3 ms·

"Come on. Implementing something you invented should be easy." Sounds like someone who has never implemented something more complicated than a hello world.

by roel_v 16y ago

rbanffy 16y ago
Sounds like you don't know me. In the past 25+ years I got my share of bugs. But having a correct implementation of a spec is kind of the only proof you can have the spec is complete and implementable. Not having one is just sloppy.
- roel_v 16y ago
  Of course I don't know you, but with 25 years you should be in a position to recognize that a spec as complicated as this one is near impossible to implement 100% correctly. Even more, there is no way to tell if the spec is implemented correctly. Certainly when you actually have to ship something - you can't just spend 10 years polishing the implementation, testing millions of edge cases.
  - rbanffy 16y ago
    > a spec as complicated as this one is near impossible to implement 100% correctly. I gather being impossible to implement by third parties was a design requirement that was accomplished. What I find surprising is they got carried away and made a spec they couldn't implement either. > there is no way to tell if the spec is implemented correctly That's why a correct open reference implementation is a must. You can always say that any corner cases can be resolved according to the code. When you make your spec excessively complex you are just asking for trouble. This spec, BTW, exists for the sole reason as to legitimate a Microsoft format as a standard competing against ODF. It exists not to be implemented correctly, but to fragment the marketplace, preventing the standardization of something Microsoft cannot control and use as leverage.
    - roel_v 16y ago
      Apart from the tin foil hat, I love the doublethink in this. So basically you're saying 'screw the spec, make an implementation and make that the spec'?
      - rbanffy 16y ago
        Make a spec and make a reference implementation. Is there something wrong about it? What do you propose? Make a standard nobody can follow?
    - tzs 16y ago
      > This spec, BTW, exists for the sole reason as to legitimate a Microsoft format as a standard competing against ODF. It exists not to be implemented correctly, but to fragment the marketplace, preventing the standardization of something Microsoft cannot control and use as leverage. ODF was never a feasible alternative. First, Sun retained veto power via the threat of patents over ODF. Sun's patent grant for standardization was limited to a particular version of the standard and any future versions whose standardization Sun participated in. If they wanted to derail attempts to take ODF in a direction that they did not approve of, they could withdraw from the committee leaving the standard unprotected from their patents. Second, Sun made it clear that ODF was only going to have those features necessary to support Star Office. There were attempts to make it more general, so that it could be a universal format, but Sun squashed them.
- recoiledsnake 16y ago
  Sitting around writing specs while the competition runs away with the market does not really help in many situations. Guess this was even more true in the 80s when Office was originally developed. Choose between 'sloppy and shipped'(most software) vs. 'never ships because of targeting perfection' (GNU Hurd?).

"Come on. Implementing something you invented should be easy." Sounds like someone who has never implemented something more complicated than a hello world.

by roel_v 16y ago

"Come on. Implementing something you invented should be easy." Sounds like someone who has never implemented something more complicated than a hello world.

rbanffy 16y ago

Sounds like you don't know me. In the past 25+ years I got my share of bugs. But having a correct implementation of a spec is kind of the only proof you can have the spec is complete and implementable. Not having one is just sloppy.

roel_v 16y ago
Of course I don't know you, but with 25 years you should be in a position to recognize that a spec as complicated as this one is near impossible to implement 100% correctly. Even more, there is no way to tell if the spec is implemented correctly. Certainly when you actually have to ship something - you can't just spend 10 years polishing the implementation, testing millions of edge cases.
- rbanffy 16y ago
  > a spec as complicated as this one is near impossible to implement 100% correctly. I gather being impossible to implement by third parties was a design requirement that was accomplished. What I find surprising is they got carried away and made a spec they couldn't implement either. > there is no way to tell if the spec is implemented correctly That's why a correct open reference implementation is a must. You can always say that any corner cases can be resolved according to the code. When you make your spec excessively complex you are just asking for trouble. This spec, BTW, exists for the sole reason as to legitimate a Microsoft format as a standard competing against ODF. It exists not to be implemented correctly, but to fragment the marketplace, preventing the standardization of something Microsoft cannot control and use as leverage.
  - roel_v 16y ago
    Apart from the tin foil hat, I love the doublethink in this. So basically you're saying 'screw the spec, make an implementation and make that the spec'?
    - rbanffy 16y ago
      Make a spec and make a reference implementation. Is there something wrong about it? What do you propose? Make a standard nobody can follow?
  - tzs 16y ago
    > This spec, BTW, exists for the sole reason as to legitimate a Microsoft format as a standard competing against ODF. It exists not to be implemented correctly, but to fragment the marketplace, preventing the standardization of something Microsoft cannot control and use as leverage. ODF was never a feasible alternative. First, Sun retained veto power via the threat of patents over ODF. Sun's patent grant for standardization was limited to a particular version of the standard and any future versions whose standardization Sun participated in. If they wanted to derail attempts to take ODF in a direction that they did not approve of, they could withdraw from the committee leaving the standard unprotected from their patents. Second, Sun made it clear that ODF was only going to have those features necessary to support Star Office. There were attempts to make it more general, so that it could be a universal format, but Sun squashed them.
recoiledsnake 16y ago
Sitting around writing specs while the competition runs away with the market does not really help in many situations. Guess this was even more true in the 80s when Office was originally developed. Choose between 'sloppy and shipped'(most software) vs. 'never ships because of targeting perfection' (GNU Hurd?).