The book Reader or "View" alters EPUB metadata

[J J]

I'm running Calibre 4.23 on Linux.

Reading an EPUB via the Calibre default "View" results in altered metadata upon closing the book.

Since bookmarks and such are kept in the database, I know of no reason the epub file should be altered. Setting the EPUB file to Read-Only resolves this issue and reading and bookmarks still work.

What is being written to the EPUB?

[J J (justme97)]

I did a little more investigation and found that "ebook-viewer" is writing meta-data under "Miscellaneous" "calibre_bookmarks.txt" even though bookmarks are controlled via the database. Deleting this entire entry via "edit book" has no effect and the book marks are still present because they are in the DB.

The content written also includes: access time stamps, Book Title, and my User ID. It's written as JSON Base64.

This should not be happening!

[Kovid Goyal (kovid)]

LTrun it off under the miscellaneous section of the viewer preferences.

 status invalid

[u922796 (u922796-deactivatedaccount)]

Why is this issue considered "invalid"? An EPUB viewer is expected to view EPUB files, not silently modify them. This is not a good default behaviour because it is surprising. Given the number of users with a similar issue, I don't understand why this surprising behaviour is still the default.

[Kovid Goyal (kovid)]

because there are more users that like being able to share their
annotations with their files easily than there are users that have
fixed ideas about what a viewer "should" or "should not" do.

[J J (justme97)]

@Kovid Goyal (kovid) said, "LTrun it off under the miscellaneous section of the viewer preferences."

It's pretty obtusely buried, but I found it.

I still think a default WRITE for a Reader is not a good choice, but I'll deal with it.

Thanks.

[BryanFRitt (bryanfritt)]

Quote:
because there are more users that like being able to share their
annotations with their files easily than there are users that have
fixed ideas about what a viewer "should" or "should not" do.

Saving annotations could be done by creating a separate file with a name similar to the original file the same folder instead of changing the original file. This way you can exchange the meta-data with someone who also has the same file, and people be can be sure that their viewing the same e-book because the checksums of the file to read will match no matter what bookmarks/meta-data gets added to the separate meta-data file(s).

Separate file is better idea, but it could at least this could create a prompt that asks the user something like "Do you want to save a copy of the meta-data directly to the ebook file? [Yes][No ] ()Remember choice for this file and don't prompt again ()Remember choice for all files and don't prompt again" with some option in 'Preferences' to start back asking again.

"I still think a default WRITE for a Reader is not a good choice"
I agree.

A program called ____ Viewer/Reader implies it's not editing/writing those file(s), otherwise it could be called ____ Editor/Writer. Even those usually have prompts asking if you want to save or not save edits when the program is exited.

[BryanFRitt (bryanfritt)]

Question: How to remove this metadata and go back to the original files?

[ReadWrite (readwrite)]

Echoing others: a WRITE under ANY circumstances for a "Reader" deeply violates the Principle of Least Astonishment (https://en.wikipedia.org/wiki/Principle_of_least_astonishment).

My two cents: it shouldn't even be an option. Writing should always be done to a separate metadata file if needed.

But if the decision is to have the Reader write, then to have that be the default only makes the POLA worse.

I ended up here because this tripped me up pretty badly. It would be great if this could please be reconsidered and fixed. Calibre is an awesome product otherwise. Thank you for making and sharing it!

[J J (justme97)]

Wow! You guys are doubling down on your intent to have a hidden write function buried in the reader.

Defaulting to "Write" was bad enough but at least there was an option (albeit buried) to disable it. Now that Check box disable option has been removed!

All the previous reasons stated for why this arguably inappropriate "write" function was the default are clearly shown to be specious by removing the option to disable that was already present. This suggests an ulterior motive and casts suspicion and doubt on an otherwise great product.

[Frank Mouson (fmouson)]

Hi. As a college student working on a research project, I encountered a similar problem using Caliber. This error made me doubt my note's confidentiality, so I contacted https://essays.edubirdie.com/report-writing-services to finish my assignment avoiding plagiarism. Their expert assistance allowed me to focus on the content of my project without worrying about technical glitches. I hope that the developers of Caliber can resolve this problem soon, as it is crucial for researchers and avid readers to have a seamless experience using this software.

[André Chalella (andrechalella)]

Frankly it is quite insane that a viewer goes around modifying files and the team doesn't even consider the possibility of it being inappropriate.

Such behavior goes so much against professionalism in software, I'd expect it from shady proprietary software.

Come on, at least one option to disable it? There are serious people using this software for serious purposes, for some of us moving bookmarks around is not even a desirable feature, it's an actual hindrance because the file changes.