Auto Save

Sorry to be such a pain, but would it be possible to have the option to switch off AutoSave? I NEVER use this in any any program, and am a bit freaked out to find it as the only way to use Scrivener.

Any chance of Manual Save or No Auto Save as an option in the preferences or at least somewhere?


Vickie, something you can try. Think of Snapshots as your analog to saving. When you open up a part of your draft for editing, and you know you are going to be making some big changes, or just want to make sure you can always revert back to the way things were, simply press Cmd-5 and then start editing. Now, even though your current version is auto-saving, you can always return to the original, as if you had never saved in the interim – but with the added benefit of having saved your new stuff all along.

It takes a little mind twisting to use different words and slightly different concepts – but think of Cmd-5 is a super fast, integrated, and easy to manage “Save as…” function. Except you don’t have to worry about where all your version copies go, and what to name them, and sorting through old copies looking for that one revision change you missed.

Oh, and as an aside, you might have to get used to auto-save in a lot of ways. Tiger made auto-save a snap, and I’ve noticed a lot of applications are heading toward that model of never worrying about saving. Unfortunately, most of them do not have any way to address the concerns that Snapshotting addresses. Hopefully Scrivener will help other developers see the benefits of a built-in “save as” system.

Thanks for replying Amber V, I coudn’t have put it better myself. :slight_smile:

I thought long and hard about auto-saving; the majority of posters to the old forums wanted it, but I was never a big fan of it. Then I started playing with Mori, and I realised that I actually really liked autosaving. Remember that “Force Save” is available for peace of mind (although auto-save is pretty safe, the thing that bothered me about auto-save in most apps was that I was relying on the program to save my precious work; on occasions I wanted to be able to be able to force a save so that I knew for myself). So in Scrivener, you have the best of both worlds; if you forget to save your work for some reason, Scrivener will do it for you; if you want to be extra sure, you can force a save yourself.

The other reason I never liked auto-save was that sometimes I would just mess around with a document, then exit without saving and know that my old version was safe and the changes weren’t committed - and as Amber V has pointed out, in Scrivener, that is what Cmd-5 (snapshot) is for. Just get into the habit of hitting it before making any big changes. That camera shutter sound will grow to be very reassuring, I hope. :slight_smile:

Thanks for your helpful words, I’ve never understood or found a use for snapshots and I can see this would be helpful against the hated autosaving for individual documents.

But will this work for changing the ordering/groupings in the Binder, Corkboard and Outliner? For Splitting and Merging documents?

How can I experiment with a structure and then decide it’s not what I want?

There isn’t even a Save As, so that I can somehow get back to my original structure.

I suppose that as a laptop user, I never have any surprise losses when thunderstorms suddenly disconnect me from the national grid.

Shouldn’t Autosaving be a choice? Wouldn’t this be easier than implementing Undo for the Binder and Outliner and for Splitting documents?


Making auto-saving a choice would actually be a massive piece of work in the current set-up. It would mean re-impleming all of the old greyed-out-icon-because-it’s-unsaved stuff, completely rewriting all of the saving code, telling every view that displays a document icon to listen out for the document becoming unsaved/saved, and so on and so forth.

Snapshots is a very useful feature - don’t neglect it! Even if you don’t see it as an alternative to saving, it is there so you can save older versions of individual documents.

There is no Save As… but there is Backup To… This does much the same thing, in that it lets you save a copy of your whole project. The only difference is that whereas Save As… makes the newly created file active, Backup To… just saves the project somewhere else while you carry on using the original one.

Can I request that there is some form of Undo or Snapshot for the structure in the binder? And for splitting/merging Documents? and for Grouping and Ungrouping? And for Convert to File/Folder?


Yes, definitely use “Backup to…” (“Save Copy As…” in most other programs) before making any big structural changes and doing a lot of splitting and merging. There is no undo for most of that, and merging can be destructive to Snapshots. You retain the primary document’s Snapshots, but the secondary document(s) being merged with it lose their Snapshots without warning.

Perhaps something should be done about that?

It certainly should; it’s a bug! That was an oversight I will fix…

First, a little tale:

I was writing my novel in Scrivener. It was late at night, the children and the wife were in bed and I was tired but still wanted to finish that chapter before calling it a night. I felt safe that my work was being periodically saved, without my needing to do so manually; and I had a backup from the night before: I always backed up before shutting down.

Now, all I needed to help me wrap up that chapter was my brandy nightcap. So I went into the living room to fetch it and gladly took the opportunity to stretch my legs a little: it had been a good few hours with me stuck to that chair. Then I headed back to the study, glass in hand. I was already anticipating being tucked away in bed for a well deserved night’s sleep, another hard fought chapter under the belt. Yes, the book was starting to shape up rather nicely…

When I got to the door I found it odd that it was wide open. I always left it closed because I didn’t want the cat to sneak in and make a bed out of some (very few) refused manuscripts I had lying around.

I came in and stopped near the door. Sitting in my chair, in front of the computer, was Mary, my six year old. The lone lamp in front of her cut out a black silhouette of child and chair. It just happened that sometimes my daughter had nightmares and wandered out of her room, looking for her parents to comfort her.

I put down my drink as a sense of impending doom started to work on my throat. I approached the desk and gently placed my hand on her shoulder, trying not to shake.

“So, what is my little girl doing out of bed this late at night? Not having another one of those nasty dreams, was she?”, I said, noticing, with growing horror, the short distance between her small fingers and the keyboard.

“No, not this time, Dad!”, came the prompt reply. She turned her face to me, giggling excitedly. I tried to caress her cheek, but my hands failed me; I knew what was coming. “This time it was a nice dream. I woke up and couldn’t go back to sleep. I had to tell it to Mom or you. I didn’t find anyone but came in here and saw the computer on. And you’ve told me to always write down any really good ideas that I have. So I thought my dream was actually a very good idea and wrote it down. Look!”, ordered she, pointing at the screen.

I obeyed, finally facing what I hadn’t dared to since I had entered the room. The display was almost entirely blank, save for a few lines in the centre. I could see by its title that this was the document I was writing on when I went down. The vanishing of the scroll bar further informed me that the dozen or so pages I had spent all morning, afternoon and evening writing were gone. All that text had been deleted, obviously to leave a clean sheet for this alternative literary work, authored by my six year old daughter. And – of course – due to the auto-saving system, it had been thoroughly overwritten on disk, relegated to the utterest digital nothingness.

The world twisted round. I sat in another chair, head in hands, eyes closed. I was incapable of coherent thought. Mary was talking but I couldn’t understand her; I just heard the cheerful sound of her words. Eons seemed to pass.

Then, all of a sudden, I stumbled out of my stupor. Undo! Control zed! Undo, Undo, Undo!!! How could I not have thought of that most intuitive of actions??? All I had to do was press the magic key combination and every single evil of this world would be undone back to the Garden of Eden… or close enough for me, at any rate.

I jumped to my feet and set my eyes on that empty screen once more; not in fear and defeat this time, but with a heart full of valour and resolve.

In two steps I was beside the desk again. Mary was now sitting back in the work chair, her short legs fully stretched on the seat, and she kept on chattering about something or other, unencumbered by my prolonged silence. I still had to repossess my commandeering post before I could correct the harm that had been done and set all things to rights. I turned to her with what I hoped to be an air of authority.

But she just went on: “…so it was a really, really brilliant dream, Dad. And your ideas are brilliant too. You do that too, don’t you, Dad? You write down your ideas too. After writing my dream, I clicked on those little notes on the left and read them. I liked them but couldn’t understand some of them. I think they are too complicated. So I came back to mine.” She went to the front of the chair, crossed her legs, planted her elbows on the desk and held her pretty face in her hands. “Don’t you like it better than yours, Dad?”, she asked, grinning at me.

This time I just froze. I kept staring at her, my organism unable to function in a world suddenly turned so unfair. My mask of sternness went unnoticed by the underage critic in front of me; I tried to swap it for something more in tune with my inner collapse, but to no avail.

Mary had changed documents. She had deleted all the text in the document that was selected, written her own little thing and then she had actually gone and changed documents before returning to the first one.

That meant no undo-type shenanigans would be taking place tonight. No siree.

I returned to the doom-chair, spent a few more eons there holding my head in my hands and then I noticed that I was not actually feeling anything much anymore, except sleep. I stood up, took the untouched brandy and gulped it down. Then I picked my daughter up — by that time she was also very sleepy — took her to bed, tucked her in and kissed her goodnight.

“Tomorrow I’ll tell her that she shouldn’t touch the computer without asking first”, I thought to myself, futilely. “Of course, she’ll still do as she pleases”.

As I was leaving the bedroom, she said quietly, peeping from between the sheets and almost dozing off:

“Dad, don’t you think I’m right? I mean, what I wrote in the computer?”

“Sure, honey; quite right”, I replied, not certain of what to say. “Now, see if you can have some more of those brilliant dreams”.

I was now almost asleep myself, dreaming that everything was fine with my novel and I had finally completed that difficult last chapter.

But before going to bed and turning that into virtuality for a few hours, there was still one final hurdle to be crossed. I returned to the study: I still had to shut down that despicable machine. Somehow, that night I didn’t think I would be making a back up of anything…

I entered and went straight to the desk. “This may be the end of the line for you, my friend!”, I childishly thought as I bent down to grab the mouse and command the thing to go to sleep, as we humans do. “Only difference is you may not wake up in the morning”, the little devil on my shoulder added.

The moment my finger was about to press “Shutdown”, my eye, very close to the monitor screen, could not avoid reading, for the first time, the beginning of Mary’s brief composition. And then the rest of it. I sat.

[size=85]I had a dream and it was Mom’s birthday. So I drew her a picture of the four of us. Mom, Dad, me and Mr. Sleepy, the cat. Oh, and Timmy too. We were all around the table waiting for Mom to blow the cake’s candles (in the picture). Then (in my dream) I took the picture to Dad’s study and made a copy of it in that machine Dad has there. Then I put that copy on Mom’s side of their bed as a surprise for her. Then I decided to put some more things in the picture, like flowers and clowns and hippopotamuses but I didn’t like that and threw it away. Just the four (five) of us is better. So I think it was a brilliant idea to have made that copy! (In my dream.)[/size]

Now I knew what she meant. And she was right. A safe copy is always a brilliant idea.

Well, I too had a copy of my own overwritten work.

In my dreams.

Second, my opinion:

Much as I like Scrivener – and I won’t enumerate here all its qualities, because we all know and love them (I always think that, within a group that loves something and is devoted to it, it is the bad aspects that are worth discussing, so that the thing can be perfected) –, I must say that, as long as Scrivener does not allow the user to either turn off auto-saving or redirect it to a folder which is not the original project folder, it is flawed. More so with the current severe limitations of the Undo function, which, in the context of a single document, is gone as soon as one leaves that document, and doesn’t work on the project as whole – undoing and redoing the creation, deletion and moving of documents, etc.

Entertaining tale. :slight_smile: This thread is five and a half years old by the way. :wink:

I would argue that a broken undo function is a foremost issue here, and shouldn’t be conflated with any operational philosophy of the software that might be hampered by bugs in it. The problem is the undo, changing the underlying saving mechanism to get around the problem in undo isn’t the way to fix the real problem: undo. And that is a serious problem. You should be able to undo a single edit after thousands of other edits in other files.

As for the mechanism, this behaviour is not at all unusual for programs of this type. You are working in a system of files, not a single file. In this way, it is more akin to a database than a word processor. There are multiple things that need to be saved, not just the individual items within the system. The system itself needs to be saved. If you choose to show hide labels on the corkboard, that is a systemic change and saving it is not thematically or philosophically bound to the status of any one outline node’s contents (what we call in more friendly terms, “documents”) in the system. Ctrl-S does not save one single piece of the project, but dozens of pieces. And that is just the stuff you can see. Vital background processes are keeping various coherent parts of this system synchronised together. Yes, you can crank the auto-save idle counter up really high, but when you do that you’re basically crippling the programs intended safety nets. Sometimes it is necessary (especially over file servers), but as a way to try and force the program to act like Word? It’s living on the edge.

And for anyone used to saving as a way of recording moments of “This is good, let’s mark it” working styles, I again recommend Ctrl-5. It even kind of looks like Ctrl-S. :slight_smile: Ctrl-5 is your save button. It’s your, “I like this, but I don’t know where I’m going…” button. It’s your, “I’m done for the day” button; that “Now about that brandy…” button. And it’s better than Ctrl-S because if you change your mind a week later you have all of those moments recorded for you in the snapshots sidebar. It means you don’t have to worry about which “MyBigBook-23-bak-SecondAttempt-81.docx” has that one bit of prose you fitfully deleted after too much wine. It’s all right where it should be, in one cohesive system that is your work in progress. Meanwhile the system itself, and all of its data are reasonably protected from crashes, power outages, and any number of unforeseen or accidental calamities.

It’s true, in this case the safety net failed you. And for that you have my sympathy. You might have some bits in your automatic backup folder. Definitely check that out, but if the lost stuff was written in a single session then probably not. Nothing is perfect, and there are always a dozen and a hundred ways to break any sort of protection (and auto-save is only necessary for a fraction of them to happen). Six year olds are unparalleled experts at it, but fitting the common solution (which has saved who knows how many author’s books at this point in time) to exceptional cases isn’t always the best route to take.

If you want, you can also change your automatic backups to trigger when you press Ctrl-S, and if you do that I would recommend bumping up the number of saved backups to something way higher than five. Now whenever you hit Ctrl-S, the whole project is tucked away for safekeeping. In tonight’s scenario, you’d be able to go in and pull out and get right back to work, all without the change of habit in Ctrl-5 and thinking in pieces instead of totalities. So there are definitely options for you.

A chilling story. I believe T. E. Lawrence left the manuscript to Seven Pillars of Wisdom at a railway station and had to rewrite the whole thing from scratch, and I also heard somewhere that Isaac Newton’s dog ate some of his work. I’d shoot myself if I lost a stack of work. Still, at least you have the six year old, who is also an important creation :wink:


In defense of the auto-save…

I’ve had my laptop crash on numerous occasions whilst writing. Without the wonder of auto-save, I could have lost hours of work - as indeed I did several times while working in other software programs.

The main selling point for me is not the fact that it saves me having to recreate my work, it’s that it saves me from having to try and figure out what needs to be recreated. The nature of Scrivener is such that in a single working session I’ll have been working on a dozen different sections from as many documents from all over my story, and in my research and planning sections too.

So auto-save protects me from the nonhuman errors. The human ones are best protected by ++, or whatever it’s equivalent is on the Mac.

Just for the record, two points:

  1. I actually like auto-saving. I just would prefer it redirectioned to another location, like most writing software does, even on the Mac; I’m thinking about Storymill, Storyist, Jer`s Novel Writer, etc. Only as a second best, would I want to be allowed to disable it.

  2. That was just fiction. In fact and regretfully, I’m childless :frowning:

That would be one of the human errors I was talking about :smiley:

I’m not sure what you mean by redirecting the save somewhere else. If you mean to create a new file each time that would create a massive burden on your harddrive. If you mean something else, how would that differ from what is already offered by the rolling back-up system and the snapshot functionality.

Remember: None of the content protection systems (Auto-save, Save, Save-As, Snapshots, Snapshots-with-title, and automatic backups) exist for your benefit. They are there to minimise the number of support requests when things go wrong. :smiley:

I think they mean the “recovery file” technique that Word and a lot of other Save-As-You-Go programs work. It’s a compromise to direct autosave, where all real-time changes are folded into a ~tmp file somewhere, rather than the file the user the clicked on. If the computer crashes (or Word, as would be more likely the case) then the next time you load it detects the recovery file and asks if you wish to restore your session, which has a small risk of not working. The key thing in all of this, and the example programs mentioned, is the word file. It is not “impossible” to do something like this with a format that uses tons of little files, but it’s way more complicated, and the risks of a recovery not working are statistically higher. It’s worth mentioning that these other programs don’t work too well if you (and indeed you can’t at all in most) import 3Gb of PDF files. This is why I used the word “system” above. Scrivener is more than just a host for large files, it is a host for systems of files. Again it’s not impossible (though as Keith pointed out long ago and far above, at this point it would be a huge undertaking—especially for something most people like as-is) but it is way more complicated.

Given that there is very little functional difference between snapshots and point-saving for individual sections, and auto-backups bound to Ctrl/Cmd-S and bulk saving, with most of the differences being to the advantage of the current system since you have a history of saves rather than one set of bits being updated every time you C-5/C-S that just seems like a lot of added burden and complexity for no net gain.

I would also point out that the MS Word implementation is quite far from ideal. It always seems to offer me a number of restore files when I don’t need them, and never has anything useful when there is a crash.

The best recovery file implementation I’ve seen is Vim’s, but that is a plain-text single-file editor, so dramatically less complicated than most of the above. The closest example to Scrivener on the Mac is, by the way, none of the above, but Ulysses. That one also uses a many-small-files format, and it also employs aggressive autosave and rolling backup. The main user side difference is its open file vs. closed file architecture. Internally when you open an item a separate set of files is created for it until you specifically save that tab. This system wouldn’t work as well in Scrivener, which works in much smaller pieces and has such things as Scrivenings mode for editing dozens of files simultaneously. It’s also worth noting that Ulysses has no per-item snapshot feature. The bi-modal open/close model is the line of protection against unwanted edits. I suppose the defunct Copywrite would be another. It also used the Snapshot+Autosave method. Anyway, my point is you can’t really compare Scrivener to stuff like Word. It’s really more like Lightroom or Aperture, which also describe “systems” instead of “files”.

I knew I shouldn’t have brought up the issue again. I had already in another thread and the quasi-flaming kept on relentlessly, no matter how plainly I explained what I was asking for. Of course, this is what happens when religion is being discussed. And yes, unfortunately, autosaving is most definitely a religious issue in the Scrivener community.

But here we go again. A detailed explanation of what I mean follows.

The main thing here is that manual saving and autosaving, which now are joined in one process, would be completely independent processes, although very similar, almost equal. Right now, saving is triggered by one of two different events: either the user selecting the save command or the passage of a predefined amount of time since the last save; but the response to either of those two triggers is the same, namely saving to the project folder and resetting the “changed” flags. What I propose, then, is to separate (duplicate) these last steps of the process also.

An example:

A new project is created. A standard project folder is created for it, somewhere, named “[project_name].scriv”. Inside it a few files are created, only those needed for the management of the project, as usual. But an autosave folder is also created at this time in the same parent folder as the project folder, named “[project_name].scras” (“as” as in “autosave”) and containing the exact same files – just a copy of initial project folder.

The user now creates a few documents, but does not use the save command yet; that means the project folder is exactly as it was when the project was created: the user has not yet decided he likes what he has written so far. As for autosaving, it has not yet kicked in (it’s set, say, to only do its thing every 15 minutes). Now the user writes a bit more and decides he likes what he now has and saves the project, for the first time. That means Scrivener goes through all the structure of the project, picks up whatever has been changed since the last manual save (that could mean created, edited, deleted or moved – this first time that means everything) and saves all that in the proper places in the project folder, which now contains some document files also.

Are we clear so far?

Let’s say the user keeps working and ads a little more content and then he thinks: “no, this is not a brilliant idea; let’s go back to the saved version”. So he simply uses the open command and reloads what he last saved (a “Restore” or “Reopen” command would be even better); in a few seconds, he’s good to try again, with no need to look for and mess with backups. (Of course, reloading would only work if one is allowed to close without saving.)

Still with me? Good! Now comes the “exciting” part…

The user writes yet another scene or chapter or whatever – a document in any case, which, for the time being, exists only in the computer memory. And NOW who comes knocking at the door? Why, it’s our good old friend Mr. Autosave! But, lo, he has changed his ways: instead of just writing, whenever he pleases, all over the user’s work, he now politely does so to another folder, just in case the user doesn’t want to keep his last changes (it happens to us all, you know?)

So just what does Polite Autosave V.2 do, exactly? Simple: It goes through the exact same process as a manual save, the only two differences being that, one, instead of picking what has changed since the last manual save, it picks what has changed since the last autosave, and, two, instead of saving that to the project folder, it saves it to the autosave folder, leaving the project folder well alone! So, in our example, since this is also autosave’s first occurrence, it also saves everything, including that last mentioned document, which, however, has not yet been manually saved to the project folder and indeed might not ever be.

It cannot get any simpler than that.

Of course, this means there is the need to maintain two sets of flags, and not just one, in order to keep track, not only of what has changed since the last manual/auto save, as it is now, but of what has changed in two different contexts – since the last manual save, on one hand, and since the last autosave, on the other. But, as the logic is exactly the same, how hard could that be to implement?

Put another way, Scrivener would be doing exactly the same thing it does now, just double: there would be two folders to save to, two sets of flags to indicate changed “things” and two types of event to trigger saving – a time event (autosaving) and a user command event (manual saving).

To state (repeat) the obvious but to be perfectly thorough in my explanation: One is now doubly safe: the version of the project kept in the main project folder contains what the user means to keep so far, without any experimental work which can be discarded at any time by reloading; the version kept in the autosave folder contains those experimental changes too, for the event of a crash – and, differently from the example above, autosave might be set to every two seconds or even one.

So, apart form being heretic, what else is there not to like about such an approach? I know, I know: “The way Scrivener works pleases most users.” Good. And how about pleasing all users (in this respect)? Because, of course, there should be a setting with three options: 1 – use autosave the exact same way it works currently (default option); 2 – autosave to a parallel autosave folder, as explained above (maybe it should also be possible to select the location of that folder); 3 – don’t autosave. And then everybody could have it their way.

What else? It uses the double of the disk space? So what, disk space is cheap these days and just text uses hardly any space at all. Besides, one would not be using snapshots and backups so much – which, by the way, have their uses and I wouldn’t like to be without.

It takes time to do a manual save, on top of the time taken by autosaves? Come on, autosave, if it is set to work every two seconds, the default, would take much more time than the comparatively infrequent manual saves (and still go unnoticed).

So please forget Windows, forget Word, forget “recovery file”, forget ~tmp files and what-not. Just read what I’ve written and don’t put words in my mouth. There would be no “massive burden on your hard drive” at all, just the doubling of the disk space, at most (which would occur too if backups were to be used). If there are “3 Gb of PDF files” within the project and duplicating them is inconvenient, options 1 or 3 above can be used (but same comment as before concerning backups). The same goes for the “human errors” mentioned: if autosave is disabled using option 3 above and then things go wrong, well, one is entitled to make decisions and to err; and the author of the program cannot be blamed for allowing users to make choices (though I insist there should be no reason to disable autosaving). And, as explained profusely, there would be very little “added burden and complexity”, quite the opposite: going back to a good version of one’s work would be much easier than it is now and constitute a substantial “net gain”. And, by the way of complexity, what really is confusing right now is the existence of a save command with almost no purpose, since the software does that already, automatically, every two seconds.

If you want to, tell me that the author has no intention to alter the philosophy of the program, ever. I can respect that. Or tell me that’s the way it is and it won’t change, just because. I won’t argue with that – no use. Tell me most users like it this way. Well, that’s no reason not to improve it. But please read what I wrote and don’t reply to what I didn’t write.

Sorry, it sounds like you might have our tone all wrong here. This forum, and indeed many forums other than ours, tend to involve a lot of cross-feed. Even if someone starts a thread, from that point on it is no longer “their thread” and multiple people can come in and add to it in whatever way they please—and we’re probably a little less strict on that score than most forums, as good and varied discussion can often breed parallel insights and improvements that might not have been arrived at if everyone stuck to their own threads and very strictly only discussed precisely what the original poster spoke of and nothing more. So no personal affront was meant by any offhand rumination, some even amongst ourselves not meant to be directly responding to you in the first place—certainly not “quasi-flaming”. Flaming usually means personal attacks or angry arguments that have digressed beyond their original topic.

To recap:

  1. The manner in which the disk state and the Scrivener software works is so intrinsicly bound together that it would take a long rewrite of vast amounts of code to do what you are asking, and at this point in history, to two applications instead of just one.
  2. Meanwhile snapshots do the same thing as Saving but on a per-outline-piece basis. They are functionally identical to Save except in that they keep a record of each time you saved for future use.
  3. And further, something that didn’t exist six years ago: setting your automatic backup feature to create a backup whenever you Save (giving it a solid purpose) gives you a superior but similar level of protection as your described 2-way system. It’s five way, and potentially more if you want. By default it is tucked away out of sight, but you can easily change that as well. You can even turn off the .zip option so that stepping back to a clean slate is as easy as pie.

There is the future to think about as well. The industry is tending toward auto-save and data persistence as well as cloud storage and collaboration, even for things as simple as a text editor. Scrivener has been at the front of that tidal movement for years, and while it doesn’t yet offer real-time collaboration it is something we are interested, in a long-term sense, in doing. Taking Scrivener back to way things are evolving from would remove that possibility entirely and very likely ultimately end up with it using an outdated model that future generations find anachronistic or difficult to use in a modern context.

Offloading permanent or semi-permanent (depending upon preference) save states outside of the live system via Snapshots and Automatic Backups affords you greater levels of protection and keeps the road open to further modernisation and better multi-user possibilities in the future. It’s also, as pointed out, very much the standard way of doing things amongst programs like Scrivener.

With all due respect, I find that hard to believe, since, as I explained, all that would be required would be the duplication of logic, objects and disk data which already exist.

As we both seem to agree, snapshots provide security (from potentialy unwanted changes) only at a document level. And as I stated in the previous post, I (and probably not only I) want to be able to jump all over the project, making global changes, and still have that security. Snapshots are good, but not for this purpose.

I don’t agree. As I said in the previous post, both snaphots and backups are very good features I wouldn’t want to be without, but not for this simple purpose. What I want is, without leaving Scrivener, to be able to just close the document, confirm that no changes are to be saved and then reopen (or just select “Reopen”, as I sugested). Simple and intuitive. I don’t want to have to close Scrivener, go to Windows Explorer, look for the backups folder, pick the last one up, unzip it if necessary, put it where the project folder is, delete the project folder, rename the backup to the name the project folder had, and finally open that. (A similar process would have to be followed in a Mac, I suppose.) All this – each time I want to discard changes. I think anyone would prefer the former over the latter, if given the choice.

Change for the sake of change is called “mystifying”. As we say in my country, “changing – yes, but only for the better”.

“Greater levels of protection” than what? Than autosaving to a parallel folder? Can’t see why.

No, it isn’t, if you are talking about Storyist, Storymill, Jer`s Novel Writer or such. They either don’t autosave or allow the user to disable autosaving. Check your own links page.