Well, I guess that’s a reason, at least. I’m intensely sceptical about its value, though, when the difference is less than a thousand lines: a subfile containin

Well, I guess that’s a reason, at least. I’m intensely sceptical about its value, though, when the difference is less than a thousand lines: a subfile containing the usual ifndef gating and a million `a := b` lines adds less than 0.05s on each subsequent inclusion, on my laptop (the first, about 0.15s.) 10,000× `include gmsl`: around 0.1s. 10,000× on including a single-file modification: around 0.7s. Cull all the blank lines and comments and merge continued lines, 197 lines left (excluding license information, which I will acknowledge has some purpose) and it’s around 0.28s. I haven’t ever used makefiles on large projects, but the performance difference feels like it should be completely negligible.