Issue #14276 has been updated by shevegen (Robert A. Heiler).


> Developing production code without an available debugger
> is not manageable.

This appears to be somewhat opinionated since I do just
fine without a debugger (except for pp, just as tenderlove
once said that he is a puts debugger, I am a pp debugger).

However had, that is not to say that I necessarily disagree
with you. If possible it would be fine to rely on a more
stable API.

> If we succeeded to do that we could present ruby to the
> new comer with a simpler development environment easing
> the entry level.

Here you make the assumption that, all the things included
and managed, automatically make ruby simpler and thus easier.

But in reality, some suggestions also work in an orthogonal
manner - having to know how testing works, such as making
minitest a mandatory part, is another barrier for newcomers.
So is having to know rdoc; I still don't know how to avoid
rdoc from making certain assumptions leading to strangely 
formatted autogenerated documentation.

That is also not to say that documentation is not important,
quite the opposite. It is very important. I started to use
autogenerated README.md files a while ago, and so far it
works quite fine as entry-level documentation.

> We could put simple documentation in ruby-lang.org to
> show how to achieve a professional development
> workflow with ruby with no fuss.

Who or what defines a "professional development workflow"?

And can this be extrapolated to other people?

Matz once said in an interview that people are different.
It's one reason why ruby is so flexible - you can adjust
it to many different use cases (http://www.artima.com/intv/ruby.html)

I am not necessarily against the proposal, mind you. I 
think your core statements made are perfectly fine.

But there is not really "one true way" in ruby alone.

What may be possible is some sort of tutorial or more tutorial
on the ruby webpage or a link from it; someone may have to
write the content of what constitutes that particular 
workflow that you may want to have first.

I agree in the sense that ruby should also focus and help attract
newcomers, without making things worse for the older ruby hackers.

> A full TDD environment available upon installation it
> is a great marketing headline in my perspective.

Ruby should not be pushed into a particular philosophy "just
like that". What about those ruby hackers who, for any reason,
do not use TDD? We could just as well say that all frameworks
must use the MVC pattern and that everyone has to use agile
and vim users write better ruby code than emacs users.

I agree with your core statement about debuggable APIs.

I assume that the ruby core team may need more information
how people use (certain) debuggers and what are the main
reasons why the debuggers break/broke. Whether it were
syntax changes or C-internal reasons that broke extensions.

----------------------------------------
Feature #14276: Ruby core debugger API
https://bugs.ruby-lang.org/issues/14276#change-69165

* Author: dsferreira (Daniel Ferreira)
* Status: Open
* Priority: Normal
* Assignee: 
* Target version: 
----------------------------------------
I would like to propose the creation of a ruby core minimal API for a Debugger.

Historically we have been relying on gems as providers of debug functionality.
This has caused problems with debugger APIs being broken and the debugger gems being outdated by ruby releases.
This happened in the transitions from 1.8.* to 1.9.* and 1.9.* to 2.**

Currently ruby community relies mostly on *byebug* gem (+ pry) for debugging purposes.

Recently a new bug [14226](https://bugs.ruby-lang.org/issues/14226) was raised due to changes in ruby internals that broke *byebug*.

I consider that the debugger should be a core part of a language like ruby.
Developing production code without an available debugger is not manageable.

For this reason I believe we should have a way of guaranteeing that each ruby release will not break the debugger.

Also having full control of the debugger API is very convenient for future purposes.
This way third party entities can rely on it to develop productivity tools.

I see two ways we can accomplish this:

- Defining core Debug API namespace where we would create the minimal interface for ruby core debug functionality (Not sure if only cruby or a broader API adopted and designed in collaboration with the other ruby implementations).
- Creating debug gem in standard lib with the same purpose. Having the guarantee that the unit tests will run for every ruby release.

The main goal in my mind would be to have a bare minimum TDD development environment available upon ruby installation
with each bare minimum API providing the base functionality to allow the development of full fledged solutions capable of rivaling with the best solutions in the market.
For that goal I would include here for consideration the same approach for other functionalities such as:
* Coverage
* Documentation
* Benchmarks
* Tests
* Performance
* Linter

I believe we have currently a good situation with regards to:
- Documentation (Rdoc)
- Tests (Minitest)

Why not try to replicate those examples for the remaining areas?
Are documentation and tests so much different from the rest?

If we succeeded to do that we could present ruby to the new comer with a simpler development environment easing the entry level.
We could put simple documentation in ruby-lang.org to show how to achieve a professional development workflow with ruby with no fuss.

A full TDD environment available upon installation it is a great marketing headline in my perspective.

Just to finalise.
The goal here is not to have ruby core replacing the existing gems.
The goal is to give a clear path for these gems while allowing others to develop even further in top of a sound base
without worries regarding future support on the APIs being used.





-- 
https://bugs.ruby-lang.org/

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>