[Libmesh-devel] Compiler warnings

2013-12-08 Thread Roy Stogner
Two questions: I've mostly been using -Wconversion privately to catch the occasional 32<->64 bit conversion bug... but I just graded the work of a student whose results got destroyed by an inadvertent implicit "double->int", and now I'm paranoid about my own codes. Would anyone mind me adding "-

Re: [Libmesh-devel] Compiler warnings

2013-12-08 Thread Kirk, Benjamin (JSC-EG311)
> On Dec 8, 2013, at 4:38 PM, "Roy Stogner" wrote: > > We don't currently add compiler warning flags to compilation in opt > mode. Is there a reason why not? I'd have expected the extra > compilation time spent generating warning flags to be dwarfed by the > optimization work itself. No, the

Re: [Libmesh-devel] Compiler warnings

2013-12-08 Thread Derek Gaston
I guess I'll vote against adding this to opt: but it's not a strong vote. My main worry is that we'll spew tons of warnings from includes further upstream... Derek Sent from my iPhone > On Dec 8, 2013, at 4:13 PM, "Kirk, Benjamin (JSC-EG311)" > wrote: > > > >> On Dec 8, 2013, at 4:38 PM,

Re: [Libmesh-devel] Compiler warnings

2013-12-08 Thread Roy Stogner
On Sun, 8 Dec 2013, Derek Gaston wrote: > I guess I'll vote against adding this to opt: but it's not a strong vote. My vote "for" isn't a strong vote, either. Ben seems fairly neutral, so unless someone else chimes in on the "for" side I guess I'll leave it be. No objections from anyone to me