On Sun, Dec 13, 2009 at 2:45 PM, Albert ARIBAUD <albert.arib...@free.fr> wrote: > Øyvind Harboe a écrit : >> >> On Sun, Dec 13, 2009 at 12:13 AM, Carsten Breuer >> <carstenbreueropen...@textwork.de> wrote: >>> >>> Hi all, >>> >>> >>> "treat warning as errors" (-wall) is a fine thing but sometimes breaks >>> the build.. >> >> This is a good thing to have as a default behavior, we want warning >> fixes fed back. It cleans up the code and finds real bugs. > > Reminds me of my teachers' motto regarding C: roughly rendered, it went > "when the C compiler emits an error, that's because it cannot turn your > source code into its equivalent executable form. When it emits warning, > that's because it will not." -- to which experience prompts me to add "... > and when it emits neither errors nor warnings, that's because your source > code does not do what you think it does." > > -Wall is the minimum that should be applied to source code. Running it > through Splint (or some other verification tool) would not hurt either.
Is there an open source lint tool that's worthwhile to use? Did you try splint on OpenOCD? >> You *can* disable error on warnings at configure time. > > I personally would never turn this off, and I hate when a codebase requires > it to be turned off because someone was not bothered enough by some warning > to fix the root cause. There are places where the policy is "when changing > code, verify that you dont get *more* warnings than before". :/ There are legitimate uses: e.g. recompiling an *old* version with a newer compiler can yield warnings that are fixed in latest version of openocd. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
