Subject: Make vs Other vs vs..
To: None <current-users@netbsd.org>
From: None <netbsd99@sudog.com>
List: current-users
Date: 01/27/2003 10:50:01
I thought I'd enquire as to whether anyone's thought of migrating the NetBSD 
build mechanism to something like Jam or elsewise. What I'm curious about is 
not whether it's just too much work, but technical reasons against shifting 
focus to another build tool or mechanism.

Perhaps there're a few things that something like Jam doesn't do so well?

The reason I ask is that Jam seems to be orders of magnitude quicker when 
building large-ish projects. For example:

Task                                   Sybmake     Jam
Full build from scratch	               116 min     93 min
Full build with nothing to do          30 min      3 min
Full build after a small change        30 min      9 min
Recompile/relink after small change    10 min      2 min

This was in Sybase and were measurements they took during their conversion to 
Jam. Course, the switch itself was extensive, but the gains afterwards made 
up for it.

Finally, Jam may be in the process of being tuned to do header file caching as 
I write this note, which means that the tool may eventually be much quicker 
than it is right now.

Apple appears to like Jam, as they bundle a version of it with Project 
Builder--I think mostly because it is under a very BSD-like license which 
doesn't preclude making binary-only commercial derivative.

Comments? :-)