Re: "Emacs 21"

[Stuart D. Herring]

>       Most of the rest are comparisons between 20 (or
>     19) and 21 in terms of error reporting, mark handling, echo area
> resizing,
>     and mode line formats.
>
> I am not sure whether you are talking about passages that describe
> differences between specific Emacs versions, or about teaching the
> user how to write code that tests the Emacs version.  Which?

There are both; the passage you quote was talking about describing
differences, and the paragraph after it (that you did not quote) was
talking about testing the Emacs version.

> The former, we should remove.  This manual is meant for the current
> Emacs version.  Discussing older Emacs versions is a digression which
> gets in the way of the intended purpose.

Are you sure?  This is the Lisp introduction, and for version 21 included
all these discussions.  Certainly 20 is less relevant once 22 is out, and
19 even less so, but do you really mean to remove all of it?  (I am not
arguing either way on the matter, just describing past practice.)

> The latter, we should update (if necessary) so that they handle Emacs
> 22 properly.

By "properly", do you mean "recognize as distinct from all prior (and
perhaps subsequent) Emacs versions, or do you merely mean that they should
not produce literally false output (e.g., diagnosing Emacs 22 as 20)?  Or
do you mean that all such tests should make logical divisions such as
"older than Emacs 21" and "not older than Emacs 21", regardless of which
divisions they make?  Finally, can such tests use `emacs-major-version'
and such, introduced in 19.23?

>       I'm not sure
>     what to do with the CVS references.
>
> If they are just intended as example version numbers, any value is
> fine.  If they are meant to be the version of a recent Emacs, just
> update them to 22.1.

They are in most cases part of example filenames that sometimes will not
exist on users' machines (e.g., lisp/TAGS and lisp/abbrev.el, the latter
for an example of `directory-files-and-attributes').  The only mention of
the version number outside of a path is as a mention of where the named
file came from (that is, the CVS Emacs sources for version foo).

So they are neither example version numbers nor versions of recent
Emacsen; they are filenames and sources of files.  The files don't seem
particularly relevant since they are from an old CVS Emacs version, and
perhaps all the text that discusses them would be better rewritten to
discuss toy (or even imaginary) files.

> Can you do these things and send diffs?

Please send along clarifications (and more guidance about the CVS names)
and I'll do just that.

Davis

-- 
This product is sold by volume, not by mass.  If it appears too dense or
too sparse, it is because mass-energy conversion has occurred during
shipping.