Obfuscated Eiffel? (I know it's not, but...)

[Jimmy Johnson]

My last post reminded me of this.

What in the world is:

     ⟳ o: other ¦ force (o, @ o.key) ⟲

and

     ∀ o: other ¦ item (@ o.key) = o

Answer:  it is the implementation and post-condition of `merge' from HASH_TABLE, but ... WHY?

I know the mathematicians here love this stuff, but it sure is hard for me.  Because I don't use these constructs often, I have to always look them up.  But that is the problem.  I don't mind the construct, if I could easily get it translated.  Where do you find a explanation of these constructs.  How do I google it?

I duck-duck-goed this:  "y = (x > 10) ? (x * 2) : (x / 2)".  Since I don't use C++ very much, this was a bit obfuscated.  Took three tries with duck-duck.  Put "C++" before it and "meaning" after it and voila, the search assistant explained it nicely.

Internet searching for the above Eiffel construct [or most things Eiffel] rarely gives meaningful answers.  It is really frustrating.

Any advice?  [other than, "write it down this time, Johnson!]

thanks,

jjj

[Eric Bezault]

On 17/12/2025 0:43, Jimmy Johnson wrote:
> Internet searching for the above Eiffel construct [or most things
> Eiffel] rarely gives meaningful answers.  It is really frustrating.
>
> Any advice?  [other than, "write it down this time, Johnson!]

Larry, will AI be able to fill the gap here?

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

[Liberty Lover]

AI can certainly fill the gap if it has the right and curated data to explore and pull from. 

As with anything: Garbage in gives garbage out. It is one of the reasons that AI on typical programming language tech-stacks produces a very high degree of bugs. It was trained on buggy code and lots of it. And the bias towards the bugs is difficult to abate simple because of the way AI training appears to work.

This is why I would love to see an Eiffel AI trained on curated Eiffel code and Eiffel-based education and research materials. This seems very much the line of reasoning for medical, legal, and other mission-critical AI training data and the models they produce. 

There is a thought in my head that at some point I would love Simple Eiffel to rise high enough that it could do that job, creating and dogfooding its own model and cutting the cord to the VERY expensive Claude Opus 4.5 model that I am presently using.

Until that day, I say ... don't put your notebook away just yet. But hold on to your hat ... the answer might be just around the corner.

Now, for me ... back to getting Simple Eiffel fully into Linux cross-platform compliance. It's getting close!!

--
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 visit https://groups.google.com/d/msgid/eiffel-users/b91e7fae-817a-435a-84c1-2ef720cc4a91%40gobosoft.com.

[javier...@gmail.com]

You can read about the Symbolic form of loops here
https://www.eiffel.org/blog/Alexander%20Kogtenkov/2020/03/symbolic-forms-loops

-- Javier