Slashdot Mirror
USPTO Issues Microsoft A Patent For 60's Technology
theodp writes "On Tuesday, the USPTO issued U.S. patent no. 6,594,674 to Microsoft for a System and method for creating multiple files from a single source file, which describes a fundamental IBM Mainframe file structure, the Partitioned Data Set, that's been around since the 60's and is familiar to virtually anyone who's used a mainframe text editor in the past five decades. To the amazement of readers of an IBM newsgroup, neither Microsoft nor the USPTO examiners seem to be aware of the existence of the Mainframe-based prior art, which is not cited in the patent."
Doesnt Reiser FS 4 also have this feature.
Creates multiple files from a single source file.
Prior art.
My foot's asleep.
In another news Microsoft applies for a patent on "circular-shaped-low-friction moving support for vehicles" as an adendum to it's "Microsoft wheel patent".
Officials are not aware of any prior art16,777,216 comments ought to be enough for any forum!
Even the System 7 Finder had a feature [still exists in OS X] for creating "stationary".
Any file with the "Stationary" checkbox checked, becomes a template of sorts and opens not the original, but a copy of the stationary, ready for saving, etc.
You figure with all the massive amounts of obvious prior art that the USPTO would be slightly less broken about this one.
Sigh
Blocklevel: Practical Information Architecture
dd if=/dev/zero of=patentedbyMS bs=1024k count=1024 /mnt/priorart
mke2fs -b 1024 -F patentedbyMS
mount -t ext2 -o rw,loop=/dev/loop patentedbyMS
~~~~~~~
"You are not remembered for doing what is expected of you." - Atul Chitnis
I can't speak for the Patent Office, but MS aren't aware of this prior art because he's simply too young. In the 60's, Gates hadn't even smoked his first crack pipe yet.
How are you going to keep them down on the farm once they've seen Karl Hungus?
I guess the answer is "Yes, a patent was granted".
/sbin/mke2fs myfile
Now I am going to be a mudraker. Here is an implementation of this patent under
Linux (so sue me). Effectively only 6 lines of shell script. Do what you want with it.
I would have thought that it should take longer than 4 minutes to implement a patent -- this one is really pushing "obvious", even ignoring the prior art issue.
# create a 2mbyte pds
dd if=/dev/zero of=myfile bs=1k count=2k
# make it into a file system (pds)
# make a mount point for the pds
mkdir mydir
mount myfile mydir -o loop
# make some files in the new pds
for i in a b c d; do touch mydir/$i; done
# and now some links to the members
for i in a b c d; do ln -s mydir/$i $i; done
# now, if you modify file a, b, c or d
# the contents of the pds "myfile" are changed.
# the pds can be unmounted, and read or
# written as a single file "cp myfile myfile2"
# would be an example.
# When mounted the symbolic links allow automatic
# internal access
Ratboy.
Just another "Cubible(sic) Joe" 2 17 3061
On a not-so-close reading wouldn't Macs' resource and data forks and ResEdit fall under this patent as well?
To the amazement of readers of an IBM newsgroup, neither Microsoft nor the USPTO examiners seem to be aware of the existence of the Mainframe-based prior art, which is not cited in the patent."
I don't understand why they are amazed. Mainframe technology is essentielly the Dark Age of Computing. Nobody knows about it, and no place teaches it.
What are "Partitioned Data Sets"? Dunno, I've never heard of them. The link certainly doesn't explain, that's just dinosaur mumbo jumbo. (And what's so special about that editor screenshot?)
This is by the way IBM is porting Linux to their mainframes. Customers might need the high reliability etc., but they still won't buy them if they can't find anyone who knows how to use the native operating systems, let alone program them.
Does this mean people running things like CICS have to pay Micro$oft on royalties? This deserves the foot, man.
This sig no verb.
neither Microsoft nor the USPTO examiners seem to be aware of the existence of the Mainframe-based prior art, which is not cited in the patent.
In all fairness, Microsoft probably knew full well that this prior art existed. Putting that in the patent application would be like sending the USPTO a stamped, self-addressed rejection letter!
It's the USPTO people who are missing this experience to find prior art. But can we really expect one (government) organization to have deep experience in every field? There's something fundamentally wrong with the idea of the USPTO.
Remember Stacker?
In fact, didn't they lose a patent lawsuit back in the early 90's over the disk/volume compression driver they included with DOS 6.0 or 6.2? It's been so long I don't remember the name, but IIRC they had to remove the patented algorithm in favor of a less efficient unencumbered alternative, change the name of the compression system and issue the changes as a DOS6.21 update release.
Am I crazy or is there really no difference, and MS has fallen into some IP time-warp? What's next; patenting the linked list?
We need a private alternative to the PTO - it's clear they can't do their jobs anymore. Hell, all they'd have to learn is "google.com" and "slashdot.org" and they could reduce the number of crap patents by a factor of 100.
"Lawyers are for sucks."
- Doug McKenzie
How is this different from zipping up files and accessing the zip like it's a folder/directory?
This sort of thing is starting to happen way too often.
I'm wondering how long it will be before someone sues the patent office for issuing a dud patent and wasting everyone's time.
Maybe that will stop the flow?
(Spudley Strikes Again!)
First action is usually a rejection on something, but not always. Could be language / semantics, or something. Could also be peior art if the examiner finds any.
The second action is then the final rejection or allowance. This needs to be based on the first action, so if the first action was a rejection on language then the second action is usually an allowance. If the first action was based on prior art and the art was good then it is usually a rejection. Most stuff is never black and white, there are some 65million colors that most patents fall into.
If you go back and forth more than 2 times on any patent then you spin your wheels as a patent examiner, and don't get credit for a third or fourth action if you have to do more than 2 actions on a patent. Thus if you don't find anything in a period of time, then you have to allow it.
As you get promoted in the patent office, you are expected to do MORE actions. So if you start out as a GS7 you have 4 actions about every 2 weeks. When you become a GS9 it is 5-1/2 actions about every two weeks. (GS is goverment job scale ratings. ) It was like that when I was there and I doubt it has changed much. The actual number of actions may not be 100% correct but it is close enough that anyone with an IQ smarter than a tomato plant can get the general idea.
Only 'flamers' flame!
Does slashdot hate my posts?
http://www.theonion.com/onion3311/microsoftpatents .html
Check out the link that someone posted in that IBM Mainframe thread!
Paul B.
An example is shown here HOWTO: Use NTFS Alternate Data Streams
I'm not sure if a similar feature existed in prior OS or not, since I'm not familiar with the mainframe OSs of long ago, but, I think this feature has been a part of the NTFS file system since NT 3.1
Of course, the whole NTFS Alternate Data Stream (ADS) thing can be evil if it is misused. I mean you cant see ADS using DIR or Windows Explorer.
[i]"To the amazement of readers of an IBM newsgroup, neither Microsoft nor the USPTO examiners seem to be aware of the existence of the Mainframe-based prior art, which is not cited in the patent."[/i]
:D
Duh, of course you don't include references to prior art in your patent application. What kind of idiot would do that?
"Oooh, does that mean we get to kick some puffy white mad zionist butt?"
It's not great news that the patent was issued, but it's not disaster either. If/when Microsoft tried to sue (Rieser? MacOSX? Take your pick) someone for infringing, then the patent is nullified by the courts.
Just getting a patent issued to you doesn't mean the story is over.
--
Obviously, nobody here read or understood the patent. The idea is that it's slow to write lots of small files because the head is constantly seeking to the inode table, to the directory, to the data, etc.
To see what this patent would do, make a 4MB file, and tar it up. Then make 1000 4KB files, and tar them up. The second file is a few percent bigger than the first one, right? Now untar the first file, and note how long it takes. Then untar the second file, and note how much longer it takes -- not just a few percent longer, but orders of magnitude longer.
Ideally, tar would just:
1. allocate and write 4MB of data
2. write 1000 inodes
3. make 1000 directory entries
However, tar has to do this:
1. allocate and write 4KB of data
2. write 1 inode
3. make 1 directory entry
4. repeat 1000 times!
Microsoft is patenting a way (an API, essentially) to allow tar to do the Right Thing in that situation. This operation could also solve speed issues with GConf.
While there are many similar things, no OS has an API that does this, right? Therefore, there is no prior art.
Next you'll tell me that the food pyramid was made for the agriculture industry, not nutrition!
A few MSDN links that show Microsoft not only was aware of, but also supported the decades-old mainframe Partitioned Data Set (PDS) file structure in its product offerings...
= /library/en-us/his/htm/_sna_a.asp
= /library/en-us/his/htm/_sna_platforms_supported_by _the_host_file_transfer_activex_control_oledb.asp
--> http://msdn.microsoft.com/library/default.asp?url
Microsoft Host Integration Server 2000
Glossary
data set members
Members of partitioned data sets that are individually named elements of a larger file that can be retrieved by name.
partitioned data set (PDS)
A data set in direct access storage that is divided into partitions, called members, each of which can contain a program, part of a program, or data.
--> http://msdn.microsoft.com/library/default.asp?url
Microsoft Host Integration Server 2000
On the mainframe platform, the Host File Transfer ActiveX Control supports the following data set types:
Basic Partitioned Access Method (PDS) data sets
* Partitioned Data Set Extended members (PDSE)
* Partitioned Data Set members (PDS)
Any examiner who approves a stupid patent (i.e. has prior art, is obvious or not novel) GETS FIRED! Bet those pathetic desk jockeys would suddenly clean up their act...
Nevertheless, this kind of information can still be missed by patent offices in a search. I am at this moment checking the (in-)validity of several patents for which applications have been filed about ten years ago. Already at that time, not every patent office (Europe, US; Japan used to be and most of the time still is very slow) was able to gather the proper prior art for every application.
Landmark prior art publications are cited for one application in the field, 'preventing' grant, whereas they are not cited at all for another application in the filed on the same subject. This happens, for two applications handled in the same patent office at approximately the same time(USPTO in this case).
Why, you might ask... Well, for the USPTO: people don't stay long, the attorney/agent business is too lucrative. And for both USPTO and EPO: it's difficult to find what you are looking for in a very short period of time, in a very very large database.
Well, that brings up all kinds of validity issues, outside and inside the courts.
Isn't a Doom .WAD file much like this patent?
So if MS has a patent on this, WTF do I have to use PGP disk in order to get this functionality?
MacOS allowed you to access the resurce fork with the same API as the data fork for read/write/open (you just specified it was the resource fork vs the data fork), but that was frowned upon -- there is a separate resouce manager that stored data in a structured format.
BeOS (BFS) allowed for an arbitrary number of streams (attributes) which were identified by name but also had a typecode identifier (so you could tell if it was an integer, a mime type, a string, an image, etc). These attributes were optimized for small chunks of data, but could hold large chunks of data as well. This was in the time frame between NT 4 and Windows 2k.
In BeOS, you had to open the main file directly first, then used a different API to get a listing of attributes or read/write/delete them.
BeOS attributes were influenced by SGI's XFS, IIRC, so they may have had a similar feature before NTFS as well.
Do you even lift?
These aren't the 'roids you're looking for.
Based mostly on the replys I'd say the patent office isn't really checking patents very well and realisticly can not actually do a halfway decent job.
This sets up a bad situation.
So what is needed is a public/advocate review phase so the public can present publcly known prior art to the patent office before patents are issued and then have to go to cort with obveous prior art.
I don't actually exist.