Re: [PATCH] macos: do let the build find the gettext headers/libraries/msgfmt

[Johannes Schindelin <Johannes.Schindelin@gmx.de>]

Hi Carlo,

On Sun, 26 Apr 2020, Carlo Arenas wrote:

> On Sat, Apr 25, 2020 at 5:59 AM Johannes Schindelin
> <Johannes.Schindelin@gmx.de> wrote:
> >
> > On Thu, 23 Apr 2020, Eric Sunshine wrote:
> >
> > > On Thu, Apr 23, 2020 at 12:17 PM Eric Sunshine <sunshine@sunshineco.com> wrote:
> > > > I'm slightly leery of seeing these applied globally on Mac OS in
> > > > config.mak.uname since various package managers on Mac OS install
> > > > packages in wildly different locations, and these settings might
> > > > cause the wrong version of a package to be picked up if a user has
> > > > a preferred version installed elsewhere.
> > >
> > > As a follow up, although slightly leery of applying this change
> > > globally to config.mak.uname, I don't necessarily oppose it either.
> > > Considering how widely adopted Homebrew is on Mac OS, baking in a
> > > bit of Homebrew-specific knowledge would make it easier for a Git
> > > developer to get up and running by eliminating some of the manual
> > > fiddling and configuration currently necessary.
> >
> > I share your concern. But in contrast to Fink and DarwinPorts, we have
> > no Homebrew-specific knob in the Makefile (does this mean that we
> > expect users to use Homebrew by default?).
>
> IMHO the reason why there is no NO_BREW flag is because brew decided to

That is not an opinion, but an assumption.

> use the "default" directory instead of one of their own (which is why it
> has so many issues with unlinked files that might or not conflict with
> the system, like gettext), this is also why a NO_BREW flag (similar to
> the other two introduced since 8eb38cad44 (Disable linking with Fink or
> DarwinPorts., 2006-07-24)) wouldn't make sense.
>
> my assumption is also that most people in macOS use brew nowadays

Yes, that is an assumption. One I share, but I would say that it is more a
suspicion because I do not really want to act on it.

> instead of fink, darwinports or any of the other alternatives, but there
> is also people that use none and it is that complication why my proposed
> patch was in what would seem like the wrong place to begin with.  for
> more context in that last point see: 59f8674006 (Move Fink and Ports
> check to after config file, 2006-12-12).
>
> on that last point, I am afraid there is still at least a problem that
> needs addressing, but will comment in patch instead for easy of
> reference.

I would have preferred a straight answer right here than a reference to
something I have not received (yet?).

Ciao,
Dscho