Mainly, completion of the core design, and the parity between interface and functionality. When we had to delay the the launch a few months ago, the main problem was that a very significant majority of the GUI in the compiler wasn’t wired up yet. Obviously, we couldn’t release that. 
The other problem is that what was wired up wasn’t yet working to core design specifications, broadly speaking. We’re talking stuff like having the ebook output produce a valid ebook that reacts to the settings you provide, or being able to compile footnotes at all, etc. Again, not problems you can have in software you sell.
Of course we knew we’d get no end of people piling on and telling us how bad we are as a result, and we know we’re going to continue getting that until the end—that’s the choice we made, and there will never be a shortage of people who prefer that mode of discourse. It should be obvious to anyone with an ethical bone in their body though, which choice is the correct one. Heck, even if you aren’t much on ethics, arguably releasing prototype code and selling it could actually lead to legal problems, not just people eager to lodge complaints about development speeds or what have you.
Indeed that is almost always necessary to some degree. Given the delays, we’ll probably have to have more of those kinds of things that I personally would like to see, but that’s just how it goes. It’s going to be a long time before it’s “done” (in the foggy sense that software never is). The goal for 3.0 though, is being able to do what’s printed on the tin. That’s just simple, logical necessity, and I don’t think many would seriously debate that.
I think you’ve highlighted one of the problems, in fact. Scrivener is far larger than most people realise, it is an extremely ambitious project. Look at its user manual, and now consider that the one you see when hitting F1 has nearly 250 pages of redacted text in it on account of things that weren’t working at the time (excepting stuff that will always be Mac specific, like support for the programmable OLED touch strip above the keyboard on their laptops). For those who never encounter those areas in the missing pages, they might be utterly baffled as to what we’re going on about—after all, it still had 600 pages of user manual, and that’s an awful lot of software, one can use it for years without straying outside of that realm.
But here’s a simple one for you:
- Create a new blank project and paste several paragraphs of lorem ipsum into the starter file.
- Click into the second paragraph and set it to “Block Quote”.
- Load File ▸ Compile….
- Select the “Modern” compile format in the left sidebar, and assign the stock section type in use to the “Section Text” layout, in the middle column.
- Compile to RTF, and open in any word processor.
If you’re a biographer, well, you can see how this might cause problems! In this case the font family didn’t override, though it should have while leaving the custom indenting intact (and not mangled), but that’s just one problem on the docket. The full integration between the stylesheet system and how it passes through and is optionally transformed by the compiler is very complex, particularly where we get into the permutations of combinations of these features. And these solutions sometimes require custom outfitting for each output type. What styles do to an ePub file (generating CSS classes and automatically generated attributes, along with the ability to code your own—all only recently made possible) is totally different than what they should do to RTF files, or Markdown files.
Indeed that is true and expected. But one thing we’ve done a good job of so far I think is in stability. What is there is pretty solid. For a program this complex and still under heavy construction you would very rightly expect crashes and data loss bugs now and then—but look around, you won’t find many. I test with it daily and extensively, and it it is very rare I hit a serious issue like that—and my definition of data loss is very broad: I include stuff like the colour of your annotations getting lost, since one might use blue, red and green comments for strong and important semantic purposes, meant to communicate to future versions of themselves for years to come.
But keep those backups—and keep doing it when it goes gold as well. I don’t use any piece of software without backups upon backups upon backups. It’s not just the software I’m protecting myself from, but the hardware, the operating system, the weather and even my own foolishness (perhaps primarily so).
Anyway, thanks for stopping by the forum.
