Re: master 1d9d07fb00e 3/3: (cl--typeof-types): Rework to fix some regressions

[Stefan Monnier]

> I see, ATM in objects.texi we have before the DAG "The Lisp Type
> Hierarchy for primitive types can be represented as follows:".

Oh, of course, I knew that [ he said, fumbling wildly.  ]

> For "primitive types" I meant types which comes to the user preloaded.
> Anyway I think we have either to change the doc or the DAG to be
> coherent.

Yeah, `transient-*` definitely doesn't count as "primitive types".
And neither do `xref-elisp-location` and `frameset-register`.

> In the manual I'd personally expect to see only the primitive
> (preloaded/predefined?) types, but maybe it's just me.

The types I manually listed in `syncdoc-type-hierarchy.el` can be
removed (or changed) at will.  I don't have a strong opinion about them.

> This leads me
> wondering if syncdoc-type-hierarchy.el shouldn't automatically render
> the DAG using all types of an emacs -Q using like to
> 'comp--compute-typeof-types' to pick all types defined ATM.  It would be
> less arbitrary but maybe too verbose?

It would still be fairly arbitrary, IMO, but easier to justify :-)
We'd probably want to weed out some of the internal types that could
show up, tho.


        Stefan