Kobo future firmware feature request thread

[speakingtohe]

Originally Posted by davidfor

Or going through the bug listing threads and marking them as fixed. Actually, that would be pretty depressing. I generally post a list of the bugs and changes I


Are you on 2.6.1? That seems a lot faster to list the shelves. But, it still could be improved.

Yes I am and it is pretty much the same. Maybe longer.(I use the one potato two potato method of timing usually)

Was about 90 seconds-2 minutes on 2.5.1 and never as fast as 90 seconds on 2.6.1. And I deleted 10 books Perhaps it is trying to teach me patience.

Helen

[davidfor]

Originally Posted by speakingtohe

Yes I am and it is pretty much the same. Maybe longer.(I use the one potato two potato method of timing usually)

Was about 90 seconds-2 minutes on 2.5.1 and never as fast as 90 seconds on 2.6.1. And I deleted 10 books Perhaps it is trying to teach me patience.

What is the biggest shelf? And how many? My impression of a speed improvement was with 1200 books on my Touch, about 30 pages of shelves including two with 200+ books and one with 900+. That was taking less than 30 seconds to open. I'll have to try the same on the Glo and see if there is something different there.

[speakingtohe]

Originally Posted by davidfor

What is the biggest shelf? And how many? My impression of a speed improvement was with 1200 books on my Touch, about 30 pages of shelves including two with 200+ books and one with 900+. That was taking less than 30 seconds to open. I'll have to try the same on the Glo and see if there is something different there.

That seems pretty fast to me. My mini is reasonable, about 20 seconds with around 800 books, largest shelf maybe 20. Sometimes it is only 11 seconds.

The Aura has around 2400 books, with 66 pages of collections and is around 2 minutes to open. Largest collection 73 books.

With 2.5.2 I had same amount of books +/-10 and around same amount of collections. But largest 84 and was almost always les than 2 minutes.

It is late so I am not awake enough to be more precise. Tomorrow or next day I will get exact numbers.

I changed my collections from series to author, with no noticeable difference and and had both for a short time.

I am absolutely at a loss to understand what the Kobo can be doing with the books that would take that long. And why they have to do whatever it is all over again when you hit the back button on the aura. Oh well sorry to be such a sniveller

Helen

[davidfor]

I just did another test. The SD card has 1184 books. About a quarter are PDF, the rest ePubs.

I first put it into the Touch. Waited a while and then plugged it into calibre to create the shelves. That created 31 pages of shelves. The big shelves have: 931, 66, 259, 211, 208 and 208 books (you can guess how that was sorted). Most of the rest are less than five books with a few between five and ten.

Opening the shelf list was about 40 seconds each time. I tried from the home screen and from the main library list. The speed was fairly consistent.

I took the card out and put it in the Glo. That took 11 minutes to process the books. Again connected to calibre and let it create the shelves. There were a few more shelves as there were already 200+ books on the Glo. But the sizes of the big shelves were the same. Actually, the numbers above came from the Glo. The big shelves should only have had books from the SD card, so they should be the same on both.

On the Glo, opening the shelf list list took 31 seconds. Where I started from or the sort order didn't matter. Again, the 31 seconds was very consistent.

A confession: This is actually with beta firmware on both devices. But, the figures for the Touch agree with what I saw with 2.6.1. And for anyone who wants to know, here's a hint about a change up there. And no, I'm not going to answer any questions.

[moffattm]

Originally Posted by speakingtohe

Aww, you are just trying to make poor Vugtitan more morose.

I am not totalyy sure there is any bug fixing going on to take away from.

Helen

I wouldn't want to make poor Vugtitan any more morose than he already is, the poor chap is already losing lots of sleep over his absent file browser.

Unfortunately I think he's with this idea. I think there's more chance of getting a file browser via a third-party plug-in rather than from Kobo. Kobo could do a few things to make developer's lives easier by setting up some sort of mechanism to integrate plug-ins with the Kobo menus.

[speakingtohe]

Originally Posted by davidfor

I just did another test. The SD card has 1184 books. About a quarter are PDF, the rest ePubs.

I first put it into the Touch. Waited a while and then plugged it into calibre to create the shelves. That created 31 pages of shelves. The big shelves have: 931, 66, 259, 211, 208 and 208 books (you can guess how that was sorted). Most of the rest are less than five books with a few between five and ten.

Opening the shelf list was about 40 seconds each time. I tried from the home screen and from the main library list. The speed was fairly consistent.

I took the card out and put it in the Glo. That took 11 minutes to process the books. Again connected to calibre and let it create the shelves. There were a few more shelves as there were already 200+ books on the Glo. But the sizes of the big shelves were the same. Actually, the numbers above came from the Glo. The big shelves should only have had books from the SD card, so they should be the same on both.

On the Glo, opening the shelf list list took 31 seconds. Where I started from or the sort order didn't matter. Again, the 31 seconds was very consistent.

A confession: This is actually with beta firmware on both devices. But, the figures for the Touch agree with what I saw with 2.6.1. And for anyone who wants to know, here's a hint about a change up there. And no, I'm not going to answer any questions.

I removed some books and timed shelf opening

Spoiler:

2520

746 collections

73
39
33
24
24
23
23
22
22
22
20
18
18
17
7x16
4x15
6x14
3x12
15x11
9x10
15x9
18x8
15x7
...

106 seconds

removed 49 books from 73 book shelf
101 seconds

Deleted books from 3 largest collections
86 seconds

Deleted books from all collections over 20 books (10)
68 seconds
690 collections
2217 books

deleted 284 books from 19 collections
64 seconds
1933 books

deleted 43 books 43 colections
1889 books 569 collections
58 seconds

deleted 54 books 54 collections
51 seconds
1835 books


Added back 73 book collection
59 seconds

made collection, added no books of 231 books
220 seconds
opened shelves again and used system clock 3 minutes and 30 seconds or 210 seconds.

Removed no books, changed driver configuration to series, which gives me 479 series collections and one 231 book collection of books with no series.
49 seconds

This was on the Aura all books on SD card, no books on device, all epub.

I timed some twice once with counting and once with system clock and as you said the time was the same or close.

I am a bit baffled.
Reducing the number of collections seemed to make a reasonably consistent difference, but creating a collection of 231 books already on the device made it take almost 3 times longer. Then changing the shelf configuration to series from authors but still having the 231 book collection was fastest yet. I know that I have less shelves but if the 231 book collection added around 90 seconds why would it take much less time. Probably a logical reason but it eludes me.

I also cannot understand why the library list is so much faster. I have read claims that the books and/or database and/or covers must be processed, but are the covers/database/books not the same for the library list? I think of the library list as a large shelf, and the shelves as subsets of the library list.

Any light you can shed on this would ease my poor broken brain

Helen.

PS still love your plugins.

PPS
Accidentally deleted reading and finished shelves. AGGGHHHH, must put on again

[davidfor]

Originally Posted by speakingtohe

I removed some books and timed shelf opening


This was on the Aura all books on SD card, no books on device, all epub.

I timed some twice once with counting and once with system clock and as you said the time was the same or close.

I am a bit baffled.

That makes two of us. The Aura is supposed to be faster than the Glo and definitely faster than the Touch. I do have less books, but I that shouldn't affect the shelf list performance. Number of shelves should have an affect and the number of books on them should as well.

Reducing the number of collections seemed to make a reasonably consistent difference, but creating a collection of 231 books already on the device made it take almost 3 times longer. Then changing the shelf configuration to series from authors but still having the 231 book collection was fastest yet. I know that I have less shelves but if the 231 book collection added around 90 seconds why would it take much less time. Probably a logical reason but it eludes me.

The only thing I can think is that when you removed your old shelves and replaced them with the author shelves, the database was cleaned up. Maybe the mass insertions needed for this caused a rebuild of the indexes.

I also cannot understand why the library list is so much faster. I have read claims that the books and/or database and/or covers must be processed, but are the covers/database/books not the same for the library list? I think of the library list as a large shelf, and the shelves as subsets of the library list.

I agree with you there. Displaying a list of 1200 books with covers sounds harder then counting the books for the shelves. But, it is faster. I suppose they have had more time to work on it and see it more often.

Any light you can shed on this would ease my poor broken brain

Nothing I can think of. But, my brain is broken from spending the day trying to work out why the comms at one of a customers sites is so much slower than all of their other sites.

Helen.

PS still love your plugins.

Accidentally deleted reading and finished shelves. AGGGHHHH, must put on again

Have you tried PeterT's recipe for a Finished shelf?

[speakingtohe]

Originally Posted by davidfor

That makes two of us. The Aura is supposed to be faster than the Glo and definitely faster than the Touch. I do have less books, but I that shouldn't affect the shelf list performance. Number of shelves should have an affect and the number of books on them should as well.

The only thing I can think is that when you removed your old shelves and replaced them with the author shelves, the database was cleaned up. Maybe the mass insertions needed for this caused a rebuild of the indexes.

I agree with you there. Displaying a list of 1200 books with covers sounds harder then counting the books for the shelves. But, it is faster. I suppose they have had more time to work on it and see it more often.

Nothing I can think of. But, my brain is broken from spending the day trying to work out why the comms at one of a customers sites is so much slower than all of their other sites.




Have you tried PeterT's recipe for a Finished shelf?

Yes that is what I use, I accidentally deleted them from driver configuration when changing from author to series.

The only thing I can think is that when you removed your old shelves and replaced them with the author shelves, the database was cleaned up. Maybe the mass insertions needed for this caused a rebuild of the indexes.

The only time I got the processing content message was when I put back the 73 book collection, which was when I would expect a database cleanup, then I added no books, just changed the driver configuration to put all books with no series in a single collection and the time almost tripled.
Next I changed to series collections from author collections with the same largish collection of no series and the time was much faster.

I will change it back just out of curiosity.

Helen

But I will change it back to what it was

[davidfor]

Originally Posted by speakingtohe

The only time I got the processing content message was when I put back the 73 book collection, which was when I would expect a database cleanup, then I added no books, just changed the driver configuration to put all books with no series in a single collection and the time almost tripled.
Next I changed to series collections from author collections with the same largish collection of no series and the time was much faster.

I will change it back just out of curiosity.

I was thinking of the two tables used for the shelves. There will have been a mass of deletes and inserts to remove the old shelves, create the new shelves and put the books on them. That might have ended up with nicely organised tables and indexes.

[speakingtohe]

Originally Posted by davidfor

I was thinking of the two tables used for the shelves. There will have been a mass of deletes and inserts to remove the old shelves, create the new shelves and put the books on them. That might have ended up with nicely organised tables and indexes.

Maybe. I finally got the authors collections with one collection of 231 books right and the time is 70 seconds for 549 collections with one 231 book collection, one 73 book collection etc. In the meantime, while flailing around in the calibre template language modifying one of PeterT's templates, I had 104 pages of collections and this took a whopping 4 minutes and 40 seconds to open the shelves with the same 1867 books. Four times as slow as with 55 pages of collections. (what is it doing with my poor books, shredding and rebuilding them?)

This seems to me to be about the same as 2.5.1 but I am not sure by any means.

However it is still an improvement over 2.3.3 or 2.2.3 whichever it was as largish collections have less impact

Helen

[sos_sifou]

We are in 2013 and still

No text reflow in PDF :-(

Please add this feature, even if it doesn't work properly, it should be enough for reading the text, then back to normal viewing for tables and equations :-)

[glezmen]

1) the web browser should be able to open links which want to open on a new tab/page

2) tab support to the browser would be awesome

[Treefingers]

Fixable UI problems:
- Keyboard: Put a period on the alphabet page. This is useful mostly for web pages but also for books/authors with periods in the name. Just shorten the spacebar a little to make room. And keep it in the same place (next to the spacebar) on the other pages. (side note: why doesn't the number page have a link to the accented letter page?)
- Sudoku: you can add small, "note" numbers in any order you choose to type them. But you have to "backspace" to remove them, and only in the reverse of the order you put them in. It would be better for the buttons to simply toggle the number, i.e. turning them off should work exactly the same way as turning them on currently does.
- Foot/end notes on standard epubs should work as foot/end notes do in kepubs, i.e. with a pop-up overlay. Many epubs don't have a proper return links, making them a chore to read if the notes are at all useful.

Tasty, tasty, gravy:
- Replace the fairly useless sketchpad with a text-based notepad. Better yet, combine the two: text entry with the ability to scribble on top.
- Go the extra mile typography-wise. You're already doing a very decent job here, but I've noticed a few bugs (e.g. I get line breaks on the inside of brackets/parentheses and around en dashes; this shouldn't happen) that could be fixed. And I'd also like to see those little details that add real finesse to typography. For example, punctuation at the beginnings and ends of lines (! and ? excepted) should actually hang over the margin so that a letter is at the margin instead.

Thanks for the great work, Team Kobo!

[samhy]

Originally Posted by Treefingers

Fixable UI problems:
- Keyboard: Put a period on the alphabet page. This is useful mostly for web pages but also for books/authors with periods in the name. Just shorten the spacebar a little to make room. And keep it in the same place (next to the spacebar) on the other pages. (side note: why doesn't the number page have a link to the accented letter page?)
- Sudoku: you can add small, "note" numbers in any order you choose to type them. But you have to "backspace" to remove them, and only in the reverse of the order you put them in. It would be better for the buttons to simply toggle the number, i.e. turning them off should work exactly the same way as turning them on currently does.

Good points. How many times do I try to remove a "note" number by pressing said number again, before I remember it's not working that way!

The keyboard is different on the Aura (does the Glo have this new lay out too?), but on the Mini it's frustrating to have to switch keyboards in order to have periods and @.

[Jaden]

Originally Posted by samhy

The keyboard is different on the Aura (does the Glo have this new lay out too?), but on the Mini it's frustrating to have to switch keyboards in order to have periods and @.

It's also frustrating that the German keyboard only has the umlauts under the special characters. But they are not special characters but used very often.

There should be different keyboard layouts for different languages, and even better if you could choose what keyboard layout you want (have a "German quertz" layout but English menu language, for example).

Guess that's not going to happen, though.