Rename IMATH_IMATH_NAMESPACE option to IMATH_NAMESPACE

[cary-ilm]

Also, replace IMATH_INTERNAL_IMATH_NAMESPACE to IMATH_INTERNAL_NAMESPACE.

These were likely originally derived from ILMBASE_IMATH_NAMESPACE, which made sense, but the double IMATH seems confusing and unnecessary.

Note this fixes the CMake options, not the #define symbols, which already have the proper values. This change makes the two consistent.

Signed-off-by: Cary Phillips cary@ilm.com

[lgritz]

I wonder if we should simplify by switching from a 2-deep nested namespace, both of which are customized, to a single (customizable) namespace and an alias.

We would still define

set(IMATH_NAMESPACE "Imath_${IMATH_VERSION_API}" CACHE STRING)

but then the namespace declaration would be

namespace IMATH_NAMESPACE {  };
using namespace Imath = IMATH_NAMESPACE;

and that's it.

This would be literally the only place in the whole codebase where IMATH_NAMESPACE is used, and users would never need to use it. Both all of our internals and all user could would always say Imath::blah, and the resulting symbols would be automatically versioned and optionally changed to anything else the builder customized.

Is there really a reason why it's still useful to implement the versioning and customization as (a) two-level, and (b) having a potentially custom user-referring name as well as a separate custom symbol-mangling name?

[meshula]

I like updating the scheme to use a language feature that didn't exist when the original scheme was invented.

[cary-ilm]

I agree about not seeing the necessity of two levels of macros, although I suspect it's going to be difficult to untangle at this point. My internal build system has a maze of compile-time definitions of IMATH_NAMESPACE, I'd prefer not to disturb its assumptions about the values. A house of cards I'd prefer not to dust right now. This change only affects the names of the CMake options that are new to 3.0, it doesn't disturb actual C++ macros.

…

On Mon, Mar 8, 2021 at 12:51 PM Nick Porcino ***@***.***> wrote: I like updating the scheme to use a language feature that didn't exist when the original scheme was invented. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#111 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFC3DGIXRF5IPAZH2VWLUWLTCU2EJANCNFSM4YZ4LZ6A> .

-- Cary Phillips | R&D Supervisor | ILM | San Francisco

[cary-ilm]

Can someone sign off on this? I believe it's a rather innocuous correction of a CMake configuration setting whose name got mangled in the port from IlmBase.