How to speed up KOReader on opening EPubs?

[iandol]

Hi, I tend to swap around several EPubs fairly frequently.

KOReader takes ages (up to a minute or two) when opening an EPub compared to Plato or the native EPub reader (Kobo Nia). My EPubs are a mix of EPub3 and kepub (generated using Calibre). I love the KOReader feature set, so want to stick with it and I wondered if anyone had any tips or configuration tweaks for speeding book loading up?

[JSWolf]

Originally Posted by iandol

Hi, I tend to swap around several EPubs fairly frequently.

KOReader takes ages (up to a minute or two) when opening an EPub compared to Plato or the native EPub reader (Kobo Nia). My EPubs are a mix of EPub3 and kepub (generated using Calibre). I love the KOReader feature set, so want to stick with it and I wondered if anyone had any tips or configuration tweaks for speeding book loading up?

KePub has a lot of extra code that's not needed for KOReader. If you sent them to your Nia as ePub, they would be smaller and would open faster. I cannot say you'd notice, but you can try comparing how long they take to open with some eBooks in KePub to the same eBooks in ePub.

[iandol]

Thanks for the hint. I tried comparing a 7.4MB EPub ("polished" using Calibre to be EPub3 compliant) copied directly vs. exported via a kepub using kobo utilities.

First start:

EPub: 47.5 secs
Kepub: 58.8 secs

I navigated to the same page in both (both quite sluggish, navigation within the epub can be slow). Then within the same session exiting to the file manager and returning to the book is quicker (both are around 8-10 seconds). Plato with the same files:

EPub: 2.4 secs
Kepub: 2.8 secs

Interestingly now if I go to the same page, then exit and reenter each book, Plato slows down to ~7 secs, but things like navigation also tend to be faster.

Making any layout changes also takes ages in KOReader. I am using a custom font, although it has been "readified" using the python script.

[NiLuJe]

See this technical explanation: https://www.mobileread.com/forums/sh...73#post4001973

That would mean this book falls into the "giant" and/or "degenerate code" category .
Is it really gigantic (i.e., omnibus or bible), or does it feature some weird CSS trickery?

[iandol]

@NiLuJe: wow, that is a great description of what is going on, I understand this much better. I tend to read larger academic books (my current one is 4496 pages according to Plato), and so probably am pushing KOReader against its advantages.

Do you know how long this cache is utilised (does KOReader have a cache policy?), and how many books KOReader can cache on what I assume is something low-specced like the Kobo Nia? Will it cache till RAM runs out?

[JSWolf]

There is no need at all to polish to convert ePub 2 eBooks to ePub 3. It does absolutely nothing useful unless you need to add ePub 3 specific features. Just let the ePub 2 eBooks stay as ePub 2 and the ePub 3 stay as ePub 3.

One thing KOReader does is cache the eBooks. You can try opening them again after they've once been opened to see if they get any faster. I don't know what Plato does.

[Frenzie]

Originally Posted by iandol

@NiLuJe: wow, that is a great description of what is going on, I understand this much better. I tend to read larger academic books (my current one is 4496 pages according to Plato), and so probably am pushing KOReader against its advantages.

The number of pages doesn't necessarily matter that much. It's more the number of HTML elements and the complexity of the CSS.

Of course any document over a thousand pages will have quite a number of elements by logical necessity, but you can fairly easily create a 10-page document that'll do as bad or worse as a 1000-page one if you (don't) know what you're doing.

Btw, you can always share the document using ScrambleEbook if it's not open access.

Originally Posted by iandol

Do you know how long this cache is utilised (does KOReader have a cache policy?), and how many books KOReader can cache on what I assume is something low-specced like the Kobo Nia? Will it cache till RAM runs out?

The most important cache in this context is written to files, in cache/cr3cache.

The default max size for that is 64 MB, but you can set it higher, um, somewhere.

What that means in number of books depends on the books. Your average book will generate a cache of about .25 to 1 MB in my experience, so I'd be inclined to phrase 64 MB as at least 64 books. But your specific scenario will doubtless generate a much larger cache file.

[iandol]

Frenzie, For the book I tested the cr3cache file for it is 5.8MB, which is certainly larger than some other cache files for more "standard" sized books.

The scrambled EPub is shared here; as far as I can tell the HTML appears straight-forward, but perhaps the large number of documents or the large number of footnotes is to blame: https://www.dropbox.com/s/r9idu2h47v...bled.epub?dl=0

Armed with all this information, I can at least make a more informed decision about which reader to use with with books. Thanks again to you all for the information you have shared.

[Frenzie]

Mainly it's just very large.

[iandol]

Ha, thanks for the confirmation!

[Frenzie]

PS I'm attaching an ebook that shows off the general principle of the kind of thing I mean. But it's small enough that it doesn't really matter. Every little word or even segment comes with its own excessive repetition of the exact same thing. But I've seen books (MS Word exports?) that do it much worse still, with a multitude of elements around every word.

<span style="color:#000000; font-family:Times, serif; font-style:normal; font-weight:normal; text-decoration:none">eour</span>

If you want to buy it, it's in the Itch.io Ukraine bundle btw. Link here.