Re: pkg/50279 (emulators/tme build failure on NetBSD/amd64 7.0)

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

The following reply was made to PR pkg/50279; it has been noted by GNATS.

From: Joerg Sonnenberger <joerg%britannica.bec.de@localhost>
To: Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>
Cc: joerg%NetBSD.org@localhost, gnats-bugs%NetBSD.org@localhost, pkg-manager%netbsd.org@localhost,
	gnats-admin%netbsd.org@localhost, bsiegert%NetBSD.org@localhost
Subject: Re: pkg/50279 (emulators/tme build failure on NetBSD/amd64 7.0)
Date: Wed, 14 Oct 2015 21:24:52 +0200

 On Thu, Oct 15, 2015 at 01:37:31AM +0900, Izumi Tsutsui wrote:
 > joerg@ wrote:
 > 
 > > > > I am doing my testing on NetBSD current with clang. I don't know right
 > > > > now what is pulling in stdint.h. I am not saying it is enough, just that
 > > > > I can't reproduce the problem in my test environment.
 > > > 
 > > > Why didn't you write it in your commit log?
 > > > What you wrote in the log ("explicitly") was very confusing.
 > > >  >> Explicitly pull in limits.h, it sometimes hasn't been included yet.
 > > > 
 > > > Anyway, please don't commit random changes you can't test.
 > > 
 > > Fine, I'll just leave things broken the next time.
 > 
 > If you mean the original "(((type) 0) - ((type) 1))" was broken,
 > it's fine to leave it.
 
 No, the original was ((type)0 - (type)1) << x, for some value of x. The
 C standard is pretty explicitly about that not being valid.
 
 > Making build broken even on the default NetBSD environment is
 > much worse than that, i.e you should revert your untested change.
 
 I have an environment where it works. As such, it was tested. I could
 not verify your problem, so I tried to apply a fix for that which
 doesn't make the situation any worse. I won't try to do that in the
 future, since it has become obvious that you prefer to do any fixes
 yourself.
 
 Joerg