Elisp threading macros

Elisp threading macros

[Garklein]

Hello,

Would it be possible to have the dash.el macros `->', `->>', `-->',
and `-some->' in subr-x.el? I find them to be really useful, and would
love to see them in the base language.

I know that `thread-first' and `thread-last' exist, but I find that my
eyes can recognize `->' and `->>' much easier. `-->' is really nice
for more complex threading.

`-some->' is useful in cases like this:

(let (foo (function-which-may-return-nil))
  (when foo
    (bar foo)))

which can be rewritten, much cleaner, as

(-some-> foo function-which-may-return-nil bar)

Thank you,
Garklein

Re: Elisp threading macros

[Robert Pluim]

>>>>> On Wed, 28 Aug 2024 10:08:05 -0400, Garklein <garklein97@gmail.com> said:

    Garklein> Hello,
    Garklein> Would it be possible to have the dash.el macros `->', `->>', `-->',
    Garklein> and `-some->' in subr-x.el? I find them to be really useful, and would
    Garklein> love to see them in the base language.

    Garklein> I know that `thread-first' and `thread-last' exist, but I find that my
    Garklein> eyes can recognize `->' and `->>' much easier. `-->' is really nice
    Garklein> for more complex threading.

    Garklein> `-some->' is useful in cases like this:

    Garklein> (let (foo (function-which-may-return-nil))
    Garklein>   (when foo
    Garklein>     (bar foo)))

    Garklein> which can be rewritten, much cleaner, as

    Garklein> (-some-> foo function-which-may-return-nil bar)

Beauty is in the eye of the beholder:

(when-let ((foo (function-which-may-return-nil)))
  (bar foo))

Robert
-- 

Re: Elisp threading macros

[Eli Zaretskii]

> From: Garklein <garklein97@gmail.com>
> Date: Wed, 28 Aug 2024 10:08:05 -0400
> 
> Would it be possible to have the dash.el macros `->', `->>', `-->',
> and `-some->' in subr-x.el? I find them to be really useful, and would
> love to see them in the base language.

We might, but probably not using the names dash.el uses.

Re: Elisp threading macros

[Visuwesh]

[புதன் ஆகஸ்ட் 28, 2024] Robert Pluim wrote:

> Beauty is in the eye of the beholder:
>
> (when-let ((foo (function-which-may-return-nil)))
>   (bar foo))
>

Not to mention, eldoc does not really work for these threading macros
which is a real annoyance when trying to read code that heavily uses
such macros.

Re: Elisp threading macros

[Henrik Kjerringvåg]

When-let is just functionally the same in the example provided. I think the point was actually about threading. I’m with Garklein on this subject. In my day job I program in Clojure and can attest to these threading macros being very valuable. 
If Eldoc doesn’t work well with these macros, it sounds like some improvements can be made to Eldoc as well.

Sendt fra min iPhone

> 28. aug. 2024 kl. 18:43 skrev Visuwesh <visuweshm@gmail.com>:
> [புதன் ஆகஸ்ட் 28, 2024] Robert Pluim wrote:
> 
>> Beauty is in the eye of the beholder:
>> 
>> (when-let ((foo (function-which-may-return-nil)))
>>  (bar foo))
> 
> Not to mention, eldoc does not really work for these threading macros
> which is a real annoyance when trying to read code that heavily uses
> such macros.

Re: Elisp threading macros

[Philip Kaludercic]

Garklein <garklein97@gmail.com> writes:

> Hello,
>
> Would it be possible to have the dash.el macros `->', `->>', `-->',
> and `-some->' in subr-x.el? I find them to be really useful, and would
> love to see them in the base language.
>
> I know that `thread-first' and `thread-last' exist, but I find that my
> eyes can recognize `->' and `->>' much easier. `-->' is really nice
> for more complex threading.

This sounds more like a matter of habit.  FWIW I find a symbol like ->>
or --> very unusual for Elisp.  All in all, there are not that many if I
run this snippet

(mapatoms
 (lambda (sym)
   (when (string-match-p (rx bos (+ (not alnum)) eos)
			 (symbol-name sym))
     (message "Note %S" sym))))

and most of the appear to be part of some DSL like rx, and not Lisp
functions (in my case only 18 out of 57).

> `-some->' is useful in cases like this:

Can you explain what, ->>, --> and -some-> do generally?
>
> (let (foo (function-which-may-return-nil))
>   (when foo
>     (bar foo)))
>
> which can be rewritten, much cleaner, as
>
> (-some-> foo function-which-may-return-nil bar)

[citation needed]

> Thank you,
> Garklein
>
>

-- 
	Philip Kaludercic on peregrine

Re: Elisp threading macros

[Basil L. Contovounesios]

Eli Zaretskii [2024-08-28 18:55 +0300] wrote:

>> From: Garklein <garklein97@gmail.com>
>> Date: Wed, 28 Aug 2024 10:08:05 -0400
>> 
>> Would it be possible to have the dash.el macros `->', `->>', `-->',
>> and `-some->' in subr-x.el? I find them to be really useful, and would
>> love to see them in the base language.
>
> We might, but probably not using the names dash.el uses.

Just for context, dash.el inherited these names from the Clojure
programming language.

-- 
Basil

Re: Elisp threading macros

[Thierry Volpiatto]

Garklein <garklein97@gmail.com> writes:

> Hello,
>
> Would it be possible to have the dash.el macros `->', `->>', `-->',
> and `-some->' in subr-x.el? I find them to be really useful, and would
> love to see them in the base language.

Personally, I have stopped using packages wrote with dash.el, the code is
unreadable, at least for me.
I hope we will not have to suffer of this in Emacs.

> I know that `thread-first' and `thread-last' exist, but I find that my
> eyes can recognize `->' and `->>' much easier. `-->' is really nice
> for more complex threading.
>
> `-some->' is useful in cases like this:
>
> (let (foo (function-which-may-return-nil))
>   (when foo
>     (bar foo)))
>
> which can be rewritten, much cleaner, as
>
> (-some-> foo function-which-may-return-nil bar)

And now we have no idea of what this code does, not speaking of the
diverse ->, --> etc... that are a mystery.

-- 
Thierry

Re: Elisp threading macros

[Garklein]

> And now we have no idea of what this code does, not speaking of the
> diverse ->, --> etc... that are a mystery.

To me, the `-some->' form is clearer than the `let'. (I am very happy
to learn about `when-let' though, useful!) I can easily read it left
to right, as if it were English and not code.

I don't see what's so bad about the names `->', `-->' etc. After all,
we commonly use functions with such cryptic names as `car', `cdr', and
`cons'  :)

On Wed, Aug 28, 2024 at 2:25 PM Thierry Volpiatto <thievol@posteo.net> wrote:
>
> Garklein <garklein97@gmail.com> writes:
>
> > Hello,
> >
> > Would it be possible to have the dash.el macros `->', `->>', `-->',
> > and `-some->' in subr-x.el? I find them to be really useful, and would
> > love to see them in the base language.
>
> Personally, I have stopped using packages wrote with dash.el, the code is
> unreadable, at least for me.
> I hope we will not have to suffer of this in Emacs.
>
> > I know that `thread-first' and `thread-last' exist, but I find that my
> > eyes can recognize `->' and `->>' much easier. `-->' is really nice
> > for more complex threading.
> >
> > `-some->' is useful in cases like this:
> >
> > (let (foo (function-which-may-return-nil))
> >   (when foo
> >     (bar foo)))
> >
> > which can be rewritten, much cleaner, as
> >
> > (-some-> foo function-which-may-return-nil bar)
>
> And now we have no idea of what this code does, not speaking of the
> diverse ->, --> etc... that are a mystery.
>
> --
> Thierry

Re: Elisp threading macros

[Garklein]

> Can you explain what, ->>, --> and -some-> do generally?

Of course!

`->' takes in a value and a variable amount of forms, "threading" the
value through them by inserting it as the first argument in the first
form, inserting the result of that as the first argument in the next
form, etc.

For example, `(-> foo x (y 3) z)' is equivalent to `(z (y (x foo) 3))'.

`->>' does the same thing, but inserts each value as the last argument
of each form, instead of the first.

`-->' substitutes the symbol `it' in each form as the previous value.
`(-> x (* it it) (+ it 9) (sqrt it))' is equivalent to `(sqrt (+ (* x
x) 9))', and can be read as "take `x', multiply it by itself, add nine
to it, and take its square root".

`-some->' is like `->', except that it will short-circuit the whole
`-some->' expression with `nil' whenever any form returns `nil'.

On Wed, Aug 28, 2024 at 2:05 PM Philip Kaludercic <philipk@posteo.net> wrote:
>
> Garklein <garklein97@gmail.com> writes:
>
> > Hello,
> >
> > Would it be possible to have the dash.el macros `->', `->>', `-->',
> > and `-some->' in subr-x.el? I find them to be really useful, and would
> > love to see them in the base language.
> >
> > I know that `thread-first' and `thread-last' exist, but I find that my
> > eyes can recognize `->' and `->>' much easier. `-->' is really nice
> > for more complex threading.
>
> This sounds more like a matter of habit.  FWIW I find a symbol like ->>
> or --> very unusual for Elisp.  All in all, there are not that many if I
> run this snippet
>
> (mapatoms
>  (lambda (sym)
>    (when (string-match-p (rx bos (+ (not alnum)) eos)
>                          (symbol-name sym))
>      (message "Note %S" sym))))
>
> and most of the appear to be part of some DSL like rx, and not Lisp
> functions (in my case only 18 out of 57).
>
> > `-some->' is useful in cases like this:
>
> Can you explain what, ->>, --> and -some-> do generally?
> >
> > (let (foo (function-which-may-return-nil))
> >   (when foo
> >     (bar foo)))
> >
> > which can be rewritten, much cleaner, as
> >
> > (-some-> foo function-which-may-return-nil bar)
>
> [citation needed]
>
> > Thank you,
> > Garklein
> >
> >
>
> --
>         Philip Kaludercic on peregrine

Re: Elisp threading macros

[Philip Kaludercic]

Garklein <garklein97@gmail.com> writes:

>> Can you explain what, ->>, --> and -some-> do generally?
>
> Of course!
>
> `->' takes in a value and a variable amount of forms, "threading" the
> value through them by inserting it as the first argument in the first
> form, inserting the result of that as the first argument in the next
> form, etc.
>
> For example, `(-> foo x (y 3) z)' is equivalent to `(z (y (x foo) 3))'.

Right, this is what we already have with thread-first, right?

> `->>' does the same thing, but inserts each value as the last argument
> of each form, instead of the first.

Oh, I wouldn't have guessed that from the symbol name.

> `-->' substitutes the symbol `it' in each form as the previous value.
> `(-> x (* it it) (+ it 9) (sqrt it))' is equivalent to `(sqrt (+ (* x
> x) 9))', and can be read as "take `x', multiply it by itself, add nine
> to it, and take its square root".

Hmm, anaphoric macros are somewhat controversial they have an even
greater cognitive overhead than every day macros, let alone threading
macros.

> `-some->' is like `->', except that it will short-circuit the whole
> `-some->' expression with `nil' whenever any form returns `nil'.

Here I have to agree with others in the thread that when-let* seems more
maintainable, as it forces the programmer to introduce names that
hopefully indicate what is being done.

---

Generally this seems like a cultural clash between traditional Lisps and
newer languages like Clojure that derive some inspired from Lisps, but
have intentionally broken with the existing "Lisp tradition".
Backporting concepts like these is certainly technically feasible, but
it feels forced and foreign for people like me who don't work with
Closure or related languages.

> On Wed, Aug 28, 2024 at 2:05 PM Philip Kaludercic <philipk@posteo.net> wrote:
>>
>> Garklein <garklein97@gmail.com> writes:
>>
>> > Hello,
>> >
>> > Would it be possible to have the dash.el macros `->', `->>', `-->',
>> > and `-some->' in subr-x.el? I find them to be really useful, and would
>> > love to see them in the base language.
>> >
>> > I know that `thread-first' and `thread-last' exist, but I find that my
>> > eyes can recognize `->' and `->>' much easier. `-->' is really nice
>> > for more complex threading.
>>
>> This sounds more like a matter of habit.  FWIW I find a symbol like ->>
>> or --> very unusual for Elisp.  All in all, there are not that many if I
>> run this snippet
>>
>> (mapatoms
>>  (lambda (sym)
>>    (when (string-match-p (rx bos (+ (not alnum)) eos)
>>                          (symbol-name sym))
>>      (message "Note %S" sym))))
>>
>> and most of the appear to be part of some DSL like rx, and not Lisp
>> functions (in my case only 18 out of 57).
>>
>> > `-some->' is useful in cases like this:
>>
>> Can you explain what, ->>, --> and -some-> do generally?
>> >
>> > (let (foo (function-which-may-return-nil))
>> >   (when foo
>> >     (bar foo)))
>> >
>> > which can be rewritten, much cleaner, as
>> >
>> > (-some-> foo function-which-may-return-nil bar)
>>
>> [citation needed]
>>
>> > Thank you,
>> > Garklein
>> >
>> >
>>
>> --
>>         Philip Kaludercic on peregrine

-- 
	Philip Kaludercic on peregrine

Re: Elisp threading macros

[Po Lu]

Garklein <garklein97@gmail.com> writes:

> Hello,
> `-some->' is useful in cases like this:
>
> (let (foo (function-which-may-return-nil))
>   (when foo
>     (bar foo)))
>
> which can be rewritten, much cleaner, as
>
> (-some-> foo function-which-may-return-nil bar)

I think very few people will agree, but "-some->" is not an acceptable
name for a preloaded function.  Or any function, come to that.

Re: Elisp threading macros

[tomas]

[-- Attachment #1: Type: text/plain, Size: 1572 bytes --]

On Wed, Aug 28, 2024 at 07:54:04PM +0000, Philip Kaludercic wrote:

[...]

> Generally this seems like a cultural clash between traditional Lisps and
> newer languages like Clojure that derive some inspired from Lisps, but
> have intentionally broken with the existing "Lisp tradition".
> Backporting concepts like these is certainly technically feasible, but
> it feels forced and foreign for people like me who don't work with
> Closure or related languages.

I think this is a very important point. Emacs Lisp (any language) isn't
"just" a set of rules, but a whole set of applications and a community
(or perhaps a set of interconnected communities) [0]. Each time someone
with big enthusiasm comes along and "hey, let's change everything, this
looks so much better", things tend to degrade into strange identity wars.

On the one hand, you can't just change the whole community "just like that":
people gravitate towards languages for reasons (I must admit I cringe myself
when I see those threaded macros, but then, my excursions into those newer
functional languages have never really worked). Most people are here because
they like things as they are.

On the other hand, for my taste at least, the reactions (my [1] reaction,
at that!) to those ideas feels sometimes more hostile than necessary.

I'd like (for myself!) to have better answers to that.

Cheers

[0] And a means of communication. And of identification. And...
[1] That's why I often shut up in those cases. But hostile silence
   is hostile, too. Oh, well.
-- 
t

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

Re: Elisp threading macros

[Alfred M. Szmidt]

   `-some->' is useful in cases like this:

   (let (foo (function-which-may-return-nil))
     (when foo
       (bar foo)))

   which can be rewritten, much cleaner, as

   (-some-> foo function-which-may-return-nil bar)

Those two forms aren't the same.  This is the macroexpand-all version:

(let ((result (let ((result foo))
		(if result
		    (progn (function-which-may-return-nil result))))))
  (if result
      (progn (bar result))))

Notice how e.g., function-which-may-return-nil actually gets the foo
argument.  And the semantics are also different.

While the above might seam cleaner, since it is shorter, it is much
harder to modify, maintain and reason about.

Re: Elisp threading macros

[Sean Whitton]

Hello,

On Wed 28 Aug 2024 at 06:05pm GMT, Philip Kaludercic wrote:

> Garklein <garklein97@gmail.com> writes:
>
>> Hello,
>>
>> Would it be possible to have the dash.el macros `->', `->>', `-->',
>> and `-some->' in subr-x.el? I find them to be really useful, and would
>> love to see them in the base language.
>>
>> I know that `thread-first' and `thread-last' exist, but I find that my
>> eyes can recognize `->' and `->>' much easier. `-->' is really nice
>> for more complex threading.
>
> This sounds more like a matter of habit.  FWIW I find a symbol like ->>
> or --> very unusual for Elisp.  All in all, there are not that many if I
> run this snippet

Right, it's not Elisp/MacLisp/Common Lisp style to use names like that.
Similar to how we use '-q' for predicates, and not '?', like Scheme does.

Elisp is a classic Lisp, which gives it a coherent aesthetic, which
helps with readability, I think.

-- 
Sean Whitton

Re: Elisp threading macros

[Philip Kaludercic]

Sean Whitton <spwhitton@spwhitton.name> writes:

> Hello,
>
> On Wed 28 Aug 2024 at 06:05pm GMT, Philip Kaludercic wrote:
>
>> Garklein <garklein97@gmail.com> writes:
>>
>>> Hello,
>>>
>>> Would it be possible to have the dash.el macros `->', `->>', `-->',
>>> and `-some->' in subr-x.el? I find them to be really useful, and would
>>> love to see them in the base language.
>>>
>>> I know that `thread-first' and `thread-last' exist, but I find that my
>>> eyes can recognize `->' and `->>' much easier. `-->' is really nice
>>> for more complex threading.
>>
>> This sounds more like a matter of habit.  FWIW I find a symbol like ->>
>> or --> very unusual for Elisp.  All in all, there are not that many if I
>> run this snippet
>
> Right, it's not Elisp/MacLisp/Common Lisp style to use names like that.
> Similar to how we use '-q' for predicates, and not '?', like Scheme does.
                         ^
                nitpick: -p

> Elisp is a classic Lisp, which gives it a coherent aesthetic, which
> helps with readability, I think.

I agree!  Elisp might have inconsistencies, but we can try to avoid
adding even more.

-- 
	Philip Kaludercic on peregrine

Re: Elisp threading macros

[Garklein]

Sorry, you're absolutely right! I had a massive brainfart.

What I meant to write was `(-some-> (function-which-may-return-nil) bar)'

On Thu, Aug 29, 2024 at 1:19 AM Alfred M. Szmidt <ams@gnu.org> wrote:
>
>    `-some->' is useful in cases like this:
>
>    (let (foo (function-which-may-return-nil))
>      (when foo
>        (bar foo)))
>
>    which can be rewritten, much cleaner, as
>
>    (-some-> foo function-which-may-return-nil bar)
>
> Those two forms aren't the same.  This is the macroexpand-all version:
>
> (let ((result (let ((result foo))
>                 (if result
>                     (progn (function-which-may-return-nil result))))))
>   (if result
>       (progn (bar result))))
>
> Notice how e.g., function-which-may-return-nil actually gets the foo
> argument.  And the semantics are also different.
>
> While the above might seam cleaner, since it is shorter, it is much
> harder to modify, maintain and reason about.

Re: Elisp threading macros

[Joost Kremers]

On Thu, Aug 29 2024, tomas@tuxteam.de wrote:
> On the one hand, you can't just change the whole community "just like that":
> people gravitate towards languages for reasons (I must admit I cringe myself
> when I see those threaded macros, but then, my excursions into those newer
> functional languages have never really worked). Most people are here because
> they like things as they are.

Yes, I think that's very much the point. It's more a matter of taste than
anything else. And taste for a large part depends on what one is used to.
When I first encountered Clojure, I quite disliked it. And threading
macros? You'd have to be out of your mind to write code like that...

But at some point, I decided to learn Clojure and I must admit, I really
like the language. It's different than Elisp for sure, but things that
irked me initially, I came to like eventually, because I started to see why
they were designed the way they are. (I'd actually love to do more with
Clojure, TBH...)

Still, that doesn't mean I think Elisp should become more like Clojure, of
course. (I'm actually neutral on that point.) I think the current
situation, with the seq library and thread-{first|last} available in core,
and dash.el readily available from GNU ELPA, is a good compromise. (And, to
the OP: from your perspective, probably the "best" you're gonna get. 😉)



-- 
Joost Kremers
Life has its moments

Re: Elisp threading macros

[Rudolf Adamkovič]

Garklein <garklein97@gmail.com> writes:

> I don't see what's so bad about the names `->', `-->' etc.

I find them cryptic as well (and I say that as a professional Fennel
programmer, using Clojure syntax all day long).

> After all, we commonly use functions with such cryptic names as `car',
> `cdr', and `cons' :)

This is where Emacs Lisp could learn from Clojure and introduce `first',
`next', and `pair' aliases.  Document the new names as preferred and
keep the old "historical curiosities" functional forever.  Of course,
`mapcar' and the like would have to be renamed as well, but all that is
doable if there is a will to move the readability of the "very core" of
Emacs Lisp forward, from 1950s into the 21st century.

Rudy
-- 
"Mathematics takes us still further from what is human into the region
of absolute necessity, to which not only the actual world, but every
possible world, must conform."  --- Bertrand Russell, 1902

Rudolf Adamkovič <rudolf@adamkovic.org> [he/him]
http://adamkovic.org

Re: Elisp threading macros

[Robert Pluim]

>>>>> On Fri, 30 Aug 2024 14:26:30 +0200, Rudolf Adamkovič <rudolf@adamkovic.org> said:
    Rudolf> This is where Emacs Lisp could learn from Clojure and introduce `first',
    Rudolf> `next', and `pair' aliases.  Document the new names as preferred and
    Rudolf> keep the old "historical curiosities" functional forever.  Of course,
    Rudolf> `mapcar' and the like would have to be renamed as well, but all that is
    Rudolf> doable if there is a will to move the readability of the "very core" of
    Rudolf> Emacs Lisp forward, from 1950s into the 21st century.

emacs lisp already has `first', `rest' and `seq-do' or `dolist'. And
`nth' to replace `caddr' etc.

Robert
-- 

RE: [External] : Re: Elisp threading macros

[Drew Adams]

> emacs lisp already has `first', `rest' and
> `seq-do' or `dolist'. And `nth' to replace
> `caddr' etc.

Neither nth nor nthcdr replaces caddr, caar
cdadar, cdddar, etc.  Each of the c[..]r
functions targets a particular part of a
list (tree) in one fell swoop.

And its name is a _picture_ of where that
part is located in the tree.  Direct access,
including mentally/visually.

(Yes, it takes (a little) getting used to.)

That alone is a good reason for not only
keeping car and cdr but having them at the
back of your mind in combination, instead
of just first/rest or head/tails.

Could different chars have been used than
c, a, d, and r?  Of course.  The c and r
aren't needed, and we could have used f for
first and r for rest, or h for head and
t for tails, or | for car and . for cdr, or
- for car and @ for cdr:

 cadaar  =  frff  =  hthh  =  |.||  =  -@--
 caar    =  ff    =  hh    =  ||    =  --
 caaar   =  fff   =  hhh   =  |||   =  ---
 caddar  =  frrf  =  htth  =  |..|  =  -@@-

 cdadar  =  rfrf  =  thth  =  .|.|  =  @-@-
 cdddar  =  rrrf  =  ttth  =  ...|  =  @@@-
 caddr   =  frr   =  htt   =  |..   =  -@@

Or Morse code: .-..  ..  ...  .--.
                L    I    S    P

The point is that there are 2 basic access
functions for the 2 components of a cons,
and accessing their combinations can be
abbreviated in a pictorial way with 2 chars.
___

Of course, no one would suggest that people
should _mostly_ use the more complex c[..]r
abbreviations instead of binding intermediate
vars with meaningful names to uses of smaller
access c[..]r variations (including car/cdr).

Still, when that's not so important, if a
reader is _used_ to the c[..]r pictures, an
occasional (caddar foo) can be handier than
(first (rest (rest (first foo)))) or
(first (nthcdr 1 (first foo))).

IOW, if you're going to use such first/rest
combinations anyway (instead of intermediate
vars), then the c[..]r patterns can be handy
and healthy. ;-)

It kinda depends on what one's used to.

(setq foo  '((1 (2 3) 4 . 5)))

(caar foo)         => 1        ..        I
(caadar foo)       => 2        ..-.      F
(cadar foo)        => (2 3)    .-.       R
(cadr (cadar foo)) => 3        .- (._.)  A(R)
(cddar foo)        => (4 . 5)  --.       G
(caddar foo)       => 4        .--.      P
(cdddar foo)       => 5        ---.      Ó
                                         or O(E)
                                         or M(N)
                                         or T(G)

YMMV.

Re: Elisp threading macros

[Augusto Stoffel]

On Wed, 28 Aug 2024 at 18:55, Eli Zaretskii wrote:

>> From: Garklein <garklein97@gmail.com>
>> Date: Wed, 28 Aug 2024 10:08:05 -0400
>> 
>> Would it be possible to have the dash.el macros `->', `->>', `-->',
>> and `-some->' in subr-x.el? I find them to be really useful, and would
>> love to see them in the base language.
>
> We might, but probably not using the names dash.el uses.

A while back, in bug#56875, I've suggested adding a `thread-as' macro
(like Clojure's -as->).  Given that we already have 2 of the 3 most
useful threading macros, I see no reason not to have the remaining one
(which might in fact be the most useful of all 3).

`-some->` is kinda neat but seems less important to me.

Re: [External] : Re: Elisp threading macros

[Alfred M. Szmidt]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 960 bytes --]

   Could different chars have been used than
   c, a, d, and r?  Of course.  The c and r
   aren't needed, and we could have used f for
   first and r for rest, or h for head and
   t for tails, or | for car and . for cdr, or
   - for car and @ for cdr:

    cadaar  =  frff  =  hthh  =  |.||  =  -@--
    caar    =  ff    =  hh    =  ||    =  --
    caaar   =  fff   =  hhh   =  |||   =  ---
    caddar  =  frrf  =  htth  =  |..|  =  -@@-

    cdadar  =  rfrf  =  thth  =  .|.|  =  @-@-
    cdddar  =  rrrf  =  ttth  =  ...|  =  @@@-
    caddr   =  frr   =  htt   =  |..   =  -@@

How do you pronounce frff? .|.|? or @@@-?

One thing that car, cadr, cdr, cdar .. have going for them is that
they are (semi-)words, with pronounciation rules.  No clue how to
pronounce -some-do-hickey-@-$-> ...  which is immensly useful when
explaning thing, it is much easier to say kə.ˈdɑɹ (CDAR) of a list
then 'the rest of the first' .. or whatever it would become.

RE: [External] : Re: Elisp threading macros

[Drew Adams]

[-- Attachment #1: Type: text/plain, Size: 1041 bytes --]

>    Could different chars have been used than
>    c, a, d, and r?  Of course.  The c and r
>    aren't needed, and we could have used f for
>    first and r for rest, or h for head and
>    t for tails, or | for car and . for cdr, or
>    - for car and @ for cdr:
> 
>     cadaar  =  frff  =  hthh  =  |.||  =  -@--
>     caar    =  ff    =  hh    =  ||    =  --
>     caaar   =  fff   =  hhh   =  |||   =  ---
>     caddar  =  frrf  =  htth  =  |..|  =  -@@-
> 
>     cdadar  =  rfrf  =  thth  =  .|.|  =  @-@-
>     cdddar  =  rrrf  =  ttth  =  ...|  =  @@@-
>     caddr   =  frr   =  htt   =  |..   =  -@@
> 
> How do you pronounce frff? .|.|? or @@@-?
> 
> One thing that car, cadr, cdr, cdar .. have going for them is that
> they are (semi-)words, with pronounciation rules.  No clue how to
> pronounce -some-do-hickey-@-$-> ...  which is immensly useful when
> explaning thing, it is much easier to say kə.ˈdɑɹ (CDAR) of a list
> then 'the rest of the first' .. or whatever it would become.

;-)

Yes.

[-- Attachment #2: c-icon.gif --]
[-- Type: image/gif, Size: 288 bytes --]

[-- Attachment #3: a-icon.gif --]
[-- Type: image/gif, Size: 282 bytes --]

[-- Attachment #4: d-icon.gif --]
[-- Type: image/gif, Size: 284 bytes --]

[-- Attachment #5: r-icon.gif --]
[-- Type: image/gif, Size: 258 bytes --]