Re: ncdigest V1 #1047

Hi Ed, Mark, et al.,

I would like to chime in regarding the second underscore issue.
My main concern is that the ability to compile-time configure netcdf
with the underscore options is maintained.
Picking a default underscore convention for a compiler/architecture
combination is necessary for the netcdf maintainers.
There will always be those who (with good reason) prefer a different
convention.

My answer to Ed's question of "Why not two underscores?":
I like to build Fortran symbols with single underscores
(and thus have to use the --no-second-underscore option with G95/Linux)
so that the C/Fortran glue code I have works (almost) identically on all platforms (without having to use autoconf).
The G77 convention of double underscores on subroutines (and single
underscores for functions) is bizarre and unique.
It makes mixing GNU-toolchain-compiled Fortran libraries incompatible
with the (single or none) underscore convention of the native compilers on other architectures (Solaris, AIX).

Again, I am advocating that the compile-time switches like no-second-underscore are tested and supported because there are
some good reasons to want it.
Making that particular switch the default setting would be confusing
and is not a good idea.

Thanks,
Charlie
--
Charlie Zender, surname@xxxxxxx, Department of Earth System Science
3228 Croul Hall, UC Irvine, Irvine CA 92697-3100. (949) 824-2987 :)

==============================================================================
To unsubscribe netcdfgroup, visit:
http://www.unidata.ucar.edu/mailing-list-delete-form.html
==============================================================================