improvements to existing site build harness

Hi,

When I created Chronicles, I did so as a branch from the webwml 
repository very deliberately, so that there was a possibility that any 
improvements I made in the build harness could be contributed back.

In my limited time poking at it, I've had some thoughts about how it 
could be made more manageable. However, if the future is for it to be 
binned, it's hard to gauge whether this would be wasted effort.

In particular, the current build harness makes use of recursive 
makefiles. I generally agree with the arguments made in "recursive make 
considered harmful"[1], in particular, I've found it would be very 
useful if the top-level Makefile had object-visibility of the final 
targets several layers deeper. 

However moving away from recursive-make would be a pretty major 
undertaking, which would be of limited utility depending on whether it 
would be appreciated upstream in the medium-term, prior to moving to 
Hugo, and how long that medium term might be. What do you think?

Are there any other pain points/nits with the current harness that fixes 
for which would be welcome and worthwhile?


Thanks!

[1] https://aegis.sourceforge.net/auug97.pdf

-- 
⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Jonathan Dowland
⢿⡄⠘⠷⠚⠋⠀ https://jmtd.net
⠈⠳⣄⠀⠀⠀⠀