Hi,
Am Mo., 17. Sep. 2018 um 10:18 Uhr schrieb Tom M. <
address@hidden>:
> I don't want to compel people to require yet another build dependency. While we have very good reasons to make glib a dependency, making e.g. python a dependency seems questionable as the current runtime solution works flawlessly for 99% of all users.
> [..]
> Personally I prefer the current C90 compliant solution that relies on the preprocessor (#423).
As you say: the current code works for 99% of all users and the need to save a few KB of RAM is very rare indeed. Adding large chunks of fairly complicated macros to the codebase just for that very rare use-case is not really worth it, I think.
The approach in #423 already includes a compile-time switch for the constant tables. If we simply replace the macro approach with a much simpler and more readable Python script to generate the code, keep the compile-time switch and have it disabled by default, then Python is only a build dependency for the 1% of users (probably even less than 1%...) that need constant tables. And my guess is that people with such special requirements have no problem installing a Python interpreter. :-)
Cheers,
Marcus