bug#36372: 27.0.50; replace-regexp-in-string skips START first chars in return value [PATCH]

[Drew Adams]

> So it isn't used in-tree at all.  Since it's an (in my opinion) pretty
> useless parameter as it's currently implemented, I think we should
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> deprecate the parameter and remove the documentation.
> 
> Then, perhaps, in ten years time we can reintroduce the parameter with
> better semantics.  :-)

1. Whether it's used in-tree is not sufficient.

2. Whether it is useless "as it's currently
   implemented" is not sufficient.  That's what
   implementation bugs are about.

3. But I agree that it is likely very rarely used,
   if at all.  (Searched my code and found no
   matches, as one example.)

3. IF it has a use, when fixed as you see fit,
   then fix it, and call out the fixed behavior
   as an incompatible change.  In that case, do
   not deprecate it.  (I don't know that it has
   a use, if fixed, hence "IF".)