Re: Release plans

[Thomas Lord]

Alan Mackenzie wrote:

Now you're playing verbal games with me. 
  

No, I'm not.

No, of course not. ["Defeatism" is] one of those handy little words which can always
be defended as rational and objective, yet at the same time give a very
good impression of an ad hominem attack and have the same effect.  It can
be used to appear to be exceptionally rude, yet without exposing the
writer to general censure.  Except, of course, you're not being rude
here, you're just being objective and rational, which is very reassuring.
  

If we just use words in simple ways, communication with
near strangers over the Internet can work out.   If we always
look for hidden meanings or sly insults, communication is harder.

"Defeatism" means a mode of strategic or tactical reasoning in which it
is assumed that the only choices are between various losses.  The
assumption in the dynamic loading decision is that either GNU Emacs
loses by not having a dynamic loader, or GNU Emacs loses by having
non-free, C-level add-ons catch on.  Defeatism is a kind of "planning
to lose" and if defeatism is the only reasoning applied then it is
self-fulfilling: loss of some kind is assured.
    

"Defeat" is utter loss; 

No, it isn't.   You can see that it is not because of the existence
of two cliche phrases:  "defeat at battle" and "utter defeat".
"Defeat" means an "undoing" or a loss.

"Defeatism" is an assumption that loss of one kind or another
is inevitable, then a choice of action aimed to minimize loss.
"Defeatism" is a kind of "null upside investment strategy"
by which I mean that defeatism is the way of spending
to minimize loss, EVEN AT THE EXPENSE of possible
gain.    When lots of people adopt defeatist tactics in the
stock market, that's called a panic.

it's when your king is in checkmate, when when
the whistle blows after 90 minutes your opponents have scored more goals,
when the enemy troups have routed your army and destroyed your strategy
to the point where the only sensible action is to surrender.
  

That's mostly in your head.   In a tragic context,
defeatism could refer to an inevitable and utter defeat
like that but the word "defeat" itself has much broader
meaning.

The inability to use dynamically loaded binaries in Emacs is like none of
these things.  It's an inconvenience, possibly minor, possibly major.
But it is _nothing_ like the utter rout implied by the word "defeatism".
  

I guess the question you are speculating about is how "important"
dynamic loading is.    I don't think either of us really knows but
I can guess too:

I'm really impressed by the roles dynamic loading has played
in the LAMP stack, particularly loading of modules into scripting
languages and loading of modules into Apache and other
web servers.

It makes sense, in retrospect, that it would be such an influential
feature.  It's a kind of combinatorics phenomenon.  If there are
M programs with dynamic loaders, and N dynamically loadable
libraries, then there are M*(2^N) possible run-time environments!
That's very flexible.   And what's more, because of the way
dynamic loading works, ANYONE can increment N without having
to bother any upstream maintainers or have patches accepted -- it's
always possible to add a new library.   The result of a dynamic loader
in the LAMP stack is a super-exponential explosion of possible
configurations of free software components and the the result of
that circumstance is the enormous success of the stack (and the
ongoing development of lots of components that "plug in" to it).

Would dynamic loading in GNU Emacs matter as much?
I'm sure nobody knows and that the answer depends largely
on the design of the dynamic loading mechanism.   Nevertheless,
the potential upside is huge, if the LAMP environment is any
indication.

The *cost* (in labor and other real expenses) of adding dynamic
loading is, I suspect, pretty low.    A crappy job of it should be
basically free.   A very good job of it should still be pretty
"cheap".   So, we're talking about a penny stock:   cheap to add
dynamic loading and a huge potential upside. 

Here is what you don't see, or at least refuse to consider: a non-free
add-on which becomes popular could be used maliciously to remove freedom
from Emacs.  Seen through your spectacles, every user is free to chose to
use that add-on or not, so there's no problem.  I'm making one last
effort in the post to help you see where you are wrong (see below).  If
this doesn't help, there's no sense in continuing the conversation.
  

Your ".nyet" license?   So, the nightmare scenario is 10s of millions or more
of new Emacs users, but all "addicted" to the ".nyet" add-on?

That sounds to me like a battle won for free software.  Not a war won
but a major battle:   The next step is to whittle away at the advantages
of the ".nyet" add-on and then free GNU Emacs has 10s of millions or
more of new users.

[*] inessential = "not composing the essence of", which is not identical
to "unimportant".
  

The combinatorics experience of the LAMP stack with
add-ons suggests that calling a dynamic loader "unimportant"
is at best premature.

I happen to believe that there is *power* in freedom.  If both the free
and non-free army is given the chance to create add-ons, the free army
(if it plays intelligently) can obtain more benefit from the
opportunity in the long run.  The same advantage, offered to both
sides, is worth more to the free side.
    

I don't think you understand power and its mechanisms, such as dominance,
deceit, lies, disinformation, demagoguery, deviousness, blackmail,
ridicule, manipulation, .... at all.  Richard most assuredly does, which
is why I am happy to trust his judgement on this matter.
  

Uh, you might be surprised.

-t