Nicolas Richard <theonewiththeevillook@yahoo.fr> writes:
> I was bootstrapping and ran into an error (shown below).
>
> The error shows a command that "make" is trying to run -- but when I run
> that manually (from the lisp/ subdir), I don't get the error:
>
> EMACSLOADPATH= '../src/emacs' -batch --no-site-file --no-site-lisp --eval "(setq max-lisp-eval-depth 2200)" --eval '(setq load-prefer-newer t)' -f batch-byte-compile net/shr.el
>
> If I remove net/shr.elc and do "make" again, I get the error again. So I
> can reliably reproduce the error, but not in a way that I can run under
> gdb...

I tested more thoroughly and came up with the conclusion that my patch
doesn't solve anything because I can't reproduce the build problem
anymore. I did this:
- I used git-new-workdir to get two copies of the emacs tree
- In one of them, I applied an extension of the patch I posted
  previously : it mostly modifies a few calls to (debug ...) into (debug
  nil ...) (See attachment.)
- In the other I used savannah/master as of
     commit b01cf822c5ac044bb550231c811a062d3a8cbeec
     Author: Glenn Morris <rgm@gnu.org>
     Date:   Tue Apr 21 06:18:09 2015 -0400

Conclusion : both trees compiled just fine.

I can only assume that I had leftover files, which caused the problem
(even though I'm almost sure I used "make bootstrap" when doing my tests
initially).

I'm now closing the bug and will reopen it if I see the initial problem again.

Should the attached patch (or some variation of it) be applied anyway ?