Archive for August, 2005

Code for GCC patch tracking

Tuesday, August 30th, 2005

I’ve gotten a couple requests for the code for the patch tracking queue that I wrote for GCC. The grabber works because it knows how to fetch raw messages from the web archive. The rest is just nice formatting for the database.

The code is all mysql + mod_python + SQLObjects + Cheetah template engine. It used to be just plain python db api, but SQLobjects ended up being a lot nicer. You can see how the code has evolved over the year in the repo. It actually has an RSS feed and simple XMLRPC support, one of which is used by a Wiki, and the other, by an IRC bot :).

Here is the SQL file for the database.
None of the searches that are done were worth indexing yet.

The code for the web app and grabber can be found in the SVN repo.

It would require a bit of modification to use elsewhere, but not much. The XMLRPC code still uses the Python DB API, mainly because we transfer pickles, and transferring pickles of SQLObjects doesn’t work. It seemed pointless to grab it from SQLObjects then convert it, then pickle it, when we could directly pickle the results of the DB API queries.

If you need a tarball or something let me know.

Attention horrible Westchester drivers

Sunday, August 21st, 2005

If you fall into one of the following categories:

  1. People who run red lights 5 seconds after it turns red
  2. People who run red lights at all only to be stopped 3 seconds later by the next red light
  3. People unclear what the sign in front of them stating “Don’t block intersection” means when the traffic on the other side of the intersection has not moved for 10 minutes
  4. People who think that it’s worth almost running over pedestrians crossing in crosswalks
  5. People who swerve from lane to lane because they are actually talking to their kids in the backseat, completely turned around
  6. People who treat lane lines as just suggestions

then please be aware we have taken note of who you are.
You are hereby banned from purchasing a flying car.
You may all continue to run red lights straight into each other, or others blocking the intersection, until none of you are left.

Why prior art doesn’t matter - combining references

Wednesday, August 10th, 2005

Engineers and CS people often scream “I know prior art” when faced with a software patent. This post explains, from a legal perspective, why prior art rarely matters.
The main reason most prior art doesn’t actually matter is simple.

  1. It is rare for a single reference (paper, etc) to render an invention obvious
  2. Combining references is subject to the teaching-suggestion-motivation test, which renders 90% of them useless

What exactly is this teaching-suggestion-motivation test?
Well, current Federal Circuit precedent (the appeals court that is responsible for patent cases) states that the prior art must provide “A teaching, suggestion, or motiviation to combine the references”.
Even if the references teach every element of the claimed invention, their is no case of obviousness without a motivation to combine. See In re Rouffet, 149 F.3d 1350 (Fed. Cir. 1998). In addition, the level of skill in the art may not be relied upon to provide the suggestion to combine references. See Al-Site Corp. v. VSI Int’l Inc., 174 F.3d 1308 (Fed. Cir. 1999).
One could argue that this is ridiculous, because it would require that every person writing papers provide explicit suggestions to combine their work with other previous work, when it really is obvious to anyone with a brain that this should be done. In the CS world, it probably is ridiculous. However, there is no precedent that does away with this requirement for any type of invention.
More details about how the patent office deals with obviousness can be found in the MPEP (Manual of Patent Examining Procedure).

There is also a case that has the potential to change this. It is currently on appeal to the Supreme Court (and is awaiting a decision on cert.). See Patently-O blog for details.

Cats

Saturday, August 6th, 2005

My friend lost his cat.
Because his cat has run off, he believes his cat is dumb.
I’ve tried to explain before that there are no dumb cats, only sneaky cats and less sneaky cats.
You can test how sneaky your cat is by bringing it to Arizona and living for a year or two.
If your cat gets eaten by coyotes, it was not sneaky.
If your cat ends up getting fatter and appears to have been getting twice as much food as you give it, it is sneaky.

In my friend’s case, his cat has probably found a new home and been renamed “Mrs. Fluffykins the third”, and was previously getting twice as much food as he gave her.
The other option is that she had an unreported auto-accident.
For those who think it’s okay to not report to their insurance company such cat <-> car fender benders: We know who you are.
The fact that most cats are uninsured is not a good excuse.

I’m back

Saturday, August 6th, 2005

After a long hiatus of not caring about having a blog, i’ve been goaded into having one again.

The current thinking is that what will happen is this:

I forsee lots of posts that consist of a link to idiocy and the words “Hey, you. Cut that out.”
And maybe the occasional picture of your rabbits devouring a kitten.

I give this a 90% chance of being true.