Re: Market pressures for more reliable software

[ewill@lexideb.athghost7038suus.net (The Ghost In The Machine)]

In comp.lang.java.advocacy, Roedy Green
<roedy@mindprod.com>
 wrote
on Thu, 14 Jun 2001 20:09:41 GMT
<jj5iitksq0v75aavlrqpouuq9rmmdepdjj@4ax.com>:
>On Thu, 14 Jun 2001 17:32:21 GMT, Ted Dennison<dennison@telepath.com>
>wrote or quoted :
>
>>"Open" doesn't quite reach it, because there's no implication that
>>the right to redistribute is present, only the right to inspect.
>
>Let's see what we can come up with for a less ambiguous term for "free
>software".

<voice mode=paranoia>

The FSF license is inherently viral.  While allowing for a number of
operations on the source code, the following requirements must
be met:

[1] The authors of previous modifications must be preserved.  You can of
    course put your name in there for your bug fixes -- in fact, you
    have to.

[2] The source code of the GPL portion must be available on a
    machine-readable medium to anyone who is sold the derived work.
    For example, if one uses the GNU-licensed readln library, one
    must make it available to anyone who uses your product.  (This
    in itself isn't too onerous, but...)

[3] If your work can be shown to be a derived work of the GPLed
    software, then the entire work must be published under the GPL.
    (Hence the term "viral".)  There might be some issues here, of
    course -- is a proprietary, sophisticated CAD program which
    happens to depend on the GNU readline library a "derived work"?
    Or merely a user of the readline library's functionality, which
    happens to solve a particular subproblem?

    Of course, if one takes, say, geda (which is a GNU-licensed
    CAD program), and modifies it by adding a sophisticated but minor
    input method (assuming it hasn't done such already), and releases
    it, then that is most definitely a derived work.  At least, I
    for one would think so...

</voice>

Of course, this may be a feature; one nice thing about GNU-licensed
stuff is that one does have the opportunity to inspect and if
necessary modify it, and then release the modifications back to the
pool.  With so many eyes, bugs can't hide too well, but there
is the issue that, if interest wanes, the software will have problems.
One might call it software evolution in action....

>
>Basically the only right you have is to inspect the software. You may
>still be restricted from selling it or modifying it. You may still
>have to pay to look.

I'm not sure how, unless you're talking about a different license
(e.g., the Sun Community-type deal).  Mozilla's license is also
slightly strange -- part of that might simply be the fact that
Netscape was originally proprietary.

>
>"white box" software as opposed to "black box" software is my top
>choice.
>
>The term has some unwanted NeXT connotations. It also might me
>confused to mean generic cheap packaging.

It's also confusable with a specific testing method, although
I'm not sure if the term is precisely enough defined or not;
"black box" testing can only use public interfaces, but
"white box" can use anything and everything it can lay its
hands on.  A rough analogy would be testing a (populated?) PC board
by only electrically probing the fingers which plug into the motherboard,
versus probing anywhere it makes sense to on the board itself.

I can't say anything about NeXT, although I've been exposed to it;
it looked neat at the time, :-) although quite underpowered by
today's standards.  (Then again, that's a hardware issue; NeXT
could probably run on today's computers without problem, especially
since there is (was?) an x86 version floating around.)

>
>some other possibilities:
>

[rest snipped]

If one wants to be ultra-pedantic about it, how about
"[license-name]-licensed software"?  E.g., readline is
GPL-licensed software.

It's not free, although it's designed to be very easy to work with.
(A lot more so than, say, Microsoft Word's EULA.  It's not even
legal to disassemble the darned thing, and one wonders whether
one can even disassemble its output -- a Word-formatted file.)

-- 
ewill@aimnet.com -- insert random misquote here
EAC code #191       25d:20h:07m actually running Linux.
                    Most likely, no neutrinos were found during this message.