On Thu, Dec 10, 2020 at 1:33 PM Paul Smith <psm...@gnu.org> wrote: > On Thu, 2020-12-10 at 18:19 +0000, DUDZIAK Krzysztof wrote: > > Operators ‘=’ ‘:=’ ‘::=’ seem to have two dimensions these control > > things expansion recursiveness (simple vs. recursive expansion) > > expansion time (immediate vs. deferred) > > No, there is only one dimension really. > > Either the content of the variable is expanded exactly once when it's > defined, or else it's not expanded when it's defined and is expanded > every time the variable is used. > > All other aspects are just side-effects of this single difference. > > I do agree that the terms "recursive variable" and "simple variable" > are misleading and/or not very informative. These terms have been used > in the docs since it was created. >
In the internal docs I've written for my company's build automation (based on GNU make), I explain recursive and simple variables via analogies to C preprocessor macros vs. C variables, which works well for most of those using the automation (which was designed for C and C++ builds). Delayed evaluation vs. immediate evaluation rvalues might work for others. Either way, I agree that the existing terminology is at best unhelpful for developers of other languages, and at worst misleading.
