svn commit: r342510 - head/games/cutemaze

[Max Brazhnikov]

On Tue, 04 Feb 2014 17:31:39 +0100 Mathieu Arnold wrote:
> +--On 4 février 2014 14:36:38 +0000 Max Brazhnikov <makc at freebsd.org>
> wrote:
> | On Tue, 04 Feb 2014 14:37:07 +0100 Mathieu Arnold wrote:
> |> +--On 4 février 2014 13:05:39 +0000 Max Brazhnikov <makc at freebsd.org>
> |> wrote:
> |> | 
> |> | This part should be reverted.
> |> | Folks, please *review* what makeplist generates and don't commit it
> |> | blindly.
> |> 
> |> I for one would love to have feedback on what the evolved makeplist
> |> should do better :-)
> | 
> | It's way too aggressive for PLIST_SUB and this leads to false
> | substitutions  sometimes. bsd.qt.mk defines a number of QT_*DIR, but not
> | all of them should  be used (actually it depends on particular port).
> | We could help makeplist by telling explicitly which substitutions are
> | right, e.g. by specifying OPTIONS_SUB.
> 
> Most substitutions are right, I mean, in a perl port, I want PERL_VER, and
> SITE_PERL, and PERL5_MAN3, and... there are a few like QT_BINDIR which
> shouldn't go in it, but it feels more like having a blacklist than a
> whitelist.

Blacklist can be longer than whitelist :).

> Patch welcomes.

I'll think about it. The problem is not specific to Qt ports. I've already been
beaten when KDE4_VERSION was improperly used for library version (just
coincidence). In fact, QT_BINDIR could be easily hidden for other ports,
but I'd like to see general solution.

Max