[ExI] The tyranny of context free grammars

Keith Henson hkeithhenson at gmail.com
Tue Dec 7 18:13:03 UTC 2010


On Tue, Dec 7, 2010 at 5:00 AM,  <Ryan Rawson <ryanobjc at gmail.com> wrote:

snip

>  I'd be careful about pining for more languages like
> perl, you have obviously never maintained significant systems written
> in perl.  To be avoided if at all possible.

I am not an expert on Perl . . .

About a dozen years ago I needed to solve a simple problem.  In those
days, a GB was a big disk.  An office network was being served page
images from a raid server backed up (if the image was not found) by a
CDROM jukebox.  It took upwards of 20 seconds to load a CD.  The image
files were all unique names and chances were high that the image you
wanted was on someone's disk on the local network, but where?

It took me several days to figure out how to do this in Perl, which
has got to be the least productive language in terms of lines of code
per day.  I think the server side and client side together were less
than a dozen lines.  It worked just fine, with the server pinging the
local network if it couldn't find the requested image and the
listening client machines telling the server if they had the image.

Done in C it would have taken hundreds of lines and a lot of effort to debug.

I agree with Ryan that Perl (and Forth) are nearly impossible to
maintain.  That's because it's nearly impossible to read code and
figure out what it does.

Keith




More information about the extropy-chat mailing list