Re: Build failure for Emacs master

[Paul Eggert]

On 04/13/2016 01:12 PM, Angelo Graziosi wrote:
> why now
>
> ./autogen.sh
> ...
> You can now run './autogen.sh git'.
>
> Really we need to run './autogen.sh git'?

Yes you need to, and yes it is silly.  This extra step was put in as a 
result of the email thread starting here:

http://lists.gnu.org/archive/html/emacs-devel/2016-02/msg00012.html

I preferred a simpler system where ./autogen.sh just works, without any 
extra steps, but I was outvoted.

> Maybe I am wrong but INSTALL.REPO, which should contain the "more 
> details on building Emacs", does not document this nor the "git" 
> option for autogen.sh nor the other "all" option.

Although INSTALL.REPO contains "more details", it shouldn't contain 
*all* details (as there are too many). It's not clear that this 
particular detail is worth recording in INSTALL.REPO.

> Maybe I am only adopting the wrong procedure to build Emacs

Yes, as you used a procedure that is not recommended in INSTALL.REPO. 
You would have avoided the './install.sh git' glitch if you had run 
plain 'make', or had run './autogen.sh all', which are the two methods 
that INSTALL.REPO suggests.