Alex Kost <[email protected]> skribis: > Ludovic Courtès (2016-05-17 12:12 +0300) wrote: > >> Alex Kost <[email protected]> skribis: >> >>> Ludovic Courtès (2016-05-16 15:45 +0300) wrote: > [...] >>>> $ git describe >>>> v0.10.0-798-g8a7680a >>>> $ tar tvf $(./pre-inst-env guix build -S emacs) |grep 'autoload\.el' >>>> -rw-r--r-- root/root 37292 1970-01-01 01:00 >>>> emacs-24.5/lisp/emacs-lisp/autoload.el >>>> -rw-r--r-- root/root 37127 1970-01-01 01:00 >>>> emacs-24.5/lisp/emacs-lisp/autoload.el.orig >>>> -rw-r--r-- root/root 22624 1970-01-01 01:00 >>>> emacs-24.5/lisp/emacs-lisp/autoload.elc >>>> >>>> Upstream’s tarball already includes those three files. >>> >>> IIUC this source is after applying our patches (including >>> "emacs-source-date-epoch.patch"): >>> >>> - “autoload.el.orig” is the original file from the upstream; >> >> Indeed, this one isn’t present in upstream’s tarball: >> >> $ wget -q -O - ftp://ftp.gnu.org/gnu/emacs/emacs-24.5.tar.xz | tar tJvf - | >> grep 'autoload\.' >> -rw-rw-r-- nico/nico 37127 2015-04-02 09:23 >> emacs-24.5/lisp/emacs-lisp/autoload.el >> -rw-r--r-- nico/nico 22624 2015-04-08 19:16 >> emacs-24.5/lisp/emacs-lisp/autoload.elc >> >> How come we’re introducing this one? I thought ‘patch’ did not produce >> .orig files unless the patch failed to apply, but here the patch >> correctly applies, only with a small offset (can be seen by running >> ‘guix build -S emacs --check’): >> >> patching file lisp/loadup.el >> patching file lisp/emacs-lisp/autoload.el >> Hunk #1 succeeded at 361 (offset -17 lines). > > Indeed, I also didn't know that "patch" produces such .orig files when a > patch applies with offset. > >> Apparently we have to use ‘--no-backup-if-mismatch’ to avoid that. > > You even found the flag, thanks! So is it OK to apply the attached > patch to core-updates?
The patch LGTM, but since this is a full-rebuild change, I prefer keep it for the next core-updates cycle. Let’s try not to forget about it. :-) Thanks, Ludo’.
