Re: Strange behaviour of read builtin

[Eli Schwartz]

On 7/1/20 7:25 AM, Greg Wooledge wrote:
> On Wed, Jul 01, 2020 at 07:51:00AM +0300, M. Nejat AYDIN wrote:
>> Is this intended behaviour of read builtin? If so where is it
>> documented? The Bash Reference Manual says only "If no names
>> are supplied, the line read is assigned to the variable REPLY."
> 
> There you go.  That is where it's documented.  (That sentence also
> appears in the man page, at the end of the description of the read
> builtin.)
> 
>> There is no mention of preserving leading and trailing white
>> spaces.
> 
> How else would you interpret "the line read is assigned to the variable
> REPLY"?  Read the line, assign it to the variable.  Do not pass Go,
> do not perform IFS whitespace trimming.

I'd probably gloss over that interpretation and assume its behavior is
equivalent to specifying a single NAME. Then I'd run

`IFS= read << " x "`

to guarantee the behavior I wanted, even though it turned out to be
unnecessary.

So I learned something today. I guess it couldn't hurt to extend that
sentence with "without processing".

-- 
Eli Schwartz
Arch Linux Bug Wrangler and Trusted User

signature.asc
Description: OpenPGP digital signature