On Tue, Aug 14, 2012 at 10:57:08PM -0400, Peter Eisentraut wrote:
> On Tue, 2012-08-14 at 17:36 -0400, Bruce Momjian wrote:
> > On Tue, Aug 14, 2012 at 05:34:02PM -0400, Tom Lane wrote:
> > > Bruce Momjian <bruce@momjian.us> writes:
> > > > On Thu, Aug 4, 2011 at 09:00:11PM +0300, Peter Eisentraut wrote:
> > > >> On tor, 2011-08-04 at 14:44 +0200, Boszormenyi Zoltan wrote:
> > > >>> I meant a mass "sed -e 's/TRUE/true/g' -e 's/FALSE/false/g'" run
> > > >>> so all the ~200 occurrences of both "TRUE" and "FALSE" get
> > > >>> converted so the whole source tree is consistent.
> > >
> > > >> I would be in favor of that.
> > >
> > > > I have implemented this with the patch at:
> > > > http://momjian.us/expire/true_diff.txt
> > >
> > > Does this really do anything for us that will justify the extra
> > > back-patching pain it will cause? I don't see that it's improving
> > > code readability any.
> >
> > I think it is more of a consistency issue. There were multiple people
> > who wanted this change. Of course, some of those people don't backport
> > stuff.
>
> I guess you could argue this particular case without end, but I think we
> should be open to these kinds of changes. They will make future code
> easier to deal with and confuse new developers less (when to use which,
> do they mean different things, etc.).
>
> If back-patching really becomes a problem, we might want to look a
> little deeper into git options. I think the Linux kernel people do
> these kinds of cleanups more often, so there is probably some better
> support for it.
So what do we want to do with this? I am a little concerned that we are
sacrificing code clarity for backpatching ease, but I don't do as much
backpatching as Tom.
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://enterprisedb.com
+ It's impossible for everything to be true. +