3 ms·

I don't entirely disagree with your perspective, but evaluating what counts as "a complete and stable set of core features" gets very tricky. For example, you

I don't entirely disagree with your perspective, but evaluating what counts as "a complete and stable set of core features" gets very tricky. For example, you said > Let's look at your example: VIM. According to criterion A.) it is finished - because it has a complete and stable set of core features, which haven't changed much in many versions (and many years). From my point of view, that's not an accurate description of Vim. Vim 8 (released in late 2016) added several new features. Most notably: > Vim can now exchange messages with other processes in the background. This makes it possible to have servers do work and send back the results to Vim. See |channel-demo| for an example, this shows communicating with a Python server. (Source: https://raw.githubusercontent.com/vim/vim/master/runtime/doc/version8.txt https://raw.githubusercontent.com/vim/vim/master/runtime/doc... ) I personally view this as a major new core feature – one that enabled a whole new class of plugins, including many Language Server Protocol plugins. For the ways I use Vim, that's a core feature. Maybe it's not a core feature for you, but that's exactly my point: what is and isn't a core feature depends entirely on an individual's usecase. > Let's look a different example - I suggest looking at a piece of imaginary vaporware. Maybe a few design documents exist - but not a single line of code has ever been written.… According to criterion B) this software does count as finished - because no new features are being added. I agree that criterion B) would characterize this as "finished", but I disagree that it would count as "finished _software_". Maybe it's a finished design document, but if no software was produced, then it's not finished software. Incidentally, it's obviously possible for software to be both 100% finished and not very useful. For example, the K&R C "Hello, world" program is finished – it doesn't need any updates and perfectly fulfills its specification. But that doesn't mean I'm planning to use it daily!

hooby 6y ago
I fully agree - that can be very tricky indeed. But I think your response on VIM does add another great point to this discussion: nobody would have considered VIM incomplete for not having JSON support, before JSON even existed. In theory, VIM could have been complete AND finished earlier without JSON-support, back in the day before JSON. But then JSON happened - and suddenly VIM suddenly no longer had a complete set of core features. And not being complete means not being finished - since being complete is pre-requisite for being finished. And doesn't that mean, that a piece of software which does NOT get new features, can't stay "finished" when things change and new requirements emerge? Doesn't a piece of software have to get new features so it can stay (close to) finished - and doesn't revert back to being incomplete? The only way a piece of software could ever stay "finished" without getting new features is, if it was used for something that no longer sees any change happen or any new requirements emerge - it basically has to be a completely non-evolving field of work with static, unchanging requirements. But any active, evolving field of work will always see changes and new stuff and subsequently new requirements will emerge. Software has to react to those changes in requirements.
- codesections 6y ago
  > And not being complete means not being finished - since being complete is pre-requisite for being finished. And doesn't that mean, that a piece of software which does NOT get new features, can't stay "finished" when things change and new requirements emerge? Doesn't a piece of software have to get new features so it can stay (close to) finished - and doesn't revert back to being incomplete? That's a really interesting question. I guess some of it comes down to whose perspective we're looking at the software from. I'd tend to look at it from the author/mountaineer's point of view: If Bram Moolenaar had declared that Vim was complete with version 7 and would never add major features like async/JSON support, I'd probably say that the program was "finished", even if it lacked certain features that I'd like it to have. (Those missing features might cause me to use a different piece of software, but I don't think they'd give me cause to complain that the hypothetical Vim was "incomplete") (And, indeed, there was a period ~5 years ago where it looked like Moolenaar might go that route, especially with respect to async support. My understanding is that his initial unwillingness to add async support/other new features was a big part of the impetus for Neovim.) > The only way a piece of software could ever stay "finished" without getting new features is, if it was used for something that no longer sees any change happen or any new requirements emerge - it basically has to be a completely non-evolving field of work with static, unchanging requirements. I'm not sure that the entire _field_ needs to be static. For example, Quake strikes me as a finished piece of software – that doesn't mean that video games (or even first person shooters) aren't evolving. It just means that Quake carved out a narrow section of that field and finished delivering on its design goals within that narrow slice.

hooby 6y ago

I fully agree - that can be very tricky indeed. But I think your response on VIM does add another great point to this discussion: nobody would have considered VIM incomplete for not having JSON support, before JSON even existed. In theory, VIM could have been complete AND finished earlier without JSON-support, back in the day before JSON. But then JSON happened - and suddenly VIM suddenly no longer had a complete set of core features. And not being complete means not being finished - since being complete is pre-requisite for being finished. And doesn't that mean, that a piece of software which does NOT get new features, can't stay "finished" when things change and new requirements emerge? Doesn't a piece of software have to get new features so it can stay (close to) finished - and doesn't revert back to being incomplete? The only way a piece of software could ever stay "finished" without getting new features is, if it was used for something that no longer sees any change happen or any new requirements emerge - it basically has to be a completely non-evolving field of work with static, unchanging requirements. But any active, evolving field of work will always see changes and new stuff and subsequently new requirements will emerge. Software has to react to those changes in requirements.

codesections 6y ago
> And not being complete means not being finished - since being complete is pre-requisite for being finished. And doesn't that mean, that a piece of software which does NOT get new features, can't stay "finished" when things change and new requirements emerge? Doesn't a piece of software have to get new features so it can stay (close to) finished - and doesn't revert back to being incomplete? That's a really interesting question. I guess some of it comes down to whose perspective we're looking at the software from. I'd tend to look at it from the author/mountaineer's point of view: If Bram Moolenaar had declared that Vim was complete with version 7 and would never add major features like async/JSON support, I'd probably say that the program was "finished", even if it lacked certain features that I'd like it to have. (Those missing features might cause me to use a different piece of software, but I don't think they'd give me cause to complain that the hypothetical Vim was "incomplete") (And, indeed, there was a period ~5 years ago where it looked like Moolenaar might go that route, especially with respect to async support. My understanding is that his initial unwillingness to add async support/other new features was a big part of the impetus for Neovim.) > The only way a piece of software could ever stay "finished" without getting new features is, if it was used for something that no longer sees any change happen or any new requirements emerge - it basically has to be a completely non-evolving field of work with static, unchanging requirements. I'm not sure that the entire _field_ needs to be static. For example, Quake strikes me as a finished piece of software – that doesn't mean that video games (or even first person shooters) aren't evolving. It just means that Quake carved out a narrow section of that field and finished delivering on its design goals within that narrow slice.