4 ms·

That OOPSLA paper is essentially the patent (it was written by the patent authors). It's basically baking many forms of immmutability (e.g., immutable to view,

That OOPSLA paper is essentially the patent (it was written by the patent authors). It's basically baking many forms of immmutability (e.g., immutable to view, but not when owned, vs. immutable in the entire object graph) into the type system. As far as I can see, if any software is patentable, that is.

DannyBee 12y ago
"As far as I can see, if any software is patentable, that is. " Why? It has been known you could do this for years. There is literally nothing special about their model, other than baking it into a type system, which everyone knew you could do, but nobody had bothered to try to sell to programmers.
- ConceptJunkie 12y ago
  You don't seem to understand the patent system. It doesn't matter how obvious something is. There's a good likelihood someone at the overworked, understaffed USPTO will rubber stamp it. Bombarding the USPTO with patents for any and everything you can possibly think of is a valid strategy, because once you get it, it's a lot of work and expense to overturn and you can start extorting people in the meantime. Microsoft didn't start out with this attitude, but when it was applied to them, they decided join in and hard. If you're a programmer who can do more than write a for loop that counts to 10, I can almost guarantee you've violated a patent. In fact, I used to joke about getting a patent on the for loop. If Fortran, or whatever language first had for loops, had been created today, there would probably be a patent granted for it.
  - DannyBee 12y ago
    "You don't seem to understand the patent system." 1. I"m a registered patent attorney. 2. I run www.google.com/patents 3. I talk to examiners frequently 4. I spent years of my life working for patent reform in DC Trust me, I understand how the patent system works. The USPTO is not horribly understaffed. They have other issues. One of the main issues is actually that the patent examiners are union, and the union is not really on board with them being asked to do significantly more work (IE searching harder), unless they get significantly more pay. In a lot of cases, they can't get significantly more pay, because there are salary caps on federal employees, etc.
    - Perseids 12y ago
      > The USPTO is not horribly understaffed. They have other issues. One of the main issues is actually that the patent examiners are union, and the union is not really on board with them being asked to do significantly more work (IE searching harder), unless they get significantly more pay. In a lot of cases, they can't get significantly more pay, because there are salary caps on federal employees, etc. Is that not something you could solve with more employees? If the task does not lend to parallelization, let each team work 2k weeks on every application instead of the current k weeks and employ doubly the amount of teams.
      - DannyBee 12y ago
        "Is that not something you could solve with more employees? If the task does not lend to parallelization, let each team work 2k weeks on every application instead of the current k weeks and employ doubly the amount of teams." The USPTO does not have the budget to double their staff. They are also getting pushed as hard as possible to lower pendency periods, which are already 3+ years in a lot of cases. If they did double their staff, the union would just demand even more concession, because they would have more power, not less. If they doubled pendency, they'd get murdered by congress. So they can't really win.
    - throwawaykf05 12y ago
      Note that the "not searching harder" problem cuts both ways. Examiners get rewarded for moving applications off their desks regardless of whether they are approved or rejected. As such, I have seen as many frivolous rejections as I've seen frivolous approvals. One rejection I saw cited some unrelated reference and literally just made arguments up. There are no stats on this, obviously, but my anecdotal guess is examiners would prefer slightly to err on the side of rejection than approval. YMMV.
- Locke1689 12y ago
  The paper is novel enough for OOPSLA -- seems fine to me.
- kenjackson 12y ago
  They do more than bake it into the type system. They allow immutable types to be assigned to mutable types. But they don't allow those values to be mutated, even from the mutable types. That is type checking fails when you try to violate the constraint of the type, not when you assign one type to another that is incompatible.
  - DannyBee 12y ago
    Which is a fairly simple application of well known typestate analysis?
    - kenjackson 12y ago
      I'm not sure it's simple, but it is related. If you read their paper you'll see they talk about the relationship of this work to Bierhoff and Aldrich's paper "Modular TypeState Checking of Alised Objects" and Naden, et al's paper on Permissions.

DannyBee 12y ago

"As far as I can see, if any software is patentable, that is. " Why? It has been known you could do this for years. There is literally nothing special about their model, other than baking it into a type system, which everyone knew you could do, but nobody had bothered to try to sell to programmers.

ConceptJunkie 12y ago
You don't seem to understand the patent system. It doesn't matter how obvious something is. There's a good likelihood someone at the overworked, understaffed USPTO will rubber stamp it. Bombarding the USPTO with patents for any and everything you can possibly think of is a valid strategy, because once you get it, it's a lot of work and expense to overturn and you can start extorting people in the meantime. Microsoft didn't start out with this attitude, but when it was applied to them, they decided join in and hard. If you're a programmer who can do more than write a for loop that counts to 10, I can almost guarantee you've violated a patent. In fact, I used to joke about getting a patent on the for loop. If Fortran, or whatever language first had for loops, had been created today, there would probably be a patent granted for it.
- DannyBee 12y ago
  "You don't seem to understand the patent system." 1. I"m a registered patent attorney. 2. I run www.google.com/patents 3. I talk to examiners frequently 4. I spent years of my life working for patent reform in DC Trust me, I understand how the patent system works. The USPTO is not horribly understaffed. They have other issues. One of the main issues is actually that the patent examiners are union, and the union is not really on board with them being asked to do significantly more work (IE searching harder), unless they get significantly more pay. In a lot of cases, they can't get significantly more pay, because there are salary caps on federal employees, etc.
  - Perseids 12y ago
    > The USPTO is not horribly understaffed. They have other issues. One of the main issues is actually that the patent examiners are union, and the union is not really on board with them being asked to do significantly more work (IE searching harder), unless they get significantly more pay. In a lot of cases, they can't get significantly more pay, because there are salary caps on federal employees, etc. Is that not something you could solve with more employees? If the task does not lend to parallelization, let each team work 2k weeks on every application instead of the current k weeks and employ doubly the amount of teams.
    - DannyBee 12y ago
      "Is that not something you could solve with more employees? If the task does not lend to parallelization, let each team work 2k weeks on every application instead of the current k weeks and employ doubly the amount of teams." The USPTO does not have the budget to double their staff. They are also getting pushed as hard as possible to lower pendency periods, which are already 3+ years in a lot of cases. If they did double their staff, the union would just demand even more concession, because they would have more power, not less. If they doubled pendency, they'd get murdered by congress. So they can't really win.
  - throwawaykf05 12y ago
    Note that the "not searching harder" problem cuts both ways. Examiners get rewarded for moving applications off their desks regardless of whether they are approved or rejected. As such, I have seen as many frivolous rejections as I've seen frivolous approvals. One rejection I saw cited some unrelated reference and literally just made arguments up. There are no stats on this, obviously, but my anecdotal guess is examiners would prefer slightly to err on the side of rejection than approval. YMMV.
Locke1689 12y ago
The paper is novel enough for OOPSLA -- seems fine to me.
kenjackson 12y ago
They do more than bake it into the type system. They allow immutable types to be assigned to mutable types. But they don't allow those values to be mutated, even from the mutable types. That is type checking fails when you try to violate the constraint of the type, not when you assign one type to another that is incompatible.
- DannyBee 12y ago
  Which is a fairly simple application of well known typestate analysis?
  - kenjackson 12y ago
    I'm not sure it's simple, but it is related. If you read their paper you'll see they talk about the relationship of this work to Bierhoff and Aldrich's paper "Modular TypeState Checking of Alised Objects" and Naden, et al's paper on Permissions.