At 09:46 AM 5/18/2004, Paul D. Smith wrote:
%% David Boyce <[EMAIL PROTECTED]> writes:
  db> Which raises the question: why wasn't the fast path in use in the
  db> truss output I showed? The answer is apparently that the generated
  db> Perl makefiles have an explicit SHELL=/bin/sh which disables the
  db> fast path.

Are you sure?  I'm surprised because there's some code in job.c which
checks the value of the SHELL variable to see whether it looks like a
valid shell; it basically does a strcmp() with "/bin/sh" and if it's
different it uses the slow path.

No, I'm not at all sure. I merely noticed that the truss output appeared to show an unneeded shell invocation and when I noticed an assignment to SHELL in the Makefile(s) I assumed it was the culprit. I'll take a closer look and see if I can figure out what's going on without internalizing the entire Perl build model (which I could do without).


Argh. I just looked more closely into some of the ancillary Perl makefiles and found this:

NOOP = $(SHELL) -c true

followed by all kinds of underbrush like

all :: pure_all manifypods
        $(NOECHO) $(NOOP)

No accounting for taste, I guess.

-David Boyce



_______________________________________________
Help-make mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/help-make

Reply via email to