Gerald I. Evenden
gerald.evenden at verizon.net
Thu Jan 19 14:24:56 EST 2006
On Thursday 19 January 2006 01:53 pm, Eric Miller wrote:
> I prefer to get rid of all compiler warnings so that real
> errors/warnings are more apparent. For the moment I can live with
> this if there are no alternatives. I _can_ get rid of the warnings
> on Linux by adding an #include at the top of the file, but then the
> program doesn't compile under Cygwin.
My comment is an aside to your main problem.
While the desire to remove all warnings from a compile are admirable it often
ends up being impossible to achieve. Last night and this morning while
cleaning up the next release of libproj4 I turned on the "-Wall -pedantic"
switches and scanned the 'make' output for 'warning's. Out of the 150+ files
constituting the library there are about 3 or 4 files that generate warnings.
With the exception of the strerror_r problem, they are trivial and impractical
As for the strerror_r problem, info/man documents that the function should
return int but the string.h header file prototypes a *char return. This is a
GNU problem that I have not resolved yet. I believe POSIX requires int.
This is a thread compatibility problem.
PS: proj_strerror_r returns int and acts like strerror_r should and should be
Jerry and the low riders:Daisy May and Joshua
"The being cannot be termed rational or virtuous,
who obeys any authority, but that of reason."
---Mary Wollstonecraft 1792
More information about the Proj