Hello,

I wanted to give the "feature/native-comp" branch a try on an old 32-bit
laptop I have lying around[1]; unfortunately compilation ended abruptly
on the first couple of ELC+ELN steps (emacs-lisp/macroexp.elc and
emacs-lisp/cconv.elc).  See [2] and [3] for fairly uninformative
backtraces; I installed libgccjit0-dbg between [2] and [3], hoping that
would make the backtrace more precise, but obviously this didn't work.

(I see that the second run started with lisp/abbrev.elc and
international/titdic-cnv.elc instead of emacs-lisp/macroexp.elc and
emacs-lisp/cconv.elc; is this expected because make(1) processes
dependencies in an arbitrary order, or is it possible that the first
aborted run left artifacts that lead make(1) to believe that the targets
are up-to-date?  FWIW there is a new folder called
eln-i686-pc-linux-gnu-639e57b3ce60d5a9 under lisp/emacs-lisp, which
contains 0-byte files named cconvPDRUUG.eln and macroexpWFTb2Q.eln.)


Is my setup supposed to be supported?  I haven't read everything related
to this new feature in detail, so maybe I missed a memo warning against
trying it out on i686.  Otherwise, is there anything I can do to make
this easier to debug?


Thank you for your time.


[1] $ uname -a
    Linux little-buster 4.19.0-8-686-pae #1 SMP Debian 4.19.98-1+deb10u1 (2020-04-27) i686 GNU/Linux

    $ apt policy libgccjit0
    libgccjit0:
      Installed: 8.3.0-6

    $ git rev-parse HEAD
    02e3ffad6d9f757599bb441704b6cf6494183174

[2]