Re: portability as a goal for debian?
On 03/08/2001 12:15:58 PM "John H. Robinson, IV" wrote:
>> On Thu, Mar 08, 2001 at 12:04:21PM -0600, David Starner wrote:
>> > Not so; awk alone can solve all Church-computable problems, which is
>> > same set of problems perl can solve.
>> yes, i know it is possible to do higher math in sed (i have a sed script
>> to compute primes. however it is very slow. but it /can/ be done), but
>> somehow i think that opening tcp sockets is a bit beyond sed/awk.
It's the difference between computability and interfaces.
I can obviously compute whatever a PDP-8 does using my PC, by running the
"simh" simulator on my PC. But just because my PC can "compute" whatever a
PDP-8 can compute, that doesn't mean my PC can read a PDP-8's magnetic
tapes or punch cards.
The only vaguely on topic relationship to the discussion is that just
because I can theoretically rewrite emacs using ed macros or rewrite the
linux kernel in /bin/ash, making those items more "portable" does not
necessarily make them easier to debug, faster, or even easier to port. In
fact, "dumbing down" a script to make it more portable on system X might
make it even harder to port to system Y, etc. You can enter a tailspin
where the changes involved in saving 1 person hour for a specific port
results in 2 extra person hours for each of the other ports, thus small
changes on various ports can slow overall development to a crawl.
Bash and perl and ash and awk and all that might be "computationally
equivalent", but the interfaces are the important part. I pick my lanugage
based on the interfaces I need for a particular job.
I've seen terrifying and undebugable, yet supposedly portable, thousand
line ksh scripts from commercial products. I hope to never work on one of