Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.mobile.android > #155958

Re: AI test. Write an app. What app should we write? It needs to be useful.

Cross-posted to 3 groups.

Show all headers | View raw

On 9/16/26 1:16 AM, Paul wrote:
> On Tue, 9/15/2026 7:55 PM, T wrote:
>> On 9/15/26 1:43 PM, Maria Sophia wrote:
>>> AI test. Write an app. What app should we write? It needs to be useful.
>>
>> I write a bit in Raku (Perl 6).  When I get stuck,
>> I typically will ask ChatGPT.  So far the advice
>> I get back and the explanations are wonderful.
>>
>> Just write something you need.  Then when you get stuck,
>> ask ChatGTP.
>>
>> He is not so good at finding punctuation errors though.
>> I am forever gettig mixed up with my brackets.
>>
> 
> Why not ask the machine for a syntax check ?  :-)
> 
> It should, actually, be good at punctuation.
> There is a logic to it, there are plenty of examples
> (reinforcement) in the training set.
> 
> *******
> 
> For extra comedy, ask your LLM-AI to write you a
> 
>      Raku lint checker
> 
> and then you would have a standalone tool to check
> your Raku source. You could have it write that
> in some other language if you want, remembering
> that it "likes to write Python programs". I thought that
> was comedy gold, when I asked it what language it
> "liked" to write computer programs in, and it
> waxed on and on about Python. The answer looked
> like an advert. I was expecting an "I can do anything"
> kind of answer, and I was shocked when it somehow
> decided Python was the One Language.
> 
> The machine likes to do anything which is within the
> training set. For example, don't ask it to write APL
> programs, because at the time, there seemed to be nothing
> substantive in the training set.
> 
> It has poor measuring qualities, and it has no idea
> what is in the training set. It can only figure
> that out, indirectly, via High Reasoning mode.
> It can detect it is guessing, while High Reasoning.
> 
> When writing programs, it seems to write a series of comments first.
> Then it fleshes out each comment and writes code. If it doesn't
> have the wherewithall to write a particular subroutine, it can
> leave that as a comment, like it was a lazy student or something.
> 
>      subroutine sqrt(x)
>      {
>         // Put square root code here
>      }
> 
> This is one reason, you must skim-read any programs it gives
> you, as it is too sneaky to admit "I whiffed on parts of my assignment".
> It just quietly leaves divots in the turf like that.
> 
>     Paul
> 

My work around it to create the structure first
with all the brackets in place, then fill the inners
in.

Where I get in trouble is with
     if 0 { say "no" } elsif False { say "NO" } else { say "yes" };

when there are if statement inside the elsif's.

Raku has no "case" statement, like bash.

And the syntax checkers tell me here is something wrong
with the last line of the program instead of where I forgot
a closing "}".  Raku says they are working on it,

Back to comp.mobile.android | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

AI test. Write an app. What app should we write? It needs to be useful. Maria Sophia <mariasophia@comprehension.com> - 2026-09-15 20:43 +0000
  Re: AI test. Write an app. What app should we write? It needs to be useful. Hank Rogers <Hank@nospam.invalid> - 2026-09-15 16:55 -0500
  Re: AI test. Write an app. What app should we write? It needs to be useful. T <T@invalid.invalid> - 2026-09-15 16:55 -0700
    Re: AI test. Write an app. What app should we write? It needs to be useful. Paul <nospam@needed.invalid> - 2026-09-16 04:16 -0400
      Re: AI test. Write an app. What app should we write? It needs to be useful. T <T@invalid.invalid> - 2026-09-16 12:13 -0700
        Re: AI test. Write an app. What app should we write? It needs to be useful. T <T@invalid.invalid> - 2026-09-16 13:12 -0700
      Re: AI test. Write an app. What app should we write? It needs to be useful. Chris <ithinkiam@gmail.com> - 2026-09-17 07:22 +0000

csiph-web