Slashdot Mirror
Google Patents Staple of '70s Mainframe Computing
theodp writes "'The lack of interest, the disdain for history is what makes computing not-quite-a-field,' Alan Kay once lamented. And so it should come as no surprise that the USPTO granted Google a patent Tuesday for the Automatic Deletion of Temporary Files, perhaps unaware that the search giant's claimed invention is essentially a somewhat kludgy variation on file expiration processing, a staple of circa-1970 IBM mainframe computing and subsequent disk management software. From Google's 2013 patent: 'A path name for a file system directory can be "C:temp\12-1-1999\" to indicate that files contained within the file system directory will expire on Dec. 1, 1999.' From Judith Rattenbury's 1971 Introduction to the IBM 360 computer and OS/JCL: 'EXPDT=70365 With this expiration date specified, the data set will not be scratched or overwritten without special operator action until the 365th day of 1970.' Hey, things are new if you've never seen them before!"
Google Labs is supposedly working on a next-gen programming development environment that allows source code statements to be physically manipulated like a deck of cards.
It's not like Microsoft was ever going to be interested in that anyway. They must get cents back from the disk manufacturers for perpetuating their ever-growing temp folders.
If you actually read the patent, it is specifically for a similar method, but designed for Distributed File Systems. This is different from just a single file being names a certain way. It is an algorithm based on the location of other related files, each different file's modified and Time to Live (TTL) dates, and the factors determined by the, keywords here, plurality of servers. If they tried to patent a regular temporary file that would be different, but this is a distributed system specifically for a file that is distributed in different parts on different systems. If you still think this has been done before, I would love to see the source for that information and gladly would recant myself given that.
The man who cannot imagine a horse galloping on a tomato is an idiot - Andre Breton
The summary is wrong. Folks, please stop reading the abstract, and read claim 1 instead.
This is what is patented:
1. A computer-implemented method comprising: selecting a file having a path name in a distributed file system, wherein the file is divided into a plurality of chunks that are distributed among a plurality of servers, wherein each chunk has a modification time indicating when the chunk was last modified, and wherein at least two of the modification times are different; identifying a user profile associated with the file; determining a memory space storage quota usage for the user profile; deriving a file time to live for the file from the path name; determining a weighted file time to live for the file by reducing the file time to live by an offset, where the offset is determined by multiplying the file time to live by a percentage of memory space storage quota used by the user profile; selecting a latest modification time from the modification times of the plurality of chunks; determining that an elapsed time based on the latest modification time is equal to or exceeds the weighted file time to live; and deleting all of the chunks of the file responsive to the determining.
I think it's time for a crowdsourced patent challenge web site run by the USPTO where there would be a period of public comment for each patent about to be awarded in order to help underpaid (and I imagine under-resourced) examiners find Prior Art.
A lot fewer patents might be awarded, but ones that are would be genuinely new -- this might also save the world billions of dollars.
Did you also inspect the quota for the user owning the file to determine if you should delete it? Were the files also stored in a distributed file system, with chunks of the file on separate systems?
Every single claim in that patent mention both of those things.
The naming of files is an example of a part of a claim. To infringe on a patent you need to infringe on at least one entire claim.
I bet when all the kids were super-excited about programming on the i386 with its "OMG VIRTUAL MEMORY!!!" the older guys who had worked on mainframes just rolled their eyes. :)
You talking about the same old grey beards that gasped when the kids opened the cover on a server and added their own memory, network adaptors, backplanes, disk drives, etc without having to call IBM out to do it?
Yeah, and then rolled their eyes again because the kids didn't know about change control, didn't notify the users about the outage, didn't verify that their backups were good (if they even had backups), and lost 6 months worth of corporate data as a result.
--
.nosig
find /tmp/* -mtime +14 -exec rm {} \;
"GET / HTTP/1.0" 200 51230 "-" "Mozilla/4.0 (compatible; Setec Astronomy)"
As an aside... I remember a few years ago - when we were still running tape backups - I went to one of our then-sysadmins and asked him to recover an important directory one of our faculty had managed to delete. I was told he couldn't do it because it would require they stop the backup system for several hours, which would throw their backup tape rotation scheme out of sync.
So we were continuously generating backups we could never actually use.
#DeleteChrome
But the USPTO is populated by idiots.
They are deserving of the disdain and ridicule reserved for the Postal Office, Congress, etc.
Which is a shame because I've always figured they had some pretty smart people there. The examiner should have taken a shit on the application and mailed it back with a note saying,"this is what your application is worth".
They are either complete morons or...are getting payoffs. And Google will just use it as club some day on a small outfit that doesn't have half a million dollars to fight a lawsuit.
When Fascism comes to America, it will call itself Anti-Fascism, and tell you to give up your guns.
Yes; maybe; and the whole summary is stupid. From claim one of the patent; the very first paragraph:
having a path name in a distributed file system, wherein the file is divided into a plurality of chunks that are distributed among a plurality of servers
So; where the mainframes of the 70s had single consolidated disks stores this is talking about doing this on a distributed filesystem. The area of application is indeed new completely opposite to the claim of the summary.
Patents are not supposed to control what you do; instead they control how you do it. Since the way that Google is claiming to do this is by going around comparing the timestamps on a bunch of different distributed chunks of a file, this is something that no mainframe of the 70s is likely to have had to to so it may even be a new way to automatically delete temporary files. I wish people would begin to understand this and commenters would point it out every time. I wonder if this isn't a bunch of patent lawyers trying to make us look silly.
Having said that; If you had a distributed file which kept a timestamp on each of several separate chunks, how would you go about deciding when to automatically delete it? My guess is that the solution you would come up with quickly is basically the one in the patent. You certainly wouldn't have great difficulty in deciding how you do it; suddenly think "maybe there's a patent that might tell me how to do this"; go to the patent office and read the patent then come back inspired and manage to solve your problem only because Google was so nice as to publish their solution. Patents are supposed to record valuable secrets that companies might otherwise keep to themselves in a way that helps humanity. This one is failing at that.
What this comes down to is that the whole idea of patents on things as abstract as software is stupid and is an illegal interference in free speech a right everyone should have under the universal declaration of human rights. The patent officers of the USPTO and the congressmen who put them there should be arrested.
=~ s,(.*),<sarcasm>$1</sarcasm>,g if any_point_you_wish();
The "summary" wholly misrepresents what the patent is about. It's not about having an expiration date in the filename at all. When someone advocating a position lies to me, as this submitter did, I figure the reason they are lying about the issue is because they realize that the truth doesn't support their position.
Rather than choosing an expiration date ahead of time, the patented method deletes a file (or not) based on multiplying the time to live by the inverse of the user's quota usage, plus the latest of several modification times. The patent covers only using that specific algorithm, and only when the TTL is represented within the filename.
Is that algorithm obvious? Several Slashdot commentors who say the are programmers read the explanation of the algorithm and still didn't understand it at all. One might say that if it's explained to you and you don't "get it", it's probably not obvious.