would it be nice to support =<< ? (not the haskell one)

would it be nice to support =<< ? (not the haskell one)

[khatar]

hello people,

this is a script to render some json elements

	json_elements=()
	read json_elements\[1+$#json_elements] <<.
		{ "user": "$USER", "uid": "$UID" }
	.
	read json_elements\[1+$#json_elements] <<.
		{ "user": "$USER", "uid": "$UID" }
	.
	print "[ ${(j:
	,:)json_elements} ]"

the thing is

	read json_elements\[1+$#json_elements] <<.

is quiet hard to read and i really miss syntaxes
like the perl qq() or the ruby %Q.

would it be possible to get a
would it be smart?

i was thinking of a =<< operator so we could write

	element=<<.
		{ "user": "$USER", "uid": "$UID" }
	.

instead of

	read element <<.
		{ "user": "$USER", "uid": "$UID" }
	.

this would be very convenient to replace this

	read json_elements\[1+$#json_elements] <<.
		{ "user": "$USER", "uid": "$UID" }
	.

by

	json_elements+=<<.
		{ "user": "$USER", "uid": "$UID" }
	.

but is it smart? is it possible?

regards

Re: would it be nice to support =<< ? (not the haskell one)

[Bart Schaefer]

On Wed, Nov 9, 2016 at 10:02 AM,  <khatar@phear.org> wrote:
>
> i was thinking of a =<< operator so we could write
>
>         element=<<.
>                 { "user": "$USER", "uid": "$UID" }
>         .


You can do this:

    element=$(<<.
        { "user": "$USER", "uid": "$UID" }
    .
    )

This is handled efficiently, with a minimum of forking.  Note that
it's necessary to follow the end marker with a newline before adding
the closing paren.

Re: would it be nice to support =<< ? (not the haskell one)

[khatar]

On Wed, Nov 09, 2016 at 11:27:41AM -0800, Bart Schaefer wrote:
>     element=$(<<.
>         { "user": "$USER", "uid": "$UID" }
>     .
>     )
> 
> This is handled efficiently, with a minimum of forking.

i got it in my mind but i have to admit i always think twice before
using $() for a little amount of data (but i never benchmark anything)

in another hand, i realize i assumed that using read with an heredoc
doesn't need a fork. was i wrong ?

anyway: thanks for the idea!

> Note that it's necessary to follow the end marker with a newline before adding
> the closing paren.

nice to point it out! i wasted hours on things like that when i was a
newbie.

regards
marc

Re: would it be nice to support =<< ? (not the haskell one)

[Bart Schaefer]

On Nov 10,  8:15am, khatar@phear.org wrote:
}
} in another hand, i realize i assumed that using read with an heredoc
} doesn't need a fork. was i wrong ?

You were not wrong.  Here-documents use a hidden temporary file, but
do not fork.