Re: rethinking @def*

[Werner LEMBERG]

> The "old" commands do not have much of semantics, and when they
> have, these semantics are inconsistent in different places in the
> manual.  Therefore new commands would not have better semantics,
> they would simply have semantics that are currently lacking.
> Therefore it seems wrong to me to have two sets of @def* commands,
> with the same syntax, one without semantics, and one with semantics.
> Only one set of commands with clear semantics should be enough.

I agree with you with about the goal.  However, I disagree with your
suggested way to that goal.

Compare this to a C library 'foo' that wants to have a stable ABI for
all of its functions, making older programs link and run successfully
with newer library versions: it is impossible to change any public
header files without breaking it.[*] Obviously, I equate the ABI with
texinfo's output formatting.

The only way to introduce a new behaviour for a broken function 'bar'
is to implement a new function 'bar2', declaring the old function
'bar' as obsolete.  I think that exactly such an approach is
appropriate for texinfo functions/macros that haven't changed for
decades.


   Werner


[*] This is a little white lie, but the details are not important
    here.