cvs commit: src/sys/kern init_main.c kern_malloc.c md5c.c subr_autoconf.c subr_mbuf.c subr_prf.c tty_subr.c vfs_cluster.c vfs_subr.c

[Poul-Henning Kamp]

In message <Pine.NEB.3.96L.1030727091934.49952A-100000 at fledge.watson.org>, Robe
rt Watson writes:

>I can't help wondering if the single most positive step in the direction
>of having formalism and analysis tools available in development wouldn't
>be to use a language other than C... :-)

There are certainly areas where C could be improved to fit the task
of modern and portable systems programming better, but all things
considered I only think that would result in rather marginal
prevention of certain "brain-o" style bugs and less rules in style(9)
to enforce.

I am on the other hand pretty sure that the biggest improvement
would not come from changing the language in which we write our
code as much as from programs which read that code, and answer the
questions which it would otherwise be very time consuming to get a
reply to.

For instance, since one can manually establish the answer to the
question "how is this line of code reached", producing more or less
kernel backtraces that lead to the line in questions, I can't see
why we shouldn't be able to automate that task.


-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.