On honoring user's ${SHELL} for BUILD_ENV_SHELL default (was: CVS commit: pkgsrc/mk/build)

[Leonardo Taccari <leot%NetBSD.org@localhost>]

Joerg Sonnenberger writes:
> On Sun, Apr 26, 2020 at 07:56:23AM +0200, Roland Illig wrote:
> > On 26.04.2020 00:07, Greg Troxel wrote:
> > > Roland Illig <roland.illig%gmx.de@localhost> writes:
> > > 
> > > > On 25.04.2020 23:38, Joerg Sonnenberger wrote:
> > > > > On Sat, Apr 25, 2020 at 09:36:17PM +0000, Roland Illig wrote:
> > > > > > Module Name:        pkgsrc
> > > > > > Committed By:       rillig
> > > > > > Date:               Sat Apr 25 21:36:17 UTC 2020
> > > > > > 
> > > > > > Modified Files:
> > > > > >     pkgsrc/mk/build: build.mk
> > > > > > 
> > > > > > Log Message:
> > > > > > mk/build/build.mk: use the login shell by default for build-env
> > > > > 
> > > > > That seems highly counterproductive.
> > > > 
> > > > Why?
> > > 
> > > Not really clear on the details, but it seems really obvious that all
> > > build system and scripts should use a basic Bourne sh, nominally
> > > /bin/sh.  What if a user was using csh?
> > 
> > Then they'd get an interactive csh for inspecting the build environment
> > and why the build failed. I see nothing wrong with that.
> > 
> > "bmake build-env" is an interactive target that is meant for interactive
> > debugging. It is not part of the regular build process.
>
> Yes, it is meant for interactive debugging. Defaulting to a completely
> different shell defeats the purpose as you can no longer easily C&P
> commands into it etc. If you really want a different shell, you can
> start that yourself.
>

I strongly agree here: IMO it's much better to have a much possible
similar environment by default instead of your favourite shell, and -
if that was desiderable - the user could always adjust that via
BUILD_ENV_SHELL.

Roland, can this please be reverted to the old behaviour?