Just lost a whole day's work with a Sigil crash

[Hitch]

Originally Posted by Doitsu

Some time ago a wrote a very simple output plugin that does this automatically. I could PM it to you or post it in the plugin thread if there's enough interest by other Sigil users.

Doits:

Me too me too? Please? We do this now manually, which is annoying.

Thanks,
Hitch

[JSWolf]

Originally Posted by Hitch

Doits:

Me too me too? Please? We do this now manually, which is annoying.

Thanks,
Hitch

I think this plugin would be popular and posting it as an official plugin would be a rather good idea.

[PandathePanda]

Originally Posted by JSWolf

What I don't get is why the ePub is not up to at least the second-to-last save. Could this be a Linux issue?

I don't know, only had something similar happen to me, but it was on Windows and I didn't save at all (So I can't complain), but haven't had it happening to me while using Linux.

[Gregg Bell]

Originally Posted by Hitch

Gregg, why in all the hells would you do that? Why would you be deleting the REQUIRED "xml stuff"? I realize that you say that the copy-paste is bringing along the Doctype declaration, etc., but, WHY delete what's there? Why aren't you pasting over it, if indeed the copied material has the requisite "stuff"? Honestly, you are ASKING for errors like that.

Hitch, I was taking one individual html file from one .epub and transferring it to another. So say, I took the "Leave A Review" file from a previous book. So I open a blank html file in the latest .epub, delete what's there. Go to the old .epub. Copy what's there. And paste that into the blank html. I'm figuring the xml stuff is the same and I didn't even know I could 'paste over' what is there. So you're saying open the blank file, leave the xml stuff, copy the old stuff, come back to the new file and just paste?

I've done that with files of course but then I get the 'do you want to replace this file?' But that's with files, not with what's inside the file.

And as long as the file passes epub check (which they always do) what's wrong with doing it my way?

Originally Posted by Hitch

You know, I've been "in" computers since 1975. Oh, not like some of the guys here; I'm no steely-eyed programmer. But I started with 80-column punch cards, in '75, and I've been involved with them in one form or another since that time. Everything from developing early word processing applications, to installing Basic4 mini-computer networks in hotels, to what I do now, and in all that time, Gregg--in 42 years--I've never, never seen anything go back and eat a saved file. Never. Not this side of a disk crash, of course.

Yes, I agree with you, it seems impossible.

Totally agree. Upon further reflection I think what happened was what Kevin suggested. I was saving the file and that lost everything. Like the screen froze. I realized I didn't have functionality so not wanting to lose anything I quick went to save it.

Originally Posted by Hitch

Also: I don't understand what it is you are copy-pasting. If it's your book, don't you just export it to HTML, and then open that HTML in Sigil?

Don't delete the head material. Just...paste over it, IF you truly have the same material in the copied stuff.

Hopefully I explained that earlier. And yeah, I go from .odt to .html to Sigil.

Originally Posted by Hitch

I gotta say, I think there's a perfectly good ePUB, someplace, on your computer. What you say does sound completely impossible.

Hitch

The file is fine now. I just went back and re-did the work.

[Gregg Bell]

Originally Posted by JSWolf

I have to agree. Unless the crash happened that affected the ePub, the ePub would most likely be corrupt.

The file is fine.

[Gregg Bell]

Originally Posted by Doitsu

You might be able to recover some files from the Sigil temp folders.

To locate the Sigil temp folders do the following:

1. Close all Sigil instances.
2. a) Windows: Select C:\Windows\Temp in the Windows Explorer.
2. b) Linux: Select /tmp in your file manager.
3. Sort the folders alphabetically and look for folders that start with Sigil, e.g. Sigil-ABC123.
4. Double-click the folder. If it contains an OEBPS folder, double-click it, open content.opf in a text editor and look for the title metadata entry:

<dc:title>My Book Title</dc:title>

4. If it's your book, check the folder date/time and check the HTML files in the OEBPS/Text folder. If they contain the lost text, copy the whole folder to the desktop and add the HTML files to a new book.

Do not modify Sigil temp files while Sigil is running!

Thanks a lot, Doitsu. but I'm Linux and anyway I already re-did the work I lost and am in good shape now.

[Gregg Bell]

Originally Posted by BetterRed

@Gregg Bell - FWIW I use Save As, with current timestamp prepended to the file name.

BR

Thanks Red. Not sure how to do this but I see there's a suggestion of doing it automatically so I'll follow the rest of the thread.

[Gregg Bell]

Originally Posted by KevinH

I still have issues with this ...

1. Sigil will not erase a previously saved file that is saved anyplace outside of its own temp folder.

2. Sigil might corrupt a file being saved to if a crash happens while in the act of saving but even that is rare as it prebuilds the new epub in its temp space and then copies the already built epub file to the location to be saved. So a good copy of the epub should either exist in temp or outside of Sigil temp area.



So an epub file stored outside Sigil's temp folder should never be erased as described in this post. It sounds more likely it was being saved to inside Sigil's temp folder (a really bad idea!), bad anti-malware software, or disk corruption unrelated to Sigil.

Did you try Doitsu's instructions for looking inside your systems temp folder to see if you can find a more up-to-date version of the epub or epub files?

I'm Linux so I couldn't follow Doitsu's instructions.

Like you said:

Sigil crashing can not impact a previously saved file no matter how it tries unless the crash happens when saving to the only copy of the file

It was the only copy of the file I'd been saving to and in hindsight, I probably did try to save the file when I realized the screen was froze. And yet the file did not get corrupted. My machine has no malware and I have no anti-malware software. I wouldn't know how to save in Sigil's temp folder.

[st_albert]

Originally Posted by JSWolf

I think this plugin would be popular and posting it as an official plugin would be a rather good idea.

I agree. I would use it too.

Albert

[Gregg Bell]

Originally Posted by JSWolf

What I don't get is why the ePub is not up to at least the second-to-last save.

Me either.

Originally Posted by JSWolf

Could this be a Linux issue?

I really doubt it. Never had a problem and have made lots of epubs in Sigil on a Linux machine.

[PeterT]

Originally Posted by Gregg Bell

I'm Linux so I couldn't follow Doitsu's instructions.

Funny; I was certain Doitsu had linux specific instructions in his post

[BetterRed]

Originally Posted by Doitsu

Some time ago a wrote a very simple output plugin that does this automatically. I could PM it to you or post it in the plugin thread if there's enough interest by other Sigil users.

Thanks, but I wouldn't use it because I have a 'universal' 2 keystroke solution.

However, it does seem others would welcome it.

BR

[Hitch]

Originally Posted by Gregg Bell

Hitch, I was taking one individual html file from one .epub and transferring it to another. So say, I took the "Leave A Review" file from a previous book. So I open a blank html file in the latest .epub, delete what's there. Go to the old .epub. Copy what's there. And paste that into the blank html. I'm figuring the xml stuff is the same and I didn't even know I could 'paste over' what is there. So you're saying open the blank file, leave the xml stuff, copy the old stuff, come back to the new file and just paste?

Yes, that's what I'm saying, IF the new material actually has the doctype declaration, etc. The example you showed us, above/earlier, didn't have that, which is what flipped me out. But there's no reason not to go to the new material, CTRL+A (select all, or, whatever you're copying), then go to the new ePUB, CTRL+A (select all in the new "empty"-ish HTML file) and then CTRL+V, paste. I don't know what the cmds are for your linux box, although I though you were running a VM, so my commands should work, if that's true.

I've done that with files of course but then I get the 'do you want to replace this file?' But that's with files, not with what's inside the file.

Stupid as it sounds, I can't tell you what it says. I do it all the time, and I guess I haven't filed away any "are you sure" last-chance salvation notices.

And as long as the file passes epub check (which they always do) what's wrong with doing it my way?

Well, for one thing...didn't you just have a catastrophic failure, with doing it your way? Why keep doing that, given what occurred? How do you know that deleting everything in that file wasn't what caused it?

My philosophy is always somewhat medical: first do no harm. If you leave the doctype, etc., in the newly-created HTML file, and replace it with identical coding, in theory, there'd be less chance for it to crash, but...hey, it's your time, do what you will.

Totally agree. Upon further reflection I think what happened was what Kevin suggested. I was saving the file and that lost everything. Like the screen froze. I realized I didn't have functionality so not wanting to lose anything I quick went to save it.

Yes, well, that's the only way that I can see for you to have lost the prior-saved file on the disk. Otherwise, this makes NO sense at all. For future reference, in case you didn't already get this from all the other guys, if you get the PSO'D (Pink Screen O'Death), or anything bizarro-world, like that, EXIT. Don't bloody well save.

Hopefully I explained that earlier. And yeah, I go from .odt to .html to Sigil.

The file is fine now. I just went back and re-did the work.

I suppose. You know, you could also simply unzip the ePUBs, and use the "insert existing file" functionality, to add the "leave a review" HTML file from one ePUB to another. Working in ePUB parts is also good experience. Instead of that copy-paste stuff. Since you don't do this stuff all day long, make a COPY of the ePUBs that you're going to unzip, etc. (n.b.: we do this all day long, run RAIDS, plus do all our work on Dropbox [another level of backup], plus version everything, plus run nightly backups, plus run full weekly backups), and we still make copies of any files we'll be working on or using. If we do, you probably really ought.

Good luck.

Hitch

[Gregg Bell]

Originally Posted by PeterT

Funny; I was certain Doitsu had linux specific instructions in his post

You're right. My bad. I just saw

2. a) Windows

and my mind shut down.

[Gregg Bell]

Originally Posted by Hitch

Yes, that's what I'm saying, IF the new material actually has the doctype declaration, etc. The example you showed us, above/earlier, didn't have that, which is what flipped me out. But there's no reason not to go to the new material, CTRL+A (select all, or, whatever you're copying), then go to the new ePUB, CTRL+A (select all in the new "empty"-ish HTML file) and then CTRL+V, paste.

Thanks Hitch. So your way overwrites what's there in the xml stuff? I can do that. But what if the xml stuff in the old epub is old? Maybe the epub check will tell me?

Originally Posted by Hitch

Well, for one thing...didn't you just have a catastrophic failure, with doing it your way? Why keep doing that, given what occurred? How do you know that deleting everything in that file wasn't what caused it?

I'll do it your way from now on.

Originally Posted by Hitch

My philosophy is always somewhat medical: first do no harm. If you leave the doctype, etc., in the newly-created HTML file, and replace it with identical coding, in theory, there'd be less chance for it to crash

But your way overwrites the xml coding at the top. Again, what if the xml coding in the new .epub doc is different than what was in the old epub? I'd be overwriting old code into the new .epub.

Originally Posted by Hitch

Yes, well, that's the only way that I can see for you to have lost the prior-saved file on the disk. Otherwise, this makes NO sense at all. For future reference, in case you didn't already get this from all the other guys, if you get the PSO'D (Pink Screen O'Death), or anything bizarro-world, like that, EXIT. Don't bloody well save.

Gotcha.

Originally Posted by Hitch

I suppose. You know, you could also simply unzip the ePUBs, and use the "insert existing file" functionality,

Is that the same as "Add Existing Files" when you right click on a file? (I couldn't find "insert existing file.")

Originally Posted by Hitch

to add the "leave a review" HTML file from one ePUB to another. Working in ePUB parts is also good experience. Instead of that copy-paste stuff. Since you don't do this stuff all day long, make a COPY of the ePUBs that you're going to unzip, etc. (n.b.: we do this all day long, run RAIDS, plus do all our work on Dropbox [another level of backup], plus version everything, plus run nightly backups, plus run full weekly backups), and we still make copies of any files we'll be working on or using. If we do, you probably really ought.

Good luck.

Hitch

Thanks Hitch. I usually do good backing up. This was just kind of a perfect storm sort of thing.

I can do the unzip thing, but really your first suggestion (your way of copying and pasting) will save time as I can see what's in the files. As long as that works I think I'll do it that way.

I really appreciate the help.