Symbolic forms of loops

[Alexander Kogtenkov]

Adoption of Unicode in Eiffel goes on. Some messages in this mailing list mentioned one such change: symbolic forms of loops. They can make code more concise and closer to traditional math. Assertions can benefit from this improvement, in particular. My blog post at the link below provides some examples of the new syntax and its relation to the keyword-based one:

 

    https://www.eiffel.org/blog/Alexander%20Kogtenkov/2020/03/symbolic-forms-loops

 

Happy Eiffeling,

Alexander Kogtenkov

[Thomas Beale]

Hi Alex,

I was just contemplating this:

⟳ c: s ¦ print (c.name) ⟲

Why not just use the initial symbol, and forget the final one, i.e.

⟳ c: s ¦ print (c.name)

That seems more consistent with the other operators.

- thomas

[Eric Bezault]

On 01/04/2020 23:40, Thomas Beale wrote:
> Hi Alex,
>
> I was just contemplating this:
> |⟳c:s ¦print (c.name)⟲|
>
> Why not just use the initial symbol, and forget the final one, i.e.
> |⟳c:s ¦print (c.name)|
>
> That seems more consistent with the other operators.

This could only work if we restrict this notation to only
one instruction. So the following:

⟳ c: s ¦ print (c.name); print ("%N")

would be a loop followed by an instruction to print a newline
(and not a loop with two instructions).

Personally, this would have my preference. I think that I already
suggested something along these lines to ISE. After all, the
symbolic form of loops is already at subset of what could be
expressed with the `across` form (e.g. no loop variant and
invariants). So this restriction on the number of instructions
should not be so much of a problem: if one wants more, just
use the `across` form.

--
Eric Bezault
mailto:er...@gobosoft.com
http://www.gobosoft.com

[Thomas Beale]

Eric,

I agree with all of that.

- thomas

[Larry Rix]

Is this in the following version? If so, I am unable to compile it (I get a syntax error).

Installation information:

Version = EiffelStudio 19.12 (19.12.10.3819 Community Edition - win64)

Monitor DPI = 96

$ISE_EIFFEL = C:\Program Files\Eiffel Software\EiffelStudio 19.12 GPL

$ISE_LIBRARY = C:\Program Files\Eiffel Software\EiffelStudio 19.12 GPL

$ISE_PLATFORM = win64

$ISE_C_COMPILER = msc_vc140

[Alexander Kogtenkov]

Dear Larry,

 

I’ve tried the following code with EiffelStudio 19.12.10.3768 and it prints TrueFalse

 

            a: ARRAY [INTEGER]
        do
            a := <<3, 5, 7>>
            print (∀ x: a ¦ x > 2)
            print (∃ x: a ¦ x = 2)

 

Is it the same for you?

 

(It looks like the download page lists a slightly higher version, I can test with it as well, but I guess, it should produce the same result.)

 

Regards,

Alexander Kogtenkov

 

 

Понедельник, 20 апреля 2020, 22:10 +03:00 от Larry Rix <lar...@moonshotsoftware.com>:
 

--
You received this message because you are subscribed to the Google Groups "Eiffel Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to eiffel-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/eiffel-users/5990ed8a-4d74-4c27-a874-98e33658b924%40googlegroups.com.

 

[Larry Rix]

Hi Alex,

I copied your code over and it ran with no issues.

So, now I am wondering what I am doing wrong.

[Larry Rix]

I have the following:

output: REAL_64

--

local

l_sum: REAL_64

do

⟳ ic: inputs ¦ l_sum := l_sum + (ic.value * ic.weight) ⟲

across

inputs as ic

from

l_sum := 0.0

loop

l_sum := l_sum + (ic.item.value * ic.item.weight)

end

Result := 1 / (1 + exp (- l_sum - bias))

end

Which produces a syntax error:

Rule Description Location Position Severity

1 Syntax Error Syntax error at line 47. AI_SIGMOID_NEURON (elotto) 47, 62

Syntax error at line 47 in class AI_SIGMOID_NEURON

        do

            ⟳ ic: inputs ¦ l_sum := l_sum + (ic.value * ic.weight) ⟲

----------------------------------------------------------------------^

            across

[Alexander Kogtenkov]

There is some issue with the parser. The current workaround could be adding a semicolon after the instruction in the loop:

 

   ⟳ ic: inputs ¦ l_sum := l_sum + (ic.value * ic.weight); ⟲

 

We’ll take a look what’s wrong. Thanks for noticing it.

 

Regards,

Alexander Kogtenkov

 

 

Larry Rix <lar...@moonshotsoftware.com>:

[Eric Bezault]

FYI, this code compiles and works fine with the Gobo compiler.
So it should be fixable: the issue is not in the grammar
but in its implementation in EiffelStudio's parser.

--
Eric Bezault
mailto:er...@gobosoft.com
http://www.gobosoft.com

> 1Syntax ErrorSyntax error at line 47.AI_SIGMOID_NEURON (elotto)47, 62