5:08 PM Trent: would you prefer to chat about it?
Michael: sure, but not for long -- I do have a day job
5:09 PM Trent: yeah, me too :)
I'm not trying to be a smart ass.. you're just not making an argument I can understand.
and I figure others wouldn't be understanding what you're saying either.
Michael: OK, I'm sorry, I'm not trying to be a supercilious asshole, I'm just drawn that way
5:10 PM where should I start? history of copyright? tort vs. contract? "derivative work"?
5:11 PM Trent:
I made the statement that you must abide the conditions of a license to
have a license and you can't copy the work without a license.. is that
5:12 PM or are you saying there's a class of conditions that one can simply ignore and still have a license?
Is it OK if we do this by the Socratic method? You get to drive, but I
mostly answer questions with questions until we get to a leaf node?
5:13 PM Trent: ok sure, go for it.
5:14 PM Michael:
All right. Suppose you publish a work with the GPL attached and you
catch me doing something you don't like with it. Let's specialize to
different flavors of "don't like" later. What is your recourse?
Trent: I tell you to stop.
and if you don't stop I threaten to sue.
and if you don't ceed to my threats, I sue.
5:16 PM Michael:
OK, that's a "cease and desist" letter, possibly accompanied by a
public naming and shaming, followed by a claim in your choice of court
of fact. Let's put German law aside for the moment; it provides special
penalties for rejecting a justified "cease and desist", but you have to
win on the merits first. What's your claim, contract or tort?
5:18 PM Trent:
heh, depends on the act.. if you're violating a condition of the
license, I'll file for copyright infringement, claiming you have no
license to copy the work.. all I need do is show there has been
5:19 PM Michael:
OK, copyright infringement. Let's take US law, I have more precedents
handy; if I knew of some other legal system where the conclusions would
be different, I'd say so, honest. That's Federal District Court. You
claim copyright infringement, I claim license.
Trent: sounds good
5:22 PM Michael:
so now you have to demonstrate that offer of contract contained in the
text of the GPL is either 1) not a valid contract under applicable
state law, 2) sufficiently limited in the scope of the license
that my conduct falls outside it, or 3) attached to conditions of
return performance in which I have failed so materially as to strike at
the heart of the contract and give you grounds to rescind it.
Trent: no I don't.
I just claim you don't have a license.
5:24 PM you
claim you do.. I ask you under what conditions did you aquire a
license.. you show me the text of the GPL, I point out the conditions
which you are failing to meet.. you must then claim you are meeting
those conditions and show you are, or you have no license.
5:25 PM Michael:
Otherwise, you have no claim under the statutory tort of copyright
infringement, and all you can do is to make claims under contract "in
rem". That might get you a clarification of my obligations under the
contract, and I either abide by that clarification or cease to exercise
license henceforth. That, in turn, might lead to a later cause of
action for breach of contract "in personam". That's a wholly
unprofitable branch of argument, but we can go there if you want.
5:27 PM The burden of proof that I am not
meeting those conditions is on you. That's how copyright infringement
claims (or any other tort claims) work in the presence of a
presumptively valid offer and acceptance of contract.
5:28 PM Trent: yeah yeah.. it's easy for me to show that you are failing to meet the conditions.
that's why we're in court.
so yeah, again, here I am asking, do you believe that you are not
required to accept the conditions of a license before you can have one.
5:30 PM Michael:
The circuit court spelled out where the district court had erred: "We
agree with Sun that significant evidence supports the district court's
holding that Sun is likely to prevail on its interpretation of the
language of the agreement and to prove that Microsoft's conduct
violated it. We agree with Microsoft, however, that the district court
should not have invoked the presumption of irreparable harm applicable
to copyright infringement claims before it determined that the
compatibility requirements were a limit on the scope of the license
rather than independent contractual covenants. We therefore vacate the
preliminary injunction and remand for further proceedings."
5:34 PM On
remand, the district court duly evaluated the scope of license and
decided that Sun could not prevail on a claim of copyright infringement
until the contract issues had been fully litigated. The court of fact
granted a different injunction on a theory of unfair competition under
California state law
and Microsoft later settled.
5:36 PM I
claim to have accepted and performed the conditions of return
performance contained in the license. Specifically, I have in fact
provided every person who has received (directly from me) a modified
and compiled copy of your kernel with the source code I built it from.
5:38 PM That
code has been run through s/_GPL//, compiled with a compiler you don't
have for a target whose object code is incomprehensible to you, and
combined on the same flash chip with binary modules that you can't
extract. But none of that is my problem.
5:39 PM Trent:
I think the key part of that citation is this: Micrsoft contends that
the disputed compatibility requirements of the license agreement are
affirmative covenants rather than limitations on the scope of the
license, and that accordingly contractual rather than copyright
remedies are appropriate if there has been any breach.
Trent: yeah.. one couldn't claim that same thing about the GPL.
5:40 PM the whole "you can't distribute this if you have linked proprietary stuff to it" I claim is a condition, not a covenant
you claim it is a convenant?
5:42 PM "you can't distribute this if you have linked proprietary stuff to it" doesn't appear in the text of the GPL.
Trent: I'm paraphrasing
5:43 PM Michael:
And wouldn't matter if it did -- that's an invalid condition (in a
contract of adhesion) under a slew of appellate decisions summed up in
Lexmark and endorsed by the Supremes when they denied cert.
5:46 PM Trent: ok, a citation, has to actually have the relevant part cited.. otherwise it's a reference, not a citation.
What the GPL actually says is that the obligation to disclose source
code on request applies to "works based on the Program", which is
explicitly defined in Section 0 in terms of the phrase "derivative
work", a term of art in the 1976 Copyright Act.
Trent: what are you citing from these references which supports your argument.
5:47 PM ok,
the GPL also defines "complete source code" which is the actual wording
that is used in the section which defines your obligations.
Michael: Just a minute, let me switch desks to where I have more dots and a chair to sit in.
5:49 PM "complete corresponding machine-readable source code"
5:50 PM and it says "work based on it, under Section 2" which says nothing about a derivative work.
5:51 PM whenever
it does say derivative work it defines what is ment by that.. I don't
think you could argue that the meaning of that wording wasn't clear.
6:00 PM Michael: sorry, I got sidetracked on the way to the new desk -- response from Russell King on my serial bug
6:01 PM "derivative work" is paraphrased (incorrectly) but hardly defined
back in a moment
6:20 PM Michael: all right, dealt with that; thanks for your patience.
6:21 PM OK,
do you want to drill down into the case law, or discuss "work based on
the Program" -- or should we perhaps summarize the extent of our
agreement so far?
6:25 PM Trent: umm.. hmm
agreement sounds like an idea.
6:27 PM I
claim the meaning of derivative work in copyright law is irrelevant..
because the GPL defines what is ment by a "work based on it, under
6:29 PM Michael:
Agreement first. Let's alternate. I think we agree that copying,
modifying, and distributing someone else's work of authorship is going
to get you in trouble if you don't have a good defense, that the
statutory defenses of "fair use" and the like are of no avail when the
material copied is whole non-trivial computer programs, and that D-Link
and Fortinet were conniving greedheads who deserved to lose in court.
6:30 PM Trent:
if you were to show to me that "a work based on it under section 2" in
fact requires no license, then I'd agree that section 3 is
ineffective.. but I think we both agree that "a work based on it under
section 2" does require a license.
s/that "/that distributing "/
6:35 PM Michael:
Let me rephrase that into agreement. If I am defending against a claim
of copyright infringement by claiming license under a valid contract of
adhesion, we agree that the court needs to determine whether my conduct
falls within the scope of license in that contract in order to render a
judgment. If not, this
defense fails and the copyright infringement claim can proceed.
Trent: that sounds right.
6:38 PM I
think that would be an easy win for the case of distributing a
proprietary extension to a GPL program, as section 2 says that
distributing as part of a whole work based on the program the
distribution of the whole must be on the terms of the GPL.
would you agree with that?
6:39 PM Michael:
So if I am claiming license under GPL terms, I have to demonstrate that
what I copied and distributed is either a "verbatim cop[y] of the
Program's source code" (Section 1), or a "modif[ied] copy ... of the
Program or any portion of it ... forming a work based on the Program"
(Section 2), possibly accompanied by a binary blob sufficiently closely
related to that source text (Section 3). Right?
6:41 PM Trent: yeah, complete and corresponding.. or an offer for the same, yadda yadda.
6:45 PM Michael:
But if I am also distributing a paperback copy of War and Peace in the
same cardboard box, I'm not claiming license under the GPL to do so,
and I don't need any namby-pamby "mere aggregation" clause to do it
either. The same applies if I put a digitized copy of War and Peace on
the same CD, or in the same flash chip, because common sense and (more
importantly) the prevailing industry practice with regard to licensed
programs of all kinds is that their boundaries are the limits of a file
in the filesystem or, in some cases, a functionally inseparable
collection of files (which may be sprinkled throughout the filesystem).
6:47 PM Trent:
yeah, ok. I can imagine a situation where a program wouldn't run unless
some big word file was present.. in which case the word file would be
considered part of the program (as it is required to run the program)..
but in general, something like that wouldn't affect the common case.
6:51 PM hehe..
just as a sidenote.. I have a completely different reasoning for why
distributing the kernel (or any software with a plugin system) with a
module that you didn't get it with is a copyright violation. Remind me
later if I still think it is relevant.
6:53 PM Michael:
Funny you should mention that. Enter Lexmark. Lexmark wrote a Toner
Loading Program (with minimal creative content if any, but that's
irrelevant for reasons discussed in the opinion) and stuck it inside
every printer cartridge they sell. It's arguably copyrightable
expression in its source code form, and it or its functional equivalent
is necessary to the proper operation of the cartridge. But they got
greedy and sha1summed it from the printer side every time you open the
printer casing, in order to create a cause of action under copyright
law and the DMCA and all other sorts of nastiness if people like Static
Control started selling interoperable printer cartridges. Don't copy
their TLP byte-for-byte, and the printer will spit your cartridge out
like so much tobacco; do copy it, and they'll see you in Federal Court.
SCC did, and Lexmark did, and now we have the second strongest kind of
law there is -- an appellate decision overturning a court of fact's
ruling _as_a_matter_of_law_, appealed to the Supremes and denied
6:57 PM (The
strongest kind of law there is is a published opinion of a majority of
the Supreme Court. The third strongest kind is a unanimous opinion of a
panel of a Circuit Court, approved by a majority of the full Circuit,
affirmed by a divided Supreme Court, and widely cited by sister
Circuits for a decade thereafter -- e. g., Lotus v. Borland. Mere Acts
of Congress aren't even close.)
6:58 PM Trent: so what did they say?
7:00 PM Michael:
The Lexmark court held that the use of the TLP as a "lock-out code",
within the sense established in the 9th Circuit's Sega v. Accolade
opinion (not previously binding on the Lexmark panel, which is 6th
Circuit), rendered it uncopyrightable as_a_matter_of_law and therefore fair game for copying without any license whatsoever, for the purpose of interoperating with Lexmark's printers.
7:01 PM MODULE_LICENSE("GPL"), anyone?
Trent: oh right, yeah, that stuff, whatever.
I think the whole MODULE_LICENSE crap is just pointless.
just a little irritant.
7:02 PM Michael:
Worse than pointless. If taken to the logical extreme of cryptographic
signatures covering one blob and another, it could render the copyright
on the entire kernel null and void.
7:03 PM This whole "uncopyrightable as a matter of law" business is very dangerous.
7:04 PM An appellate court cannot overturn a lower court on a point of fact without demonstrating "clear error", a far higher hurdle than the "abuse of discretion" involved in interpreting the law incorrectly.
7:05 PM Trent:
hmm ok.. it's a nice law to know.. and I'll try to remember it next
time I'm reverse engineering stuff.. but getting back to the GPL.
As a result, we (at least in the US) are faced with a prima facie
bizarre doctrine in which you don't just lose the ability to recover
damages for copyright infringement in a particular case, you
7:06 PM Trent: tis not the first time that this has occured.
Michael: That's what happens to people who try to put one over on the legal system by perverting copyright into patent or trademark.
7:07 PM Trent:
in general, the doctrine is that trying to use copyright to enforce
stuff that is largely not protected by copyright is means to lose your
of course, that applies to the kernel as much as it does to anyone
else.. so be sure to have that advice handy in case anyone sues someone
for inserting a kernel module.
7:09 PM Now,
back to a sane fact pattern. I've written a binary module that calls to
and from entrypoints that are unique to the Linux kernel, as indicated
by the EXPORT_SYMBOL_GPL (let's call it EXPORT_SYMBOL_UNIQUE_TO_LINUX
7:10 PM Trent: ok
7:12 PM Michael:
Stipulate that the court agrees with me that, if my engineering purpose
was to interoperate sanely with the kernel, I had no practical choice
other than to claim MODULE_LICENSE("GPL") with a big ol' disclaimer
explaining why in my documentation. Reimplementing sysfs from scratch
is dunderheaded and telling me to s/_GPL// and ship my own kernel is
7:14 PM Now,
what makes me think I get to do what I need to in order to
interoperate? Lexmark, Lotus v. Borland, and Baker v. Selden, all the
way back to the era of tear-out forms in the back of accounting
Trent: except that you don't need to do that.
7:15 PM you need to export that code, yes.. sorry.
7:17 PM personally,
I don't see what legal right anyone would have to sue you for doing
that anyway.. or what this has to do with the licenses you are allowed
to use on your code.. personally I think this whole
MODULE_LICENSE("GPL") thing is a kernel specific side issue.
Michael: The heck I don't. Borland could
have told all of Lotus's customers that they had to rewrite their
macros from scratch in order for Quattro Pro to be able to read their
1-2-3 spreadsheets. But the appellate court held that customers'
interest in reusing their efforts outweighed Lotus's interest in
keeping them locked into 1-2-3. The transcript of oral argument before
the Supreme Court is a rare example of legal analysis in the raw, and
makes truly fascinating reading to this outsider.
7:18 PM Sorry, that crossed with your comments
7:19 PM Trent: yep, no worries, it's an interesting point anyway
7:22 PM Michael:
OK, so we agree that I can modify the kernel source code and ship that
modified source code together with its compiled form, and that I can
write a new kernel module using the minimum practical amount of copied kernel code and ship that kernel module under my own licensing terms. Can I do both on the same CD?
7:23 PM Trent: ..
7:24 PM I don't think you can ship your kernel module under your own licensing terms.
with or without the kernel in toe.
because the module is useless without the kernel.. it is part of the kernel.
7:25 PM it can't reasonably be considered a separate work.
7:26 PM Michael: ... sorry .... child moment ... back now
7:27 PM OK, let's talk about the copyright law boundaries of a "work of authorship". OK?
(putting aside "work based on the Program" for the moment)
7:28 PM Suppose I write you a letter in Sindarin Elvish. Without a copy of the Silmarillion, you couldn't read it to save your life. OK?
Trent: I don't think that is analogist.
7:29 PM Michael: It isn't. But it's relevant.
7:30 PM Trent: in answer to your earlier question, can I do both on the same CD? I'd say no.
My letter is on the borderline of being a derivative work of Tolkien's
opus, because it contains creative expression that could not be
conceived without Tolkien in the background.
Trent: because the module + the kernel is a work based on the kernel.
7:31 PM Michael:
that "based on" phrase has no legal meaning outside the GPL. Can we
come back to what it means in the context of the GPL after we establish
what a "work of authorship" is?
Trent: one sec.. I will listen to you.. it's just that I'll have this on my mind and won't give you the full attention you deserve.
7:32 PM Michael: and, more to the point, what a "derivative work" is?
That's OK, let's go your direction first.
7:33 PM Trent:
the derivative work definition in copyright law is irrelevant. You
don't have permission to distribute the kernel in binary form unless
you provide complete corresponding source code, as defined by the GPL.
OK, there's another point on which we agree. You don't have permission
to distribute the kernel in binary form unless you provide complete
corresponding source code, as defined by the GPL.
7:34 PM Trent:
if you wanted to argue that you could distribute the kernel in source
form with a proprietary module, I'd say that would actually be ok..
which is the fault of the linux developers, because they made it able
to load proprietary modules without modification.
7:35 PM Michael: The "Linux kernel" is a work of authorship with a well-defined boundary, namely, the contents of a kernel.org release tarball.
Trent: this is why gcc doesn't have a module system.
7:36 PM the
authorship of the kernel that you are distributing is the authorship of
whichever parts you are distributing.. would you agree?
7:39 PM Michael:
Actually, GCC does have a module system. There is, or used to be, a
perfectly reasonable engineering interface between GCC front ends and
back ends. NeXT wrote a C++ / Objective C front end for it. They
eventually forked over the copyright to the FSF as a result of some
major strong-arm tactics by Eben Moglen personally, in which (as I
understand it, I wasn't there myself) he threatened to rouse the free
software rabble (including large numbers of students at the
universities where NeXT had its only profitable market niche) and muck
about with the interfaces
7:40 PM until NeXT collapsed under the instability.
7:41 PM Trent:
not to get too sidetracked here, but do you think he did that for
monetary gain? Or do you think he did that because he's a idealist on a
Notice that he did not darken the doors of a courtroom, in which he
would have LOST. Notice also that this is utterly, diametrically
opposed to his affidavit in Lotus v. Borland.
7:43 PM Allow
me to recommend a perusal of the IRS Form 990s for the FSF, the League
for Programming Freedom, and the Software Freedom Law Center. And the
SEC records relating to the acquisition of Cygnus Solutions by Red Hat.
The GPL has made Eben Moglen a very rich man.
It is also the sole foundation of his professional career since clerking for Thurgood Marshall.
7:44 PM Trent: yep, so which is it? he's in it for personal gain or he's on a crusade? Can he be both?
7:45 PM maybe he's doing his best to get together a war chest..
It "earns" him all his speaker fees. It "earns" him all his outside
retainers from the likes of Fluendo and Vidomi. It sustains his
teaching career at Columbia University. It underlies his protection
racket -- there can be no other term -- at the Software Freedom
7:46 PM And all of this money depends on maintaining the fiction that the GPL is not what it plainly is -- a contract of adhesion.
Trent: cause, ya know, at some point someone cough Microsoft cough is going to test the GPL.. he's gunna need a heck of a lot of money to fight it.
Michael: Bull. Fucking. Crap.
7:47 PM Trent: don't worry, I'm not going to defend Eben Moglen much.. he's a lawyer.. all lawyers are scum. period.
7:48 PM Michael:
The GPL is the best thing that ever happened to Microsoft. It is the
most beautiful smokescreen that Bill and Steve could ever have
conceived for consolidating an oligopoly in toolchains and libraries --
far, far more effective than an O/S or application monopoly.
7:49 PM Trent: see, this really is your downfall.. you're big on the conspiracy.
think RMS is in on it too? or is he the dupe?
7:51 PM Michael:
Where are the RISCs of yesteryear? Moribund, or shrunk into embedded
systems at negative gross margin! Where are the C++ compilers from
AT&T and Sun and Borland and CodeWarrior that went toe-to-toe with
VC++? Abandoned and rotting in the hulks of the corporations that once
spent millions of dollars every year on their maintenance and
7:53 PM Who
and what did this to them? Free-as-in-beer! Why write interoperable
code when GCC is free, ubiquitous, and tuned to within an inch of its
life for x86 targets?
Trent: You do know that Borland and CodeWarrior still write compilers.
and that GCC has the worst performance of any x86 compiler.
7:54 PM (I used to work for Codeplay.. we had a C++ compiler that kicked GCC (and the Intel compiler)'s ass on every platform.. )
7:56 PM Michael:
I know lots of Borland people, technical and business. Lots of Sun
people, too. Microsoft fought them to a standstill but GCC-based IDEs
and Eclipse are what
GCC doesn't have to be good. It just has to be free, and better on x86 than it is elsewhere.
I know the guy that owns that code now.. he's sitting on it till he
sees an opportunity to make a obsfucating compiler for DRM shit. yawn.
7:57 PM Michael:
All you need is mindshare among CS undergrads, especially those in
developing countries. So you give away the student edition and let 'em
learn your dialect. The rest follows.
7:58 PM CodeWarrior is walking dead, courtesy of XCode. Even Adobe is switching over.
7:59 PM Trent: yeah, XCode is the shit if you're writing objective-c.. I hear.. I don't write that crap.
Michael: Me neither.
Trent: if Apple weren't total leaches, they'd open that.
Michael: Dude, it's GCC!
8:00 PM Trent: it's an IDE that calls GCC to compile, yes.
Michael: As is Tornado.
Trent: it's really the GUI designer that is the value of XCode.
Michael: Now can we go back to what a "work of authorship" is for a minute, and why it's bogus to fetishize the fork/exec boundary?
8:01 PM XCode's GCC can't even compile _itself_. That's the magic of the GPL.
Trent: from the GPL perspective, the IDE and GCC can reasonably be considered separate works.
8:02 PM Michael:
XCode's GCC is a Canadian cross. It's unreproducible without the
compiler that compiled it, which is safely locked away inside Apple's
(or possibly CodeSourcery's) walls.
8:03 PM That's
a trick that Cygnus refined for a decade, and sold out to Red Hat for
$674 million in stock, without even having ownership of the source code!
8:05 PM Trent: idk what cygnus sold for.. their customer list most likely.
The money is in cross-compilers. Repeat after me. The money is in
cross-compilers. All commercial compilers are cross-compilers. No
commercial compiler is self-hosting. The GPL does not require
distribution of the toolchain that was used to compile the executable form of a program and THAT IS NO ACCIDENT.
Trent: hmm.. well actually it does
8:06 PM Michael:
nope. "For an executable work, complete source code means all the
source code for all modules it contains, plus any associated interface
definition files, plus the scripts used to control compilation and
installation of the executable."
8:07 PM Trent: compilation, yeah.
but that's a nice good point there. I see what you are saying
8:08 PM Michael:
The "scripts" are not the toolchain. And that, incidentally, is the
other half of the poison pill in the LGPL (apart from the bullshit
about linking creating derivatives);
an executable, the required form of the "work that uses the Library"
must include any data and utility programs needed for reproducing the
executable from it.
8:09 PM Why
do you think the LGPL has a GPL "upgrade" clause in it? So that code
contributed to LGPL projects can be migrated into a GPL project without
the GPL project becoming subject to the "data and utility programs"
8:10 PM Trent: your desire to attribute these things to malice on the part of the drafter is not justifiable.
8:11 PM Michael: When's the last time you tried to recompile a Red Hat GCC from its ostensible source?
Trent: I don't use redhat.
8:12 PM Michael: Me neither. I use Debian and Ubuntu. Their shit may be shit but it's reproducible and point fixable. Unlike any distro that Oracle runs on. More magic.
8:13 PM To understand the big payoff, you have to look at the back cover of an Economist magazine.
8:15 PM You will find one and only one software advertiser there. Oracle.
8:16 PM The
world runs on industrial-strength databases. Ten years ago, there were
four. Today, there are two. Sybase and Informix are in the dustbin, and
the only competition left standing is IBM DB2.
8:17 PM MySQL is training wheels for Oracle. MS SQL Server is a back end for Exchange.
8:18 PM Oracle and DB2 are the back ends for Oracle, SAP, and every financial application in every bank in the world.
8:23 PM Oracle
makes and breaks server architectures and operating systems. Oracle
made Tandem, and when they didn't need them any more, they broke them.
Oracle made Fujitsu, and keeps them on life support as a threat to IBM.
When the workstation market dried up, Oracle made Sun, and when Linux
became usable as a server O/S, Oracle broke Sun and made Red Hat and
SuSE. Oracle is toying with breaking SuSE over the Novell/M$ deal and
making Ubuntu instead. Oracle's Unbreakable Linux is just there to put
the fear of God (spelled Larry Ellison) into Red Hat; but if Red Hat
doesn't toe the line, Red Hat will be broken just as easily as the rest.
8:29 PM Michael:
Oracle almost certainly funded the SFLC, laundering it through OSDL and
swearing them to secrecy. Oracle broke the OSDL and tossed out Stuart
Cohen, reforming it in their own image as the Linux Foundation. Oracle
is satisfied with its duopoly with IBM, and intends to keep it; Larry
has turned his attention to SAP, and that's what today's Economist
back-cover ads are about. The only entities that could possibly
undermine Oracle's control over the market for industrial-strength
databases are Microsoft and Google, and Google has other fish to fry.
8:31 PM Is
this boring you? Do you want to go back to the meaning of "derivative
work under copyright law" and the subtext of the peculiar style in
which the GPL is drafted?
8:44 PM Trent: sorry, had cleaning people here.
8:47 PM so, ya don't think open source databases will just crush oracle one day?
8:50 PM more importantly, do you think mysql is so shit because of than oracle conspiracy or just because they are ass programmers?
8:59 PM Michael: s'ok. I'm back.
9:00 PM MySQL
isn't that bad. If you treat it like the racing thoroughbred it wants
to be. Which means, do the exact same thing to it every day and never
so much as look at it funny.
9:01 PM You cannot recompile a database with a new compiler and expect it to work the same. The code is just too damn fiddly.
9:02 PM The
same goes for 3-D graphics drivers; four out of five times you
recompile them the optimizer rearranges things enough that it slags the
9:04 PM That
is why ATI and nVidia do what they do. Not because a Third World cloner
has a snowball's chance in hell of reverse-engineering their GPU from
the driver source, but because they couldn't begin to deal with the
support requests from users of "overclocked" drivers.
9:06 PM Just to take a well documented example; read that guy at Stanford's explanation of exactly what you have to do to BSD DB4.x to make it survive under the kind of real load that a campus-scale slapd puts it under.
Trent: excuses excuses.. everyone has one for nvidia.
Michael: Not an excuse. An explanation. One that fits the facts.
9:07 PM I don't like it. If I want to slag my GPU that's my look-out. But Fry's would never carry an nVidia card again.
9:08 PM Trent: tis crap reasoning too.. I know dudes who work on the nouvea project.. they don't slag their GPU by recompiling.
9:09 PM Michael:
I hate to piss in your Wheaties, but the economics of retail
distribution are utterly destroyed by a small percentage of returns,
and the customers for 3-D graphics boards are born overclockers.
9:10 PM Trent: oh right, yes, overclocking would trash their margins, for sure
I'll pay that.
9:11 PM Michael:
You cannot protect a piece of hardware that is being clocked internally
to within a hair from its thermal breakdown point from all the idiotic
things a million code-bummers might attempt to squeeze out another
tenth of a frame per second.
9:12 PM Trent: yep.
9:13 PM GPUs need a tainted flag ;)
Commercial and industrial applications for 3-D graphics boards are
certified against specific driver binaries. They have to be, or they
crash left and right.
9:17 PM I
think you read too much "conspiracy" into my explanations for things.
MySQL is quite open about being a mid-range RDBMS, as is SQL Server.
9:18 PM Architecturally and SQL-extension-wise, MySQL is training wheels for Oracle and SQL Server for DB2. Ask a senior DBA.
9:19 PM Trent: I don't debate that... I read conspiracy into your statement that Oracle is funding the SFLC.
9:20 PM for anything other than the normal justifiable reason of not wanting their favourite platform to go away
9:23 PM Michael:
I don't know for certain. I do know that OSDL non-members contributed a
significant fraction of that seed money, and that Oracle is the only
company that matters (except perhaps Google, which goes its own way)
that wasn't an OSDL member at the time. http://www.eweek.com/article2/0,1895,1758393,00.asp
9:25 PM Trent: yeah. the intention is the problem dude, not the money trail
9:26 PM Michael: The money trail is illustrative of the racketeer's modus operandi.
9:27 PM Go look at those Form 990s. (You have to register with guidestar.org to get them but it's free-as-in-beer.) Moglen makes $150K here and $150K there, and pretty soon it adds up to real money.
Trent: or that they just don't want their customer's prefered platform to go away.
he's a lawyer, he makes money.. that's what lawyers do.
9:28 PM Michael:
Their customers could give a shit what "platform" they're on. They want
to know how many TPC/C they get per million dollars a month, and they
want to know how many 9's of uptime they get.
9:32 PM Lawyers
-- good ones -- do not lie in public day in and day out for twenty
years, on a topic on which they have been entrusted with the dignity of
a professorship at a prestigious private university, in order to siphon
cash from a web of "non-profits" whose sole line of business is to
shake down hardware makers periodically for the privilege of having
their platforms continue to be a supported target for the only compiler
that matters outside of the Wintel-verse.
9:34 PM Trent:
if he lies that much, it should be real easy to find a whole lot of
instances where he has lied and point them out.. why don't you make a
An attorney is a sworn officer of the law and can be put in jail for
life for that sort of thing, if someone who cares enough to fund a RICO
prosecution to the bitter end could be found.
9:39 PM Tell
me something. What if I'm right and there is no such thing as a
"copyright-based license". What if Eben Moglen -- professor of law and
legal history -- is perfectly aware of this fact, and lies every time
he opens his mouth to talk about the legal foundations of "copyleft".
What if Mr. "Anarchism Triumphant" is a garden-variety shyster with a
nice pyramid scheme that brings in a cool million a year, on top of
grants of founders' stock in startup after startup riding on the "IP
law is dead, cloners rule" bandwagon. Would you spend your free time
playing frisbee with your kids and learning to fix motorcycles instead
of fixing bugs in the GNU shitpile?
9:40 PM Trent: even if you're right, he's still fighting for my beliefs for me.. I'm happy.
Michael: What beliefs would those be?
9:41 PM Trent: and he can't step out of line.. otherwise his empire will crumble.
that copyright is fundamentally wrong (at least for software)
9:42 PM Michael: Would you care to examine those beliefs under a bit of gentle Socratic teeth-pulling?
9:43 PM (Keeping in mind that I mostly agree)
(but the "mostly" hides a big difference or three)
Trent: ya know what ya should do.. and I know this may be hard depending on your skills.. you should write a novel.
9:44 PM Michael: Dialogue is not my strong suit. :-)
Trent: nice big disclaimers like "this is a work of fiction, any similarity to existing people is fictional, etc"
Michael: I'd like to have Upton Sinclair'
9:45 PM Trent: which, of course, everyone who reads it will ignore.
Michael: But I'm afraid I'm too lazy to make a proper muckraking journalist.
Trent: give Eben Moglen a sensible name :)
9:46 PM Michael:
And I have a dearly beloved wife and child that I do not want to put
into Eben Capone's sights to any greater degree than I already have.
Trent: pseudonym it then.
9:47 PM you can use my name, just cut me some of the royalties.
when Moglen tracks me down I'll just say I made the whole thing up for a laugh and didn't he read the disclaimer?
Michael: If I ever use this dialogue I'll cut you a square deal. Deal?
9:48 PM Trent: hehe
9:49 PM Michael:
But seriously, I'd love to pull a few teeth over the appropriateness of
copyright for software, thesis-antithesis-synthesis style.
9:50 PM Trent: ok, just don't expect too much concentration on my part.
I'm coding a ptrace based user level chroot/unionfs util.
9:51 PM Michael: OK. For starters, what if the GPL (construed as a contract of adhesion) were universally applied to software.
(cool, ptrace is nifty, I learned a bit about it fixing ltrace to work on ARM again)
Trent: ok, half of the GPL would be pointless then.
9:52 PM Michael: Is there any software to which you would not want to see the GPL applied?
Trent: cause all the talk about other licenses would be useless.
Michael: Which it is already. But let's not go down that road.
Trent: well, there's plenty of software which is private.. and typically has no license applied.
9:53 PM so
let's instead, say that for any software which currently has or would
have a proprietary software license applied, we apply the GPL instead.
9:55 PM Michael:
What about software with a storyline and characters? Interactive
fiction? Video games? If you apply the GPL not just to the game engine
but to the "creative content" in a Hollywood sense, you'll never see
another high-budget film with an interactive sequel, because copyright
in the "mise en scene" goes bye-bye. Is that what you want?
9:57 PM Trent: never see a high budget film again.. say you aint kiding.. please.
9:58 PM Michael: no, but you'll never see a James Bond interactive trailer or a Midway X-Men III.
10:00 PM Trent: personally, I don't see how the argument that the GPL would apply to the content can apply.
10:01 PM and in that way, GPL on games sucks
You will also find the Tolkien estate prosecuting small-time makers of
Nethack clones with LotR characters in them, because they become a
vehicle for their characters and mise en scene to leak into the
commons. Hardly fair, right?
Trent: it's not as valuable as GPL on utils.
10:02 PM Michael:
That would be a poor social compromise. So let's leave works with
storyline and characters out of the presumptive GPL legislation.
10:04 PM Trent: kinda begs for the entire abolishment of copyright.. might be too extreme for you to consider debating.
10:05 PM Michael:
Now let's go to a different extreme. How about firmware? Do you want to
destroy the market for FPGAs, which have made it infinitely more
practical for skilled software developers to migrate into hardware
design, and sped up the prototyping and even small-scale manufacturing
of peripherals and embedded devices at least tenfold? If not, you
probably don't want to legislate the GPL for firmware.
There's a case for abolition of copyright. It's called China. Do you want to live there? I don't.
10:06 PM Trent: how do you know I don't?
10:07 PM if it were possible to effectively abolish copyright tomorrow, I would.
Michael: I would be very surprised if you were of mainland Chinese origin and living in mainland China.
10:08 PM Would
you now? Personally, I'd like to know how J. K. Rowling is going to get
herself out of the plot corner she has painted Snape into. Abolish
copyright, and you'll never know.
Trent: I'll get over it.
10:09 PM Michael: She's literally richer than the Queen. It's not about her. It's about me. I like books. I would miss books.
10:11 PM I
would also miss the Economist. And remastered archival recordings of
the pioneers of jazz. And going to the movies. Because all of those
things would cease to exist without copyright.
10:12 PM Trent: there'd still be books.. you don't honestly buy the argument that works only exist because of the "incentives" of copyright.
there'd be more movie cinemas than you could poke a stick at.
lot more than today.
and they wouldn't charge $45 for a popcorn.
10:13 PM Michael:
First-run cinemas would close their doors the same day. And on the rare
occasions when I eat movie popcorn, I don't really care how much I paid
for it. A good babysitter costs us $60 a night anyway.
Gonna cost more when there are three.
10:15 PM Works
would exist. Writers mostly write because they'd lose their minds or
die of drink if they didn't. But editors edit and publishers publish
because they can make a decent living at it. I would never see those
new works. That would make me sad.
10:19 PM Trent: editors and publishers would just switch to a subscription model.
10:21 PM Michael:
There will never be another Thelonious Monk or John Coltrane. But
thanks to copyright, it was worth someone's while to preserve for fifty
years the only reel of acetate containing the only known recording of
the two of them playing together, and worth someone else's while to
gently extract the tiny magnetic wiggles and run it through a chain of
signal reconstruction techniques that were literally inconceivable when
it was recorded, which would not themselves have been developed without
copyright. Then it was worth someone else's while to package this
reconstructed signal on a shiny disc with informative liner notes and
ship it to me for less than half of what I pay for a haircut. That's
10:25 PM Trent: and all you had to do was give up your right to make copies for your friends.
10:26 PM not that anyone does that.
and yet copyright continues.
10:27 PM the
illusion is that copyright makes book publishing profitable.. but it
isn't. we prefer the dead tree to the pdf from #bookwarez. That's what
keeps book publishing profitable.
10:30 PM Michael:
For 99 cents I downloaded a recording of Richard Strauss's
Metamorphosen, a work for 23 string soloists that I will probably never
have the privilege of hearing live. 99 cents! It takes me half an hour
of intense concentration to listen to it, and it's another hour before
I recover the quotidien indifference needed to piss away my one life on
Earth billing $84 an hour. Copyright costs me NOTHING. Copyright is the
greatest blessing in my life and yours. With regard to creative works.
10:32 PM Trent: there's costs.. you've just learnt to ignore them.
10:35 PM Michael:
I have wonderful friends. But I can't scrape up 23 virtuoso string
soloists among them. And without 23 virtuoso string soloists and a
great concert hall emptied of its audience, there can be no recording
like this. That costs bucks, pal, more bucks than I make in a decade.
If you've got room to fit ten years of my earnings into 99 cents, I'd
like to hear your theory.
10:36 PM Trent: the work already exists.
10:41 PM Michael:
But I would never in a lifetime have found it without a publisher's
efforts. And the great recording of Henk Bading's String Quartet No. 4
by the Stadler Quartett, a piece so unique (having been composed in a
31-tone scale and perhaps performed publicly no more than once in the
composer's lifetime) that it may never be performed again, will never
become available to me without the economic incentive of copyright.
10:42 PM Trent: yeah, big call, I'd love to see some way to show that.
10:43 PM Michael:
So even if I were to give you, undebated, the premise that no other
great work will ever be created, and that I can be happy for a lifetime
on the stale dregs of a dead culture, the argument against copyright
10:45 PM Trent: right.. because copyright is an aid to culture, not a hinderance.
10:46 PM you said china has no copyright.... they have no culture?
they're living on the dregs of a dead culture are they?
10:47 PM or they're living off the rest of us lucky people with copyright.
10:49 PM Michael: This CD was published in 1999: http://www.xs4all.nl/~huygensf/english/cd.html. Perhaps the next one will contain the Stadler Quartett recording excerpted here: http://www.xs4all.nl/~huygensf/english/music.html#badings.
Have you ever mastered a CD? I have. It was done as a labor of love for
a friend, who wrote and performed a gospel mass and needed a demo CD in
order to submit it for consideration to the publishers of a gospel
hymnal. A thousand years of royalties at the pittance she is paid would
not recompense me for the time I put into it. But it may be the only
composition she publishes in a lifetime, and that matters to her more
than I can explain to you.
10:50 PM Trent: and, as such, copyright fails yet again?
10:51 PM Michael:
The Chinese culture is alive and well. But it's a patronage system,
like Europe before copyright. I don't want to live in a patronage
Trent: why not?
10:54 PM Michael:
In a patronage system, Kate's mass would still have been composed and
performed. I would still have had the pleasure of singing it privately
under the sponsorship of the church I attend. But no one outside Santa
Cruz would ever hear it, or have the opportunity to sing it as part of
their own church service, or for that matter to record it in Carnegie
10:55 PM The
Byzantine system of copyright clearance pays Kate a pittance and me
nothing. But it keeps her work alive. That's what's at stake.
Trent: I don't understand.
10:59 PM Michael: Read the history of Allegri's Miserere at http://en.wikipedia.org/wiki/Miserere_(Allegri).
Absolutely exquisite piece. We perform it every year that we can scrape
up a soprano soloist who can float that top C into the stratosphere,
error in Grove's or not. That piece would never have been heard outside
the Sistine Chapel without copyright.
11:01 PM Trent: again, I don't know how you can make that claim.
11:02 PM Michael:
And without copyright, I would not have the scholarly pleasure of
consulting the authentic Vatican edition and marvelling at the strange
ways that exquisite art comes about.
11:04 PM What
do you think motivated Mozart to transcribe the Allegri Miserere from
memory, at risk of excommunication (which would have utterly destroyed
his employability in the
eighteenth-century Holy Roman Empire)?
11:06 PM Trent: idk, I'd like to think it was the belief that people should hear it.. not profit.
Michael: Care to guess again?
11:07 PM Trent: and if it was profit, well shit, way to be a cretin.
11:10 PM Michael:
Who cares whether Mozart was a cretin? He received whatever he received
from Charles Burney -- and I bet it clinked all the way into Leopold's
pocket -- and Burney got the manuscript. Burney got his fellowship in
the Royal Society, and I got the _Miserere_.
11:12 PM No
copyright, no publishers. No publishers, no insistent, conniving,
irresistible vacuum pulling works out of their author's hands (and
sometimes even their heads) and into my living room. It really is that
11:14 PM Trent: except it isn't.
Michael: OK, your turn. Gimme a f'rinstance.
11:15 PM Trent: publishers would exist without copyright.. in fact there would be more of them.
Michael: Oh, everybody's a publisher now. But most of us suck at it.
Trent: they don't need the control that copyright provides.
11:16 PM Michael: Rubbish. Tell me, have you ever paid more than $1000 for a work of art?
Trent: not personally.
11:17 PM Michael: I have. And it's mine. And it's the only one of its kind there's going to be, unless I decide otherwise.
11:18 PM Trent: you don't need copyright for that... you control it.
you own the copy, it's property.
11:19 PM Michael:
The artist is locally famous. At least three of his murals grace entire
walls downtown. His paintings hang in my favorite movie theatre and in
a couple of other shops around town. But mine's better than any of them.
Trent: I do happen to have two pieces of art that I own.. which cost over $1000 each.. I just didn't pay for them myself.
they were gifts from friends.
11:20 PM Michael: Good for you; you have the right kind of friends.
Trent: none of this is relevant to copyright.
11:21 PM Michael:
Sure it is. Because I might send you a snapshot of this painting, and
you might decide you agree that it's as good as I say it is, and ought
to be in the coffee-table book of his life's work you publish ten years
But that snapshot isn't going to cut it if you want to do his work justice.
11:22 PM So somebody has to shell out for a good camera. Who's gonna pay for that?
11:23 PM And somebody has to shell out for a color calibrated prepress seat. Who's gonna pay for that?
you keep reiterating the argument that copyright pays for stuff.. and I
keep reiterating the argument that copyright isn't necessary.
11:24 PM Michael: And somebody has to shell out for a fine art offset press and perfect binding. Who's gonna pay for that?
11:25 PM Hint:
That's about a million and a half in capital expense. You don't shell
out a million and a half so that you can publish a coffee table book
with a total print run of a couple thousand.
11:26 PM You
shell out a million and a half so that you can print fifty coffee table
books a year and one of them gets on Oprah and has a print run of fifty
11:28 PM If
you can't stop the guy next door from ripping off your perfect binding
and running fifty thousand color copies the day before the Oprah
broadcast airs, you're not gonna shell out a million and a half for a
fine art offset press.
11:29 PM And if you ain't gonna do this painting justice when you publish it, I ain't gonna send you a snapshot to begin with.
Trent: you can't stop him anyway
again, the illusion of copyright
11:31 PM Michael:
Wrongo, boyo. He can run his fifty thousand color copies but he'll
never get distribution. What, Amazon's gonna risk having their whole
fucking website taken down by court order because they'll get three
bucks margin from him instead of two from you?
11:32 PM You like Tom Lehrer's music?
That shit cracks me up. My sister-in-law gave me the box set a couple Christmases ago.
11:33 PM No
copyright, no box set. In fact, no reissues worth listening to at all.
'Cause the only way anybody ever got the original records was by
writing to Tom Lehrer's P. O. box in Cambridge, Mass.
11:37 PM Tom's
a university-level math instructor. Used to teach here in town, though
I've never met him. Didn't need the money to begin with; performed live
for his buddies at Harvard (or was it MIT? I forget) as a
stress-reliever. But thanks to copyright, he played Carnegie Hall, and
I've got the box set. That shit cracks me up.
Trent: I'm gunna have to go soon.
11:39 PM Michael:
Cool -- it's been fun. Would be kinda nice if you let LKML know I don't
bite off bat heads in dialogue, even if you don't buy a single thing
I've had to say.
Trent: heh, ok.
11:40 PM I can reply to that other dude who tried to explain your position and say that he got it totally wrong if you like.
11:41 PM not that I completely understand it myself now :)
11:42 PM Michael: Schwartz? He's got it pretty close to straight. How would you feel about my just posting this dialogue, raw?
11:44 PM Trent: no-one wants to read this stuff.
11:45 PM Michael:
All right. How about I put up a web page and send LKML a link if and
when it fits into the flow of something I'm writing anyway?
11:46 PM Trent: you do whatever you want.. I won't stop you.
11:47 PM Michael: Fair enough. Ditto from me, just don't excerpt in a way that misrepresents my position if you can help it.
Trent: I will post this to the list:
in case anyone cares, after speaking with Michael for a few hours I've
found he's not nearly as abrasive as this mailing list banter might
suggest. He makes some good arguments once you stop him from spouting
conspiracy stuff and, although I don't agree with all of them, I think
he has some good points. He suggested posting a transcript of our chat
but, frankly, I don't think anyone wants to read that.
Michael: That would be nice. Cheers, mate -- see you in the funny papers.