Subject: build.sh efficiency
To: NetBSD current list <email@example.com>
From: William Allen Simpson <firstname.lastname@example.org>
Date: 11/05/2003 11:19:36
When Metzger was looking into the openssl performance issue yesterday,
he was walking me through my saved build.sh outputs. Normally, I just
look at the tail....
If I am reading it correctly, it appears to run through the whole tree
deleting objects, then again making objdirs, and again depending, and
again compiling, etc.
It seems to me that this is highly inefficient disk activity. The
memory cache is going to lose locality, with everything read in multiple
times (unless your memory is as big as the src+obj).
Would it be possible to run through the tree once, doing multiple tasks?
Or maybe only twice?
After all, unlike the old `make build` process, build.sh knows at the
outset the tasks that need to be completed, based on the command, flags,
Maybe with a faster build process, folks would actually use it to test
before checkin. (Just a hope.)
William Allen Simpson
Key fingerprint = 17 40 5E 67 15 6F 31 26 DD 0D B9 9B 6A 15 2C 32