This is about the most recent version of LibreOffice on Windows 10. I can’t speak for other versions.
My daughter worked hard on her social studies essay. I type things in for her because she’s a really bad typist, but she tells me what to write… but I didn’t remember to manually save her social studies essay yesterday, and for some reason the ThinkPad rebooted, LibreOffice crashed and we lost the whole thing… because autosave was not automatically on when I installed it.
No, recovery didn’t work. We just got a blank file.
I rewrote it for her based on the information we had and what I remembered and tried to make it sound like what a 13-year-old would write because it was basically my fault and she did do the work. I did have her sit with me as I wrote it in case she didn’t like something I wrote, but it was sort of cheating. I’m okay with that cheating since I know she worked hard on it.
First, though, I went into the settings and turned on autosave.
I like LibreOffice, but why the hell is that not on automatically? Honestly, I don’t really understand why someone wouldn’t want their documents autosaved, but I’m pretty sure most people would want that.
This isn’t fucking 1993. I shouldn’t have to remember to save a document anymore and it shouldn’t be lost forever because of it.
Like I said, I like LibreOffice. I don’t really want to trust documents to Microsoft or Google. But this was really annoying.
Why do you trust ctrl-s though? You are a software engineer, you know that a bug in the piece of code that saves the document would affect both calls, regardless of whether its invoked by a timer or by the end user pressing keys, right?
I mean we have all been bitten by op’s problem In the past but it was exactly the same issue, autosave not enabled (most likely didn’t exist) what’s with all these, I don’t trust software to do it’s job so I do things by hand?
Particularly from software developers or other technical users. Found a bug in a piece of software, report it, you don’t need to change your behaviour for the next 20 years and tell everyone anecdotes about you still don’t trust a regression.
Every single one of us has been bitten by auto save that didn’t work. I’ve personally lost hours worth of code to auto save glitches and poorly timed save runs. People don’t trust it because in the past it has had and/or caused problems with their workflow.
Ctrl+S is a manual confirmation that I saved it, and is a step taken before running any code, especially through a terminal in an IDE where if the auto save hasn’t kicked in will mean the changes aren’t reflected.
There’s a couple things… First, it’s a habit to be constantly pressing CTRL+S. I’ve been doing it for many years, I’ll continue to do it probably until I stop using a keyboard. It’s such an easy keystroke, since my hands are almost always hovering over the keyboard. Second, in some software you can create new documents without first creating a file on disk. This means that when I go to hit CTRL+S, it prompts me to save the file. That’s not to say that some software can’t save a recovery version of the document in the event the software crashes, but I’m not going to bet money on it working 100% of the time. I’d rather be proactive and personally make sure my work is saved. Gives me peace of mind.
I already covered your first point, you don’t need to.
As for your second point, autosave still does its job. The fact that you haven’t chosen a name and a folder for your document doesn’t mean that the software hasn’t created one on disk that keeps getting autosaved. When you decide to finally save the document, that file gets renamed and placed where you want it.
I mean this is trivial stuff that got solved a long time ago, I don’t see people on this thread saying I don’t trust electronic payments, I only write checks but somehow everyone think a basic feature is broken everywhere
ctrl-S is deeper, older code. And yes, a bug in that would affect both manual and automatic saving. Meaning the bug has greater exposure and therefore would be detected faster.
More easily detectable bugs are less of a problem, because lack of alarm indicates lack of those bugs.
It’s this: (P => Q) => (!Q => !P)
Basically P is the bug existing and Q is someone detecting it. The more powerful the implication arrow on the left side of that equation, the more powerful the implication arrow on the right side. Or if you prefer probabilities: a greater conditional probability on the left means a greater conditional probability on the right.
Worse bugs that affect more systems are less worthy of the user’s attention.
If current_time > x invoke deper,older code that you somehow trust
Alternatively, more modern implementation suggested by someone else in this thread
At every keystroke, invoke deeper older code that you somehow trust
While not impossible, pretty hard to slip a bug into something like that and if it happens it gets identified,reported and fixed like all bugs. Users tend to be quite vocal about data loss.
Also some software developers tend to overcomplicate things, this is not rocket science