3 ms·

Third party code would be mostly a problem for XP and older with things like HyperTerminal and Pinball. Even for MS they were a problem: https://blogs.msdn.micr

by yuhong 9y ago

Third party code would be mostly a problem for XP and older with things like HyperTerminal and Pinball. Even for MS they were a problem: https://blogs.msdn.microsoft.com/oldnewthing/20080225-00/?p=23343 https://blogs.msdn.microsoft.com/oldnewthing/20080225-00/?p=...

pgeorgi 9y ago
With OpenSolaris, there was some l10n code in libc that they weren't able to get clearance for some reason. Windows has ~30 years of legacy. There are probably some bits of Adobe code in font rendering - although at least in this case, Adobe still exists so at least it's possible to call somebody to try to negotiate. If there's some significant piece of code in some ancient video codec (to pick some silly example) that came from a third party in the late 90s who then went under, it'll be a mess to track down if there's even anybody who needs to be called. Now, such a codec is well isolated: "Free Windows doesn't support the AncientCrapware4Windows codec. Does anybody care?", but what if some fundamental parts of the shell was licensed back then and dragged forward all those years? It's the long tail that hurts and just checking _if_ there's anything to do is lots of legal work that scales by lines of code (and OpenSolaris was tiny compared to Windows). For a product that (as the article claims) approaches a retail value of $0, that's a steep investment.
- yuhong 9y ago
  Yea, both are true. The Adobe code is called ATMFD (which last time I checked don't even have PDB symbols on the MS symbol server), and there is also the Indeo codec that was already partially disabled years ago due to even MS not having been able to patch it (indeed, MS created a special telemetry stub DLL that calls the real one to log how often it is used). I don't believe either is required for Windows to run though and they are also well-isolated components.
  - pgeorgi 9y ago
    It seems I wasn't clear enough on that, but either were just examples. Another, OpenSolaris' libc components, is more intricate. Unless Microsoft's lawyers can make sure that something like this (deeply integrated component with unclear terms with regard to relicensing) does _not_ exist, publishing the Windows source code is a rather expensive task.

by yuhong 9y ago

pgeorgi 9y ago

With OpenSolaris, there was some l10n code in libc that they weren't able to get clearance for some reason. Windows has ~30 years of legacy. There are probably some bits of Adobe code in font rendering - although at least in this case, Adobe still exists so at least it's possible to call somebody to try to negotiate. If there's some significant piece of code in some ancient video codec (to pick some silly example) that came from a third party in the late 90s who then went under, it'll be a mess to track down if there's even anybody who needs to be called. Now, such a codec is well isolated: "Free Windows doesn't support the AncientCrapware4Windows codec. Does anybody care?", but what if some fundamental parts of the shell was licensed back then and dragged forward all those years? It's the long tail that hurts and just checking _if_ there's anything to do is lots of legal work that scales by lines of code (and OpenSolaris was tiny compared to Windows). For a product that (as the article claims) approaches a retail value of $0, that's a steep investment.

yuhong 9y ago
Yea, both are true. The Adobe code is called ATMFD (which last time I checked don't even have PDB symbols on the MS symbol server), and there is also the Indeo codec that was already partially disabled years ago due to even MS not having been able to patch it (indeed, MS created a special telemetry stub DLL that calls the real one to log how often it is used). I don't believe either is required for Windows to run though and they are also well-isolated components.
- pgeorgi 9y ago
  It seems I wasn't clear enough on that, but either were just examples. Another, OpenSolaris' libc components, is more intricate. Unless Microsoft's lawyers can make sure that something like this (deeply integrated component with unclear terms with regard to relicensing) does _not_ exist, publishing the Windows source code is a rather expensive task.