the “reader application” has to design the opds to work with ubooquity

No, OPDS is a standard, so (in theory) all clients should be able to work with all servers as long as the specifications of the format are correctly implemented.

It is however rarely that easy with software, as you can see. :)

As far as I can tell, Ubooquity serves valid OPDS (I checked again with this validator today). But from what you tell me, there must be a difference between the feed provided by Ubooquity and the feeds provided by other servers.

The following could help me invetigate further:

• The version of KOReader you use, as well as the platform (Kindle, Kobo, Android...).
• Some examples of OPDS pages (XML files) that work correctly for you. You can get them by navigating the feed directly in your browser (to get clickable links, display the page source once the feed is displayed, using the CTRL+U shortcut in your browser)

I did not manage to reproduce the issue on my side.

Is the error log exactly the same as before ?

This line in the logged XML should not be possible after 3.0.1:

 <title>Fantasy & SciFi</title>

Ampersands (and othe special characters) are now escaped in titles.

Would you mind posting the new log ? Perhaps I missed a second issue.

Yes it's supposed to be fixed.

I'll run some additional tests by putting "&" in as many places as possible and see if something breaks.

Did you make any changes to the duplicate detection that you would like

me to try out and see if there are any improvements apart from the new

flag?

Thanks, for now I just made the feature optional.

I'll look into optimizations after the release of 3.1.0 (which will be the official Ubooquity 3, not in beta anymore).

Version 3.0.2 ( https://vaemendis.net/ubooquity/downloads/Ubooquity-3.0.2.zip ) can be configured to activate Djvu specific logs.

This is done by setting the environnement variable DJVU_LOG to true.

Example of command line launch with this configuration, env variable are passed with the -D prefix:

java -DDJVU_LOG=true -jar Ubooquity.jar

Since 3.0.0, Ubooquity looks for duplicate files after scanning collections.

This can be a veeeeery long operation on big collections and cause issues like the one you encountered, so I deactivated it by default and added an option (in advanced settings) to re-enable it.

Here is the fixed version (3.0.2) : https://vaemendis.net/ubooquity/downloads/Ubooquity-3.0.2.zip

Let me know how it goes.

That's something that's been on my mind for a long time, but as few people use OPDS feeds (and although I use them) it is usually deprioritized in favor of more popular features.

it will load 3 pages of booktitles and when I hit nr 3 it will start loading the next 3

I just tested this with KOReader and Ubooquity, that's the behaviour I get too, but only when page 3 is accessed from page 2 (using the > button), not if you jump directly to the last page using the >> button in KOReader.

I don't know why KOReader does not load the next page when you go directly to the last one, that's perhaps a question to ask on their Github bug tracker.