EPO != Europe's patent office

The patent offices of europe are the patent offices for those countries, not the EPO. The EPO is another patent office.

EPO patents can be gained in the UK just as easily as they can be gained at the EPO in any other european country.

It's as difficult to get a patent in the French patent office as in the UK one.

Maybe it's time...

to invest my HEC in 'Intellectual' property, as opposed to physical property, as the arse appears about to fall out of that. Come up with some itty-bitty little process, get a patent, sit back and wait for the "infringement" cash to just roll in...

Sorry.

I was patently indulging in a little fantasy there.

Ah well,

Back to work then.

Re: EPO != Europe's patent office

emmm.... the UK is a contracting state of the European Patent Convention, the law which the EPO uses to grant patents, and as such, amended UK Patent law (the Patents Act 1977) to correspond to the EPC. So EPO law and UK law should be the same...

before anyone gets all shirty, I'm not commenting on "software" patents.

Re: AC

No, don't worry, not going to get shirty.

However, the EPO can still be applied for in the UK. That the EPO doesn't have the right to issue patents by EU treaty isn't a problem for them and not, apparently, for those applying to the EPO for these patents.

So the gist of my point is that using the ability to get a patent number from the EPO is NO argument that the UK PTO must also grant one.

If that is the basis on which this judge ruled, he's either clueless or been led up the path.

They think it's not software

What the patent is,

You can link between a DLL and a program by name (existing prior art).

You can link between a DLL and a program by ordinal number (existing prior art).

You can have fixed ordinal values and still use names, read up on old .DEF files from the Microsoft compiler. That claim is not new.

Some entry and exit points are already fixed ordinal. Again that claim is not new.

What they're proposing to patent is splitting the DLL from the interface to the DLL, so that you compile against a fixed DLL with fixed ordinals which access the correspond dynamic DLL by name, same as ODBC does for example. i.e. Call the .LIB file a .DLL, pretend

Oh boy, it's like being in Computer Science 101 class again.

This illustrates the problem with these patents, just because nobody's ever tried to patent this doesn't mean it isn't common practice. The more common something is the less likely it is to be patented. i.e. this should be disallowed as the patented subject is what happens now in the real world whenever we load a driver. ODBC is an example of this. Nothing in this patent is new or novel.

Then we get to the EPO's games. They changed (despite previous national court rulings) that TRIPS requires it makes anything 'technological' patentable so they slotted in a phrase in Dec 2007 to expand the scope of patents.

TRIPS doesn't say this and the interpretation of TRIPS is subject to national law. So TRIPS can say 'all children need be branded with a hot iron' and it means squat if there's no law to implement it.

We've already been over this a million times, software isn't patentable because trade secrets are more effective than patents. Nokia wanted then to protect it's market and I reckon Symbian is just acting as a front for Nokia in this.

They're trying to pretend that the interface to the DLLs is not an algorithm 'as such' I just don't see how they can claim that. It's clearly an algorithm and not patentable, even if it was novel.

Forget about it

I used to be all in favour of software patents, licenses and the like but I'm beginning to think there are just too many patently unreasonable and probably unenforceable strings being attached to pretty much everything in sight.

Only this morning I ran a Whois on a domain that turned out to be registered through 1&1. I got back some obnoxious boilerplate about how I agreed to 1&1's terms by running a Whois on that particular domain. How am I supposed to know who the registrar is without first running a Whois and therefore how can I agree to anything just by running the query? In short it is a case of "By reading this contract you indicate your acceptance of its terms".

Their terms wouldn't have any hope of standing up in court of course. But the very fact that companies are willing to even try this manner of reserving rights they never had in the first place indicates to me that their rights need further restriction because they cannot be trusted. The last thing they need is yet more empowerment.

ODBC32.dll

Then this was used in ODBC, you don't link to the ODBC driver, you link to a DLL ODBC32.dll which is linked by ordinal (but you could also link by name, and the patent points out this is prior art).

When I call something link SQLExecute, I am linked against the version independant ODBC32.dll, that in turn has a bunch of functions that pass the data to the driver via late bound-by-name calls.

e.g. The call SQLExecute is just a function which passes the call data to the drivers (also a DLL), which is late bound by exported name to SQLExecute in the driver.

Provided of course the method is supported it is mapped, e.g. this states which functions are mapped by which version.

http://msdn2.microsoft.com/en-us/library/ms713848(VS.85).aspx

Man it's like they just taking the mick, trying to cut a huge amount of existing techniques from under us.

RE: They think its not software

I'm no expert in the technical field but if the method claimed could be proved to have been common practice before the filing date of the patent the EPO patent could have been successfully opposed. This is true for any technical field including "computer assisted inventions".

One of the problems facing the judges / technical boards etc is often not to determine if something is new but if it is inventive since inventiveness or obviousness is difficult to judge objectively. One person's flash of genius is another's common practice.

@Barnaby

Barnaby, unfortunately two problems with this

a) courts seem to take the stance that if the PTO approves it, you must prove that the patent is wrong. Meanwhile

a.1) PTO's assume that courts will sort it out, so don't work too hard to see why a patent needs to be approved

this means

b) you have to take it to court and this is expensive. Especially since courts demand you accede to their filing procedures even though you are unable to know what they are and how to apply them (so need a solicitor, who needs to be better [hence more expensive] than their solicitor).

So although you may be able to make up a really GOOD defense, you can only apply this in court and the patent troll (effectively, because the patent is ridiculous) will ask for less than the cost of defense.

Patents are only sensible on REAL objects, not software or algorithms. The reason for this is simple: real life can bite you in the bum severely. An example may help show how:

Set up a gearing system that uses a lot of gears and covers a large range of gearing ratios.

On paper or as an algorithm, this will ALWAYS produce an answer that works.

Implement in real life and you have problems with backlash and gear resistance. Your patent would be on how you changed the optimal paper solution to overcome these problems.

Software has no real life bite-in-the-bum scenarios.

Algorithms have no real life bite-in-the-bum scenarios.

When you apply your software analogue to a real system (like the gearing problem: use it to work out how to create the gears that will do this and then cut the gears), you get bitten. And that may be innovative enough a solution to afford social value to awarding a patent.

@They think it's not software

It's pointless arguing the technical merits of such patents. They are almost certainly going to be thrown out if challenged. This one is a little less laughable than hundreds of others, some of which put into question the sobriety, let alone the sanity, of the world's patent examiners. (Did I just hear Einstein groan?)

So what's the point. Well, there are two ways to make money out of a patent which is BS. One is simple: extortion. You rely on the fact that most companies would rather pay over a few cents of their earnings dollar than go to litigation. All it costs is a mail-out to likely victims. They may not even be using your "invention" but why let a small fact like that get in the way of good cash flow. They will simple pay up to make you go away. Eventually of course you will hit someone with both spine and principle and the jig is up, but even if the patent is ultimately invalidated you get to keep the loot you ripped off in the meantime.

The second method is even more insidious. It's called a "blocking patent" and is designed to put roadblocks in the path of your competitors development efforts. A "prominent handheld communications device manufacturer" is famous for this tactic. You claim a patent knowing that the competitors searches will dredge it up sooner or later. Their technical team will then have to waste precious time evaluating your claims, modifying their designs to get around you, writing docos to have a paper trail in case of litigation, getting their attorneys to OK things etc. etc. etc. Of course, all along they know that your clams are pure BS but they will go to almost any lengths to avoid testing them in court. Courts are no longer interested in little things like truth, justice, fairness or even common sense, and you could well lose a case fought on narrow legal (i.e. BS) principles. The patentor wins in either case. His aim was to get you to waste time and money and what could waste more time or money than a flock of lawyers?

So there you have it. The sharp end of "intellectual property rights". A toll gate or a roadblock - take your pick.

Fleshing out the details of how old this is

OK, trying to flesh out the details

Here is a sample ODBC driver (remember this driver is called from the odbc32.dll lib).

http://www.ibphoenix.com/main.nfs?page=ibp_60_odbc

I've downloaded the source code to OdbcJdbc_Source-1-2-1 from this page.

So my call to odbc32.dll SQLExecute is passed by the .Lib binding indexed by ordinal to the corresponding call in the odbc32.dll which then late binds via the name to the drivers implementation which is at line 1543 in odbcstatment.cpp.

OdbcJdbc_Source-1-2-1\OdbcStatement.cpp(1543):SQLRETURN OdbcStatement::sqlExecute()

Now look at the .DEF file for this driver

ODBCJDBC.def

And there is the export by name:

LIBRARY "OdbcJdbc"

;DESCRIPTION "OdbcJdbc -- Jim's ODBC/JDBC Driver"

EXPORTS

SQLAllocConnect

SQLAllocEnv

SQLAllocStmt

SQLBindCol

SQLCancel

SQLColAttributes

SQLConnect

SQLDescribeCol

SQLDisconnect

SQLError

SQLExecDirect

SQLExecute << here

SQLFetch

SQLFreeConnect

SQLFreeEnv

SQLFreeStmt

SQLGetCursorName

SQLNumResultCols

SQLPrepare

In the Symbian patent, the 'Extension.dll' is the Odbc32.dll (and it's corresponding lib) and the driver is the platform dll (and it's corresponding lib). The particular database platform is not known when you link to odbc32.dll.

I've picked ODBC as an example, but this was not the first, it's just it predates the web and it gets more and more difficult to locate the docs for the older stuff.

I was able to do this only because I could get the source, most proprietary software is protected by trade secrets. If I was not able to get the source, then this patent may stand and would hold back Europe for 20 years as Symbian played it's stupid games.

@Joe M

"It's pointless arguing the technical merits of such patents."

Agreed they intended to make a toll gate or road block.

But I disagree that there isn't a point in arguing the technical merits of the patent.

You know this makes me damn angry. Reading the details of the judgement makes it clear that Symbian understood the mechanics of .DEF files on Windows. So they must have known that GDI (gdi32.dll) calls the platform specific display driver to draw to the screen for example.

They must have (like I have) dealt with example after example of one DLL calling a platform specific dll by name by ordinal, by mixtures of both, by querying for the name the calling the ordinal, in fact every combination of this.

The C:\Windows\System32\drivers folder is full of hundreds of platform specific driver files whose calls are called indirectly through known interface DLLs.

Yet they tried to get this patent anyway? Yes I know gdi32.dll is proprietary to Microsoft so we couldn't get the source code to prove it, but they must know it, even if they mistakenly think it couldn't be proved.

And we would never have noticed this patent except that it was mentioned on a news site?

@Greg Fleming

In FBW systems, there IS no patent.And the BITB scenario is solved with hardware:triple redundancy and vote contention algorithms (and algorithms aren't patentable either).

But the main thing is that FBW doesn't have patented stuff in the software. HArdware, yes. Software no.

So even in this very delicate and demanding role, software patents aren't used.

re: Fleshing out the details of how old this is

thanks to Anonymous Coward for 'fleshing out the details of how old this is':

as far as I can see if this patent were granted they would have a block on most (low level) ways of 'coding to interface not implementation'!

AndyD 8-)#

@ Mark

Aye, but the distinction between what is software and what is hardware is often blurred. Maybe not to you or me, but to corporate legal departments. Very often the software *is* the hardware, so inextricably bound into it and that's what the final commercial product is. The electronics are so just much junk without all the wee bits of of embedded assembler.

But yes, we're not talking about algorithms or the real-world implementation of those algorithms in software in FBW, we are talking about a real-world finished hardware "product", so I agree with your point there.

I just wonder how this would ever pan out in the implementation of true pure AI systems if they ever eventually materialise.

Not that I ever think that they will, at least not with digital computers as they are currently defined.

@Greg

Yes, it is blurred. Unfortunately because accountants now define what is right. And that means what makes most money or stops other from taking more than "their share". So they are predisposed to see how it is blurry.

But the software patent equivalent of a hardware patent would be something along the lines of:

"A method by which verminous biological elements are restrained or removed from encroching activity via a restraint method that cause reduction in mobility by means of a one-way or unidirectional entrapment ..."

So now find any way of building a mousetrap is patented.

This doesn't happen (often) in hardware because people are SUPPOSED to see the entirety of the thing they bought, and therefore able to see the specifics of how the operation took place. However, closed software has now been ingrained in the public conscious and you don't SEE the method taken to produce the result. You just see the result. That's what gets patented.

And in the mousetrap example, that isn't patenting the trap used but the result: a trapped mouse.

Take any software patent and build something that enacts it. Damn near 100% guarantee that the code used will be completely different even to a partly-cognisant layman. So the patent isn't worth squat.

Until software patents give up on secrets AND copyrights to protect their patent, we will continue to see the result patented (trapped mice) rather than the method (the trap).

A lot of the reporting and comment on this is wildly inaccurate

Including the original story here in El Reg, and the version on the BBC Web site,

1) The appeal wasn't against the refusal of the UKIPO to grant the patent; it was against the refusal to examine the patent on the grounds that it was a software invention.

The judgement of the High Court doesn't mean that the patent is now granted; it simply means that the subject matter has been ruled patentable and that the application can proceed to examination. As and when it does get examined (presumably after the next appeal) the various objections made in these comments as to novelty and inventiveness will, if valid, no doubt be raised by the examiners, who can still reject the patent if it fails either of those tests. All that the judgement means, if it's upheld, is that they can't refuse to examine it on grounds that it isn't patentable.

In other words, the appeal judgement doesn't mean that the patent is now granted.

2) If you look at the text of the actual application (available on Espacenet) it's clear that the invention is actually rather more subtle than commentators here seem to appreciate.

The claimed invention isn't about the sort of indirect linking to DLLs present in OBCD. It's about a method of maintaining binary compatibility between DLLs in different versions of an operating system where the original DLL has had to be been extended in incompatible ways by the OS vendor and one or more of that vendor's customer.

This is possibly a scenario that's peculiar to Symbian's customers. They provide a DLL which is extended by (say) Nokia with some functions and also by (say) Sony Ericsson with different incompatible functions, and then subsequently has to be re-release by Symbian in the next version of the operating system with their own (also incompatible) extensions without breaking compatibility with any of the versions previously used by their customers.

Presumably the reason for Symbian's appeal is that the patent application protects their business model; the problem is solves is one that anyone trying to provide an OS that is both customisable and upgradable is likely to come across.