[darcs-users] Re: Preference editor

Samuel A. Falvo II sam.falvo at falvotech.com
Thu Jul 1 01:29:51 UTC 2004

On Wednesday 30 June 2004 09:07 am, Björn Lindström wrote:
> I don't find that very obvious. Let's stay with what we're all used to
> from the shell, "set".

This must be a cultural thing then, because here in the U.S., acquiring 
the contents of a variable is "determining the value of" a variable.  
This is a terminology that stems directly from algebra, and pervades 
pretty much all aspects of computer science too.

I find it absolutely retarded that I have to use "set" to list the 
contents of environment variables in the shell.  As far as I'm 
concerned, that is a design bug.  Also note that "set TERM" does 
nothing,  but "echo $TERM" will happily retrieve the current contents of 
a variable.  And this all assumes Bash -- if you are using a csh 
variant, or if you use an Amiga, you'll likely use setenv and getenv.

I claim that what you claim to be "obvious" is actually anything but.  
darcs is not the shell; let's stop pretending that it is.

It's also considered a strict no-no when a single function in an API 
performs multiple different behaviors, depending on its arguments.  One 
need look no further than realloc() to see how much of a nightmare that 
this can cause.  So, if the rule of thumb, "One function, one behavior" 
is so ardently supported at the programming API level, why should this 
rule be broken for the command shell level too?

All I ask for is consistency and logical design.  Whether you choose to 
use "value" is of no material importance to me; but please do not 
overload the meaning of "set."

For that matter, I propose we choose a different word than "set" -- 
"change" is my recommendation.

Samuel A. Falvo II

More information about the darcs-users mailing list