Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 4 Apr 2013 15:04:41 +0000
From:      "Teske, Devin" <Devin.Teske@fisglobal.com>
To:        Mark Felder <feld@feld.me>
Cc:        Devin Teske <dteske@freebsd.org>, "<freebsd-questions@freebsd.org>" <freebsd-questions@freebsd.org>
Subject:   Re: OT: posix sh problem
Message-ID:  <13CA24D6AB415D428143D44749F57D7201EF2627@ltcfiswmsgmb21>
In-Reply-To: <op.wu0tttba34t2sn@tech304.office.supranet.net>
References:  <op.wu0nsgsf34t2sn@tech304.office.supranet.net> <13CA24D6AB415D428143D44749F57D7201EF2411@ltcfiswmsgmb21> <op.wu0tttba34t2sn@tech304.office.supranet.net>

next in thread | previous in thread | raw e-mail | index | archive | help

On Apr 4, 2013, at 7:42 AM, Mark Felder wrote:

> On Thu, 04 Apr 2013 08:54:30 -0500, Teske, Devin <Devin.Teske@fisglobal.c=
om> wrote:
>=20
>> Wait, you can't? Then I've been doing something wrong all these years=85
>> #!/bin/sh
>> printf "line1\nline2\n" | while read line
>> do
>> 	echo "line=3D[$line]"
>> done
>=20
> You sort-of can, but it's not portable at all. As detailed here: http://w=
ww.etalabs.net/sh_tricks.html
>=20
>>> One common pitfall is trying to read output piped from commands, such a=
s:
>>> foo | IFS=3D read var
>>> POSIX allows any or all commands in a pipeline to be run in subshells,

And for most purposes that's fine. Read-on=85


>>> and which command (if any) runs in the main shell varies greatly betwee=
nimplementations

=85 that is only if you truly need the variables to be read into the main s=
hell. This is most always not what you want.

The page you linked about doesn't talk about the special case of "while", i=
n example:

	foo | IFS=3D while read var

On the back-end nothing changes (the same caveat applies -- variables set o=
n the right side of the pipe may not be available to the main shell; as-per=
 the quoted text). However, the high-level task changes from:

	I want to read some text from a pipe into some variables

to instead:

	I want to read some text from a pipe and process it word-by-word (in your =
case) and act on the words in a loop

So in other words=85 the only reason for wanting the variables in the main =
shell is if you want to act on the last set of variables for the last line =
after the loop has run (and presumably already processed the last line). Th=
is is what I am saying anyone will rarely ever want. In other words, once t=
he loop (potentially running in a sub-shell) has completed, you likely don'=
t care about the variable contents and are willing to throw them away anyho=
w.



>>> =97 in particular Bash and ksh differ here. The standardidiom for overc=
oming this problem is to use a here document:
>>>=20
>>> IFS=3D read var << EOF
>>> $(foo)
>>> EOF
>=20

But you're not processing a single line; you're processing the entire input=
 at-once and performing an action (writing to the screen) that also doesn't=
 care whether it's in a sub-shell or not.

SO=85

I say rock-on with the original syntax. It's portable. You don't need those=
 vars when the loop ends.

>=20
> I was having problems with the variables magically becoming empty, rememb=
ered I had Rich's site bookmarked, checked to see if it mentioned and it wa=
s. I'll admit there's a high chance that due to lack of sleep user error wa=
s the culprit.

I'm interested in why you need the variables after the loop has completed. =
Put your code in the loop where the variables are defined and have values.
--=20
Devin

_____________
The information contained in this message is proprietary and/or confidentia=
l. If you are not the intended recipient, please: (i) delete the message an=
d all copies; (ii) do not disclose, distribute or use the message in any ma=
nner; and (iii) notify the sender immediately. In addition, please be aware=
 that any message addressed to our domain is subject to archiving and revie=
w by persons other than the intended recipient. Thank you.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?13CA24D6AB415D428143D44749F57D7201EF2627>