-server-buffer)
             (error
              "nnheader-parse-nov returned nil for article %s in group %s"
              artno artfullgroup)))
          (headers
           (goto-char (point-min))
           (setq novitem (nnheader-parse-head))
           (unless novitem
             (pop-to-buffer nntp-server-buffer)
             (error
              "nnheader-parse-head returned nil for article %s in group %s"
              artno artfullgroup)))
          (t (nnheader-report 'nnir "Don't support header type %s." foo)))
       ;; replace article number in original group with article number
        ;; in nnir group
        (mail-header-set-number novitem idx)

I think the error is coming from there.  Now, the case statement
calls gnus-retrieve-headers and then looks at the result.  If the
result is nov, novitem is set and then checked for non-nil-ness.  If
the result is headers, then also novitem is set and then checked for
non-nil-ness.  And if the result was neither nov nor headers, then an
error is generated.

Thus, I expect that the last line from the above snippet is never
reached for novitem being nil.

What's wrong in this reasoning?

Could you position point inside that function and do M-x edebug-defun
RET and then use SPC to single-step through it to see what's happening?
-- 
~/.signature