Regarding the "TheOtherGCCBasedFortranCompiler" Wiki Page.

[Brooks Moses]

Tobias Schlüter wrote:
> Brooks Moses wrote:
>>Until I edited it just now, the bulk of the page consisted of a set of 
>>completely unsupported allegations regarding Andy Vaught's GPL 
>>violations, and a factually incorrect statement that releasing binaries 
>>compiled under g95 would cause the software to become Free Software [1].
> 
> How can one support the 'allegation' 'A doesn't respond to mail requesting GPL
> compliant source'?  If you searched the old g95 list archive you could find
> one or two mails where this is mentioned, does that count as support?  As for
> the 'alleged' obfuscation of the code, you can look at sourceforge's cvs which
> Andy updated approx. once in three months, just have a look at
> <http://cvs.sourceforge.net/viewcvs.py/g95/g95/check.c?rev=1.42&view=log> and
> pay special care to look at the numbers of lines changed.

Those two complaints were removed from the page several months prior to 
my editing; thus, I was not referring to them.  Meanwhile, I would have 
considered the second half of your sentence to be quite sufficient 
support for claims about obfuscation of the code.

I was also mostly referring to the allegations of what constituted GPL 
violations, which to me did not seem well supported by the actual text 
of the GPL -- I think such things need direct quotes, not paraphrases.

I also think that things like "doesn't reply to mail" that _are_ 
difficult to support in any meaningful way shouldn't be made as public 
allegations in a forum like that, regardless of their truth.

>>I have cleaned it up somewhat, by providing documentation of the 
>>substantiable claims of GPL violations (which are, honestly, a pretty 
>>minor subset of the ones alleged), and by correcting the statement 
>>regarding binaries compiled under g95.  I have also provided 
>>documentation as to why the remaining claims of GPL violation are 
>>erroneous, which I am leaving in place for now.
> 
> I don't see how the mentioning of years can be "prominent notices stating that
> you changed the files and the date of any change", esp. WRT to the "any" part.
> And even if that were the case, Andy didn't update the copyright years.  I
> don't know if he does now, I didn't bother looking into g95's in a while.

It's a debatable position, I agree -- that's why I stated it as "the 
opinion of this author", not as fact.  The "I don't know if he does now" 
part is why I think these things are best accompanied by "As of $DATE, 
such-and-such is the case" statements.  Even if we looked now, it would 
be out-of-date soon.

I admit that I made an unsubstantiated assumption that he had updated 
the copyright years, and I should have done that.

> I don't see why you kept both your second points, what's the difference?

Convergent evolution of two different points, which should have resulted 
in one of them being removed.  Mea culpa.

>>It is my opinion that this entire set of allegations and complaints 
>>should be removed entirely, and replaced by a very short, simple, and 
>>non-accusatory statement of why it was not possible to continue to 
>>include g95 as part of the GCC suite of compilers.  GFortran should 
>>present itself as a professional effort, and should be above this sort 
>>of petty bickering.
> 
> It was not possible, because Andy Vaught stopped responding to mail, closed
> CVS for everybody besides him, stopped using CVS and even before then, never
> responded to any offers of cooperation with GCC (even though g95's webpage
> clearly stated inclusion in gcc as the project's goal).  His only response WRT
> that was that he would never change his coding style, and conform with the GNU
> coding standards.  If you're interested, I'll try to dig out the relevant
> mails.  They're still on sourceforge's mailing list server.

I'm not sure I want to spend the time to read all the relevant emails to 
see all the details -- my impression is that the situation was rather 
unpleasant.  And I think the goal of the "official story" should be to 
omit as many unpleasant details as possible, so thus I suspect that 
there will be very few details left in it at all.  :)

(To explain a distinction: I, personally, am willing to take your word 
for things.  I am also perfectly happy with putting up an "official 
story" publicly without much support.  I just think any accusations made 
to the public at large should be clearly documented -- my complaints 
were meant to be "I don't like how this was presented", not "I don't 
believe you".)

I think what's appropriate to the page is something akin to the short 
phrasing on http://www.fortran-2000.com/ArnaudRecipes/CompMuseum.html: 
"In January 2003, Andrew Vaught decided to develop g95 single-handedly 
with occasional source code releases. Consequently, the project forked 
and a new project, gfortran, was integrated first in an experimantal 
branch of GCC then in the main line."  Or, in short, a polite way of 
saying "Andy took his ball and went home, so we regrouped and carried on 
without him." -- I don't think there's any benefit to describing how he 
took his ball and went home.

(That page also links to the announcement email at 
http://sourceforge.net/mailarchive/message.php?msg_id=2877571, which I 
suspect is one of the ones you'd have dug up.)

I would probably add, for purposes of that page: "Since that time, the 
two compilers have diverged significantly," and leave it at that.

>>[1] Yes, it "violates" the GPL license on the runtime library.  As per 
>>discussions on Groklaw and elsewhere, including GPL'ed code in your 
>>software does not "turn it into free software".  The GPL is a 
>>conditional license grant, nothing more.  If you release the software 
>>under non-free licenses, you are simply copying the included code 
>>without permission and thereby violating the copyright on it, and 
>>thereby liable for financial penalties to the copyright holder, as per 
>>normal copyright law.
> 
> That's splitting hairs.  You can't legally redistribute a binary that's linked
> against g95's library if the binary's not Free software (in the GNU sense).
> Now, if I use that binary, and ask for the sources, it's only up to a court of
> law what will happen.   I'd say, I have a right to the sources, otherwise the
> author of the binary would have had no rights to include the library into his
> binaries.  But now I'm splitting hairs :-)

I don't think it's splitting hairs.  This particular point has come up 
on Groklaw (www.groklaw.org; a newsblog partly about the SCO-IBM court 
case, but also about lots of other free-source legal issues) a number of 
times, and one of the key reasons it's important is that the "it will 
make your software into Free Software" perception is why many corporate 
managers are deathly afraid of it.

Consider it from the point of view of the CEO of BigCompany, who's 
thinking about the following:  Suppose that a low-level programmer at 
BigCompany sneaks thirty lines of GCC code into their compiler, and they 
ship the binaries, and someone finds out.

Now, if your view of things holds and their whole compiler becomes Free 
Software as a result, and the court can make them give you the source 
code, they've just lost the company's biggest asset over the 
infringement of maybe $50 worth of code.  This is big, this is scary, 
this leads to things like "As a condition of your employment at 
BigCompany, you are not permitted to touch GPL-licensed code even 
outside of work."

One the other hand, if this is treated as any other copyright violation, 
then the company goes to court, settles some quantity of legal fees with 
the copyright holders that's not too far out of scale of the substance 
of the infringment, removes the infringing code, fires the fellow 
responsible, and goes on with life.  This has happened many times to 
many companies before, and CEOs know how to deal with the risks.

I don't think it's splitting hairs to claim that those are signficantly 
different outcomes.

(An alternate outcome is that the settlement includes a provision that 
BigCompany can keep the code in their product.  In some cases where this 
went to court, companies have decided that they would rather release 
their product under the GPL than remove the GPL-licensed code from it, 
and thus settled with the plaintiff on that basis.  But this was a 
voluntary choice on the part of the company, not something imposed upon 
them by the court.)

The GPL is a license.  Licenses provide rights; they do nothing else, 
and can do nothing else, because they are unilaterally imposed by the 
copyright holder.  If they include restrictions as well as granting 
rights, they are called contracts, and require proof of agreement on 
both sides, which the GPL explicitly does not do.

It is quite clear that the author of the binary would have had no rights 
to include the library into his non-GPL binaries, and was doing things 
he had no right to.  What's at issue is the legal punishment to the 
author for violating your copyright, and I can think of no sane reason 
why the GPL could be held by any reasonable court to cause those 
penalties to be larger than the penalties for violating the copyright of 
non-GPL-licensed code.

So, yes, it's only up to a court of law what will happen -- but that's 
true in any lawsuit, frivolous or not, and I think that in this case one 
can make relatively safe speculations about what the court of law will 
decide.

Also, FWIW, I believe that a case of this nature did in fact actually go 
to trial in Germany recently, so there is at least one point of evidence 
of what a court will decide.

> Even though I might appear hostile, I appreciate getting an outsider's view of
> what this looks like.

Thanks!  I suppose I probably appeared a bit hostile in my initial post, 
as well; in retrospect, I wish to apologise for that.

> ps I find adding 'alleg*' weird -- this is not some rumor that making its
> travels around the internet, but this is _the_ place where these facts (we
> might discuss their interpretation) are stated.  Yes, my perspective is skewed :-)

Feel free to take it out, then.  :)

- Brooks