Re: RFC idea

Philip Newton wrote:
>
>On 25 Sep 2000, at 10:03, Ben Tilly wrote:
>
> > I think David is confused about this situation, but what he
> > said is not entirely false.  Anyone who wants can get Perl,
> > make changes under the GPL, and release the hacked up version
> > under the GPL.  You would now have a GPL-only fork of Perl
> > which it is unlikely anyone would actually use, but you would
> > have a version of Perl with rather more strict redistribution
> > requirements than the current one.
>
>About the "it is unlikely anyone would actually" use -- isn't that what
>happened when ActiveState brought out ActivePerl at first? From
>what I gather, it was mostly because DOS/Windows ships without a
>compiler, and virtually all D/W apps ship only as binaries. So people
>are not used to having to compile stuff, let alone fetch a compiler. So
>if there's one super-duper, Larry-blessed version that makes you
>fetch a compiler, download source, untar (oops, have to fetch tar and
>gzip first and compile those), compile, install, etc., and one version
>that's different, modified, renamed, hacked up beyond recognition --
>but that ships as an easily-installed binary, a lot of people would go
>for that.

You gather incorrectly.

When they first came out it was the fact that they had done a
port of Perl to Windows.  Oh, they also gave access to lots of
native Windows APIs as well.  And this could all be integrated
into IIS.  (I believe that this was done on contract for
Microsoft, so they had access to information not generally
available at that time.)

This was before you could easily compile your own version of
Perl, and when your own version of Perl would be useless for
doing web work or Win32 administration.  It was not just
impossible to build your own version, you simply could not
get another version that did the same stuff.  I remember
those days.  I remember when you could use this version of
Perl that Sarathy put out (a port of 5.004_03 IIRC), or the
version that ActiveState put out, but for a lot of things you
could only use what ActiveState put out.

And you couldn't build your own.  Because you did not have the
source.

Since then those goodies have been largely integrated back into
Perl.  I think a couple of pieces couldn't be because of the
contracts with Microsoft, but equivalent functionality is
available.  Except you cannot create ActiveX servers in Perl
without paying.  (Conspiracy theorists should note in private
that you can in Python.)

The GPL is meant to prevent that exact situation.  Oh, you can
launch your own fork, put lots of goodies into it.  But you have
to distribute source to any customer who asks, who can then give
it away to the whole wide world.  And you have to give away all
of your extra doo-dads.  Which makes it hard to have a monopoly
that people disagree with.  My understanding is that the Artistic
License was intended to allow people to ship their extra doodads,
but not have them interfere with your branding of your stuff.

Now that is, in a nutshell, the argument for copyleft.  The draft
I produced has copyleft provisions, but they only apply to
people trying to use IP from Perl in changing the existing Perl.
There are also arguments against copyleft provisions, which is
why I have several times now asked how much people care about the
above situation.  Should Perl's license attempt to prevent the
above scenario from happening?  Should Perl rely on trademarks to
do the same thing?

>And apparently, shipping OSs without a functioning compiler is not
>the sole prerogative of MicroSoft; for example, HP-UX comes with a
>K&R only C compiler that (so I'm told) is explicitly intended only for
>rebuilding the kernel. For everything else, you're supposed to buy
>their commercial compiler (which does do ANSI). (Didn't Sun also ship
>some system without a C compiler?)

That is a non-issue as far as I care.  The point of copyleft
provisions is that anyone who wants to can readily hang up
their shingle, take your software contribution, and compete
with you.  That not every user can do that is minor.  Those
that want to learn how to, can.  (gcc is available.)

> > I take it as virtually axiomatic that if there are two free
> > versions of Perl out there, the one that has Larry behind it
> > will be the one that people will choose. :-)
>
>Philosophical people maybe, but an awful lot of couch potato-type
>newbies will choose the version that's easy to install and doesn't
>require mucking around with compilers.

It has happened before and shall again that kind people will do
as Sarathy did and offer their own compiled versions for free.

I really think this is less of a barrier than it looks.  All
else being equal, Larry Wall's version will win.  And if the
offering of precompiled versions is needed to make Larry's
version equal, that will happen.

Cheers,
Ben
_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.

Share information about yourself, create your own public profile at 
http://profiles.msn.com.