gnucash master: Multiple changes pushed
Geert Janssens
janssens-geert at telenet.be
Fri Jul 4 10:30:34 EDT 2014
On Friday 04 July 2014 03:03:27 jralls at ceridwen.us wrote:
> > On July 3, 2014 at 5:22 PM Geert Janssens <janssens-geert at telenet.be> wrote:
> > On Thursday 03 July 2014 20:07:46 John Ralls wrote:
> > > On Jul 3, 2014, at 7:25 PM, Geert Janssens
> > > <janssens-geert at telenet.be>> >
> > > wrote:
> > > > Same issue but this time it may need some more careful testing.
> > > > TXN_TYPE_NONE may be used in other locations where '\0' is
> > > > assumed
> > > > instead of NULL (guile comes to mind). I don't have time to go
> > > > deeper into this right now so I'll just revert back to gcc for
> > > > the
> > > > time being.
> > > >
> > > > John, you don't *have* to fix this while you're on holidays :).
> > > > There is an easy alternative and you can revisit it later if
> > > > you
> > > > like.
> > > >
> > > > Oh, just a side question: do we want maint to be buildable with
> > > > clang as well ? If so we may need to backport (cherry-pick) the
> > > > current round of fixes.
> > >
> > > Maybe you should just add -Wno-non-literal-null-conversion to
> > > CFLAGS.
> > > Copy the code I wrote yesterday for -Wno-deprecated-register
> > > without
> > > the AC_LANG([C++]) block.
> >
> > I could do that but prefer not to until we know for sure the
> > warning is>
> > irrelevant. I can imagine clang doesn't throw this warning just for
> > fun. Perhaps what we did was generally accepted before but is now
> > considered risky ? I don't know. If we can avoid the warnings by
> > small code modifications I'd prefer that approach.
<snip>
>
> That's one way of looking at it, though there have been complaints
> that Clang enforces a particular coding style via somewhat bogus
> errors and warnings. In C 0, '\0', FALSE, and NULL are just different
> ways of saying the same thing; in C++11 we should say nullptr when we
> mean a null pointer, but for C code I think it's an unnecessary
> exercise to go through the code to make sure we always use NULL.
>
With that you just did the analysis I was suggesting. I think you are right that in this case
the warning is just too cautious and not useful in our code.
So I have chosen to disable it (on condition the current compiler understands it) as you
proposed.
The maint backporting still needs to be done though.
Geert
More information about the gnucash-devel
mailing list