Re: old bootstrap error emerges again

[Eli Zaretskii]

> From: Dieter Deyke <address@hidden>
> Cc: address@hidden, address@hidden
> Date: Sat, 18 Mar 2006 09:22:56 -0700
> 
> I am seeing the "DOC" error too.
> My setup: english Windows XP Pro, MinGW-3.1.0-1, cygwin tool set, no msys.
> Logfile:
> [...]
> In toplevel form:
> url/vc-dav.el:29:1:Error: Cannot open doc string file 
> "c:/Users/deyke/emacs-build/work/etc/DOC"
> make[1]: *** [compile-SH] Error 1
> make[1]: Leaving directory `C:/Users/deyke/emacs-build/work/lisp'
> make: *** [bootstrap-gmake] Error 2

Thanks, this gives me something that I can work with.

Here's the problem: Emacs shouldn't even look for the file named DOC,
it should look for DOC-X.  This is because loadup.el has this
fragment:

      (if (memq system-type '(ms-dos windows-nt))
          (setq name (expand-file-name
                      (if (fboundp 'x-create-frame) "DOC-X" "DOC") "../etc"))
        (setq name (concat (expand-file-name "../etc/DOC-") name))
        (if (file-exists-p name)
            (delete-file name))
        (copy-file (expand-file-name "../etc/DOC") name t))
      (Snarf-documentation (file-name-nondirectory name)))

So, on Windows, since x-create-frame is bound, it calls
Snarf-documentation with the argument "DOC-X".  Snarf-documentation
then records this name in the variable internal-doc-file-name, and
that name is dumped in emacs.exe.  So when byte-compiling for some
reason calls for a doc file, Emacs should look for DOC-X.

Can you see where the above breaks on your system?  In particular,
when does Vdoc_file_name (defined in doc.c) gets assigned the file
name which ends with "DOC", not "DOC-X"?

Thanks.