Re: Bootstrap broken between 2893cb6a21af3384cf5d6dc2b6bbdd5ebba8e1ad and 469fe6a040d6df612c00ea169670c6d8d7e58563

[Kévin Le Gouguec]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Andreas Schwab <schwab@linux-m68k.org> writes:
>
>> make[3]: Leaving directory '/home/abuild/rpmbuild/BUILD/emacs-29.0.50/lisp'
>> Debugger entered--Lisp error: (file-missing "Cannot open load file" "No such 
>> file or directory" "rmail-loaddefs")
>>   require(rmail-loaddefs)
>>   eval-buffer(#<buffer  *load*-463495> nil 
>> "/home/abuild/rpmbuild/BUILD/emacs-29.0.50/lisp/mai..." nil t)  ; Reading at 
>> buffer position 1601
>
> Huh.  I'm unable to reproduce this with "make -j16 bootstrap" on
> Debian/bullseye.

100% reproducible over here, on openSUSE tumbleweed (make -j12
bootstrap) and Ubuntu 20.04 (make -j16 bootstrap).

> john muhl <email@johnmuhl.mx> writes:
>
>> looks like what i ran into in
>> https://lists.gnu.org/archive/html/emacs-devel/2021-12/msg01587.html
>>
>> as noted there switching to -j1 lets the build complete. my bisection
>> traced the first broken commit to 35dbc79
>
> That commit seems entirely unrelated to generating rmail-loaddefs, but
> perhaps it changes something in the order the compilation happens in?

I bisected as well before reaching out to emacs-devel and came to the
same conclusion as John (35dbc79 is the first commit that misbehaves)
and you (looks thoroughly unrelated).

So both your conclusion (something incidentally changed in the order of
compilation) and Eli's in <835yrng2bo.fsf@gnu.org> (missing Makefile
dependency) seem likely?