I have amused myself with implementing this and other brevity constructs useful for working with strings, vectors and other sequences.
In the repl one can now write:
(define-struct person (first-name last-name age height) #:transparent)
(define persons
(vector (person "John" "Doe" 42 1.80)
(person "Jane" "Dae" 43 1.81)))
(define i 1)
(string (@ persons i .first-name 0) (@ persons i .last-name 0))
The file string-lang.rkt implements a form {...} that allows one
one to reference the i'th element of lists, vectors, strings etc using
the syntax {v i}.
For substrings, sublists and subvectors and subsequences one can write {v i j}.
See the test file for examples.
There is one problem though.
The syntax
{john .age} should evaluate as (person-age john)
since john is an instance of a person structure.
Using information from structure-type-info I can construct the
identifier person-age, but I can not figure how to retrieve the
actual accessor without using eval.
This means that {john .age} only works in the repl and not
in a module context.
Any pointers for improving the macro dot (which implements this)
is appreciated.
--
Jens Axel Søgaard
2012/5/10 Matthias Felleisen <[email protected]>:
>
> Thanks for the nice illustration. It would be
>
> (define initials
> (string (string-ref (id-firstname (vector-ref person i)) 0)
> (string-ref (id-lastname (vector-ref person i)) 0)))
>
> but even with that we don't get the density of some path expression:
>
> (define initials (string (@ person i firstname) (@ person i lastname)))
>
> And now we're comparing with
>
> initials = person[i].firstname[0] + person[i].lastname[0]
> or
> initials = (@ person i firstname) + (@ person i lastname)
>
> Hmph.
>
>
>
> On May 10, 2012, at 5:18 PM, Justin Zamora wrote:
>
>> In my experience, the heaviness of Racket doesn't come from words like
>> "define", etc. It comes in certain categories of programs that deal
>> extensively with strings, vectors, and structured data. For example,
>>
>> initials = person[i].firstname[0] + person[i].lastname[0]
>>
>> This is very readable and useful. It can be written quickly and is
>> read and understood easily The overloading of "+" (or your operator of
>> choice) and the implicit coercion of characters to strings is exactly
>> what is wanted here. Even evaluating person[i] more than once doesn't
>> clutter up the expression.
>>
>> Compare this to the equivalent Racket:
>>
>> (define initials
>> (string-append (string (string-ref (id-firstname (vector-ref person i)) 0))
>> (string (string-ref (id-lastname (vector-ref person i)) 0))))
>>
>> For this sort of thing, the Racket version is much harder to write,
>> read, and verify. It would be nice to have something akin to
>> at-expressions that would allow such expressions to be written more
>> clearly.
>>
>> Justin
>>
>> On Thu, May 10, 2012 at 4:00 PM, Matthias Felleisen
>> <[email protected]> wrote:
>>>
>>> I will assert something about readability:
>>>
>>> Racket programs look heavy when compared with Haskell programs.
>>>
>>> This is probably true for Python instead of Haskell, too. It is also true
>>> for ML. I conjecture that part of that heaviness comes from wide lines,
>>> long names, deep nesting. Who knows. I don't even know how to measure this
>>> kind of property.
>>>
>>> At this point, I can express certain ideas more easily in Racket than in
>>> Haskell, Python, ML or whatever, which is why I am fine. But if this
>>> advantage ever disappeared, heaviness would definitely be a factor to weigh.
>>>
>>> -- Matthias
>>>
>>>
>>>
>>>
>>>
>>>
>>> On May 10, 2012, at 3:49 PM, ozzloy-racket-users wrote:
>>>
>>>> i didn't assert that word length has nothing to do with readability, just
>>>> that word frequency has more impact on reading time than word length.
>>>>
>>>> On Thu, May 10, 2012 at 3:39 PM, Luke Vilnis <[email protected]> wrote:
>>>> I can only speak for myself but I think it's a bit much to assert that
>>>> word length has nothing to do with readability. Heck, maybe that's even
>>>> true for you, but not for everyone. I have certainly felt it to be an
>>>> issue. If the "define" keyword was 50 letters long it would definitely
>>>> have an impact on my ability to read code - it seems to be an issue of
>>>> degree, not existence.
>>>>
>>>> On Thu, May 10, 2012 at 3:26 PM, ozzloy-racket-users
>>>> <[email protected]> wrote:
>>>> am i the only one that thinks not having abbreviated names for anything is
>>>> good?
>>>> i like not having "def". especially if it's going to be redundant.
>>>> i see this as a slippery slope i don't want to go down.
>>>> it annoys me when switching to other languages to have to ask: which way
>>>> of shortening "function" does this language go with? was it "fn"? maybe
>>>> "fun"?
>>>> if the language has a strict policy of not using short versions of words,
>>>> i don't have to guess.
>>>>
>>>> as for "def" being easier to read than "define", that's not true. word
>>>> frequency has more impact on reading time than word length for normal
>>>> reading. having more aliases makes both less frequent, so adding "def"
>>>> could plausibly make reading both take longer. most people read whole
>>>> words at a time, rather than letter-by-letter.
>>>>
>>>>
>>>> On Thu, May 10, 2012 at 2:56 PM, Grant Rettke <[email protected]> wrote:
>>>> There is always pretty mode in Emacs.
>>>>
>>>> On Thu, May 10, 2012 at 1:45 PM, Ray Racine <[email protected]> wrote:
>>>>> FYI for those who may not know. Racket supports λ as an alias for lambda.
>>>>> ctrl-\ in DrRacket.
>>>>>
>>>>>
>>>>> On Thu, May 10, 2012 at 1:59 PM, Nikita B. Zuev <[email protected]>
>>>>> wrote:
>>>>>>
>>>>>> +1 for `def' as alias for `define'.
>>>>>> May I also suggest `fun' for `lambda' alias?
>>>>>> Three letter names are the best =)
>>>>>>
>>>>>> (well one can always do it with require rename-in)
>>>>>>
>>>>>> --
>>>>>> Regards,
>>>>>> Nikita B. Zuev
>>>>>> ____________________
>>>>>> Racket Users list:
>>>>>> http://lists.racket-lang.org/users
>>>>>
>>>>>
>>>>>
>>>>> ____________________
>>>>> Racket Users list:
>>>>> http://lists.racket-lang.org/users
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> http://www.wisdomandwonder.com/
>>>> ACM, AMA, COG, IEEE
>>>>
>>>> ____________________
>>>> Racket Users list:
>>>> http://lists.racket-lang.org/users
>>>>
>>>>
>>>> ____________________
>>>> Racket Users list:
>>>> http://lists.racket-lang.org/users
>>>>
>>>>
>>>>
>>>> ____________________
>>>> Racket Users list:
>>>> http://lists.racket-lang.org/users
>>>
>>>
>>> ____________________
>>> Racket Users list:
>>> http://lists.racket-lang.org/users
>
>
> ____________________
> Racket Users list:
> http://lists.racket-lang.org/users
--
--
Jens Axel Søgaard
string-lang.rkt
Description: Binary data
test-string-lang.rkt
Description: Binary data
____________________ Racket Users list: http://lists.racket-lang.org/users
