add for/string and for*/string

[AlexKnauth]

The for/string and for*/string forms are a lot like for/vector and for*/vector, so I use the same common helper function for all four macro definitions.

[rfindler]

Is shrink-string calling make-vector?

…

On Mon, Mar 25, 2019 at 8:33 AM Alex Knauth ***@***.***> wrote: The for/string and for*/string forms are a lot like for/vector and for*/vector, so I use the same common helper function for all four macro definitions. ------------------------------ You can view, comment on, or merge this pull request online at: #2555 Commit Summary - add for/string and for*/string File Changes - *M* racket/collects/racket/private/for.rkt <https://github.com/racket/racket/pull/2555/files#diff-0> (59) Patch Links: - https://github.com/racket/racket/pull/2555.patch - https://github.com/racket/racket/pull/2555.diff — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2555>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAYWsIIHT-gdZC2xVhpD_1GAALSjpw8dks5vaNBCgaJpZM4cG5PW> .

[AlexKnauth]

Thanks, fixed now.

[mflatt]

Would this be better in racket/string? (I think we included too much in racket/base in the first place.)

[rfindler]

Test cases, please.

…

On Mon, Mar 25, 2019 at 8:49 AM Alex Knauth ***@***.***> wrote: Thanks, fixed now. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2555 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAYWsJfjN9XogqZsCR_HQsmkZFv6MSj9ks5vaNPggaJpZM4cG5PW> .

[pmatos]

And docs?

[rfindler]

And don't forget the `@history[]` that I always do! :)

…

On Mon, Mar 25, 2019 at 9:27 AM Paulo Matos ***@***.***> wrote: And docs? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2555 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAYWsFZw3153sQsuFp5dwn1qF2rrFlAkks5vaNzZgaJpZM4cG5PW> .

[gus-massa]

May I ask for for[*]/bytes too? I like to maintain the symmetry of the universe.

I actually needed for/bytes once to implement a no so secure version of crypto-random-bytes:

(define (random-bytes n)
  (list->bytes
   (for/list ([i (in-range n)])
     (random 256))))

[LeifAndersen]

At this point I can't help but wonder if something like define-for-macro is in order...

[AlexKnauth]

for/vector, for/string, for/bytes, for/fxvector, and for/flvector would all be very similar, since vectors, strings, and bytes are all "fixed-length-array" style data structures. All of them could share code, but what works for them wouldn't work as well for other data structures like lists, streams, or hash-tables.

[mfelleisen]

Why don't we have

(for {sum|vector|string|bytes|...} ((i e)) e0 ...)

and even an expression in this position? For efficiency I am sure but when did that hold us back in the past?

[mflatt]

In suggesting racket/string, I really meant to avoid adding code to "for.rkt", and not just avoiding bindings. Sorry for being unclear.

Maybe "racket/private/vector-wraps.rkt" is useful. Its macro would define a string-copy! that isn't needed, though, and maybe other useless things.

[AlexKnauth]

All the logic for for/string and for/bytes is already in the compile-time helper function for_/vector though. It would be a shame not to reuse that code.

(Currently its already duplicated, in both "racket/private/for.rkt" and "racket/private/vector-wraps.rkt")

[soegaard]

I am curious: Is (for/string ...) faster than (vector->string (for/vector ...)) ? /Jens Axel Den man. 25. mar. 2019 kl. 21.04 skrev Alex Knauth <notifications@github.com

…

: All the logic for for/string and for/bytes is already in the compile-time helper function for_/vector though. It would be a shame not to reuse that code. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2555 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAcLxfZJfrjI8DL1ihFeElMyLxuMIiJuks5vaSvPgaJpZM4cG5PW> .

-- -- Jens Axel Søgaard

[jackfirth]

Why don't we have

(for {sum|vector|string|bytes|...} ((i e)) e0 ...)

and even an expression in this position? For efficiency I am sure but when did that hold us back in the past?

@mfelleisen Following up on this for posterity, this concept is now implemented in the reducers library I wrote:

> (for/reducer into-string
               ([ch "hello world"] [_ (in-range 5)])
    (char-upcase ch))
"HELLO"

There's a for*/reducer macro too.