This is the mail archive of the cygwin@sources.redhat.com mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

RE: bash: setenv: command not found


Today, Chet Ramey wrote:
> > call me crazy but I like orthogonality, and logically named
> > commands.
> 
> You like two completely different ways of doing assignments?

um, yeah... if they're different concepts. I don't exactly think
local shell variables and the global process environment should be
manipulated with the same syntax.

> First, export foo=bar is just convenient shorthand for
> 
> foo=bar
> export foo

yep, I'm familiar with it:

foo=bar sets a local shell variable
foo=  unsets a local shell variable
export foo promotes a local shell variable into the process environ
unset foo  demotes a process environ *AND* unsets the same shell variable

> There is only one way to do assignment statements in sh: var=value.
> All variable assignments are local, with the single exception of
> variable assignments preceding a command.  `export' is the only way
> to make a variable part of the environment.

but where is "unexport"? how do I demote a process environment variable
without removing it from the local shell?

It doesn't matter. Fred had the right point: it's POSIX. You're just
doing what the spec says. I just don't much care for this aspect of
the spec, like I said, call me crazy. ;) I'll just file it in my "get
over it" file and move on with life... it's not like I'm not *used*
to this, I just don't *like* it. Besides it's such a minor item in
comparison to all the other features in bash.

> sh-style shells are much more consistent.

Agreed, they are consient, I just don't think they're orthogonal. I
tend to really like orthagonal design; I learned assembly programming
on a PDP-11 (well, ok a simulated one, 'cuz the uni's original died
a bit before I took the class) then I went out and studied x86
assembly....

p.s. for those that replied to my last... Chet's original note
finally showed up, and I figured out the problem... lousy ISP's
mail servers are taking 5-10 MINUTES to SYN_ACK... lots of stuff
is going into retry queues all over the net, no wonder my lists
are such a mess. :(

-- 
now the forces of openness have a powerful and
  unexpected new ally - http://ibm.com/linux


--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]