# How to auto agree the recovery dialog

Hello!

Is there a way to automatically agree the recovery dialog on startup? There are two button presses (before to start and after to finish the dialog) that seem totally unnecessary.

I would have expected a flag to something like that or at least a user configuration, but I found neither of them.

Regards

edit retag close merge delete

Sort by » oldest newest most voted

No there isn't, and shouldn't be.

The dialogs are there to (1) inform use that a crash happened last time (which is by the way not always obvious: sometimes it happens at shutdown, and without any messages; user could close application, agree with saving changes, and not realize something went wrong); (2) allow user to choose to attempt recovery or not (and that is essential, because in some cases, when a crash was caused by something in the document, an attempt to recover could crash again, and not being able to skip recovery would make LO crash at startup, without a way to uncheck the option other than resetting the user profile); (3) show user if all files were recovered successfully or not (and it might be a crucial information). Not showing the dialogs would give surprises to users when opening some document: other documents could open "by itself"; a user could not realize that a recovered document needs checking before saving in place of the original (sometimes a recovered document could be corrupted, and the older saved document be in a better state) - and dialogs draw attention to that.

more

While it's not immediately clear to me that it shouldn't work if the user explicitly configures that, I don't think that matters for my question. I think that forcing the user to do something is usually the wrong route, while suggesting decent defaults is an important part of designing the user interface. And the default is very sane, for the reasons you have pointed out.

I would find that intuitive to use as a cli flag (in the spirit of the very helpful --norestore) much rather as part of the user profile configuration. Just because the crash quite often is obvious.

Regards

( 2019-10-09 17:41:48 +0200 )edit
1

I think that forcing the user to do something is usually the wrong route

Usually, yes. But recovery is not something that is part of normal program operation. It's the function that works when the program is misbehaving. And it's impossible to know in advance what and how is misbehaving. Trying to automatize failure handling (in the sense of "agreeing to recovery questions" as asked here) is the wrong thing to do (and may even cause dataloss on its own). (To the contrary, ignoring failures is a sane option sometimes, and --norestore is OK in those cases).

( 2019-10-09 17:51:19 +0200 )edit

I tend to disagree. I could argue for the usecase. For instance I know a few people that always call libreoffice with --norestore since they are annoyed by the dialog, yet they aren't sure enough that they will never need the recovery. Let's talk about the case you mentioned before: in the startup of the operating system, I know that the shutdown or a crash of the whole os preceded.

But recovery is not something that is part of normal program operation.

For some users it might. That's all I'm saying. It's important to have mental image of the user you are designing this for and set the defaults accordingly. Artificially limiting the use cases by forcing the "normal behavior" is very rarely a good idea. I strongly disagree with the philosophy.

( 2019-10-09 19:18:22 +0200 )edit
1

You seem to misunderstand what I'm saying. It doesn't matter if you (as a power user and having some special needs) might think of a normal operation. You might well have a workflow that might have the crash desired/unavoidable/normal; and you might want the dialog to be hidden in that case. Yet, the program itself is never guaranteed to do the right thing when recovering. A crash is always something unexpected from app PoV, and it might happen with arbitrary program state. A result of recovery depends on the state of recovery information, which could happen to be damaged during the crash, too... there are many variables, and all of them are unknown at the time before starting recovery (and your expectations don't affect the result at that point, even if you had the success a thousand times before). If you need to agree to ...(more)

( 2019-10-09 19:30:24 +0200 )edit