Your comments

I am not familiar with the way Apple filesystem works, but I suppose you could try to add quotes around the path:

"/Volumes/Macintosh HD/comix"
Also, are you using the web admin page or th desktop interface to configure Ubooquity ?

Yes, you can do that by creating a specific theme and hiding the download button in the CSS.

Once you have generated your new theme (or chosen the theme you want to edit), you have to modify the books.css and comics.css files by hiding the #details_download id.
#details_download {
    visibility: hidden;
}

I did it for the default theme, you can download the modified version here.

Hi,

thanks for the offer, but Ubooquity is closed source software.
Although I am not against the idea of opening it someday, I have no plan to do it so far.

So for now, the only way to customize Ubooquity is through themes. I also plan to make the Javascript part (so mostly the online reader) customizable in the next release.
Hi,
there is no option for that.
I agree with you on principle.

However, I checked the time required to scan a collection of 6000 files (almost 200 GB) on my core i3 PC (admittedly much faster than what can be found in a NAS, but still...). The scanning time once the file are in the database (i.e. the periodical polling time) is under one second. And that's because the files themselves are not accessed during this check. So I'm not concerned about the CPU usage.

And even though I can understand the annoyance of having to wait for the polling to be triggered (or to have to manually trigger it), for now the development and testing costs of this feature exceed the benefits I see in it.
(I'll keep it in my backlog nevertheless, but it might never be done)
At the beginning of Ubooquity, I had plans to include this feature.
Then, after writing the first version of the scanner (which periodically polls instead of monitoring the file system), I realized that the check of the files (the part that would be eliminated by the OS hook notification) was extremely fast as it is just a date comparison. Files are not actually read unless they have been modified (in which case a rescan is necessary in any case).

So the gain would very probably be quite small, both in terms of time and CPU resources.
Did you have a specific problem in mind that would be solved by such a mechanism ?