Re: REQ: Test /etc/rc clean-up

[Sheldon Hearn]

On Fri, 20 Aug 1999 11:59:05 MST, Doug wrote:

 However I'd REALLY like to emphasize again that if we're going to do
 this the proper fix is to use case wherever possible.

 I have offered several times to do the work if it has a chance of
 being committed, that offer is still good.

Hi Doug,

The several times before, did folks come up with objections, or was it
just a case of mass apathy? :-)

I don't want to piss into the wind of wisdom from ages past, but I like
the sound of what you're suggesting. I guess if there _are_ sensible
objections, they'll crop up when you send a diff?

Oh, and thanks for the offer. You no doubt understand that this is
probably not something that'll get pumped straight into CURRENT without
a goodly number of "works for me" messages in prviate mail from folks
using lotsa weird setups. ;-)

Ciao,
Sheldon.


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Re: REQ: Test /etc/rc clean-up

[Warner Losh]

In message [EMAIL PROTECTED] Sheldon Hearn writes:
: -if [ X$start_vinum = XYES ]; then
: +if [ X"${start_vinum}" = X"YES" ]; then

I never understood why you check against X"YES"?  XYES always seemed
much better than X"YES" since the latter is somewhat obscure.  Both
are identical...

Warner


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Re: REQ: Test /etc/rc clean-up

[David O'Brien]

 However I'd REALLY like to emphasize again that if we're going to do
 this the proper fix is to use case wherever possible. There are
 numerous reasons for this, not the least of which are making the
 variable case insensitive (and therefore more user friendly)

I have to really agree with Doug here.  I've seen people use "foo=yes"
(vs. "foo=YES") in their rc.conf.  They got frustrated when things
didn't work.  And in IMHO for a no good reason.

   I have offered several times to do the work if it has a chance of
   being committed, that offer is still good.

What do you think Sheldon?

-- 
-- David([EMAIL PROTECTED]  -or-  [EMAIL PROTECTED])


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Re: REQ: Test /etc/rc clean-up

[Harlan Stenn]

I understand that folks use X$foo becuase if $foo evaluates to -whatever
then there is a *chance* that test will misunderstand.

I gather the reason for using the X trick *and* the quotes is because there
might be some whitespace in there, too.

Given that "case" is a builtin and using "case" instead of "test" is both
faster, easier to read (less clutter), and supports mixed-case easier, I
prefer "case".  But I went down this road before.

H


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Re: REQ: Test /etc/rc clean-up

[Jordan K. Hubbard]

 I gather the reason for using the X trick *and* the quotes is because there
 might be some whitespace in there, too.

Actually, that's mostly just historical legacy.  When the quotes, it's
safe even if the expansion is empty or contains whitespace.  I got
kinda annoyed with this last night and did the following:

  http://www.freebsd.org/~jkh/etc.diffs.fix-it-right

Which I've sent to Sheldon for review but haven't heard anything
back from him yet.

- Jordan


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Re: REQ: Test /etc/rc clean-up

[Doug]

"Jordan K. Hubbard" wrote:
 
  I gather the reason for using the X trick *and* the quotes is because there
  might be some whitespace in there, too.
 
 Actually, that's mostly just historical legacy.  When the quotes, it's
 safe even if the expansion is empty or contains whitespace.

The X also protected test from the case where the expansion included a
string like "-x", although with most modern implementations of test (or
shells with test as a builtin) this is no longer a problem. 

  I got
 kinda annoyed with this last night and did the following:

I agree with some of your changes here, but can you explain your objection
to using case? My argument is that case is a builtin so it makes things
just a little bit cleaner, and more importantly it makes case insensitivity
for the options that much easier to implement which is a huge win in user
friendliness. For example, what happens to if [ "${pccard_ifconfig}" !=
"NO" ] if the user makes the flag "no"? I'd say that the fact that this is
going to go off anyway violates POLA, all "stupid user" arguments aside. 

Doug


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message