calibre 0.6 betas now available

[ldolse]

I hadn't tried that yet, as I have a hunch it wasn't beta 10 that put the books in that state in the DB (originally added to the DB in .5), wanted to see if there was any other debug processes that I could enable.

Re-adding them is worth a shot though. Will try to find the original source books and try again. Will try restoring the .5 backup and see if I can get anywhere with re-upgrading. If I want to upgrade to .6 a second time do I need to do anything aside from swapping the db file?

[riemann42]

Kovid,

Love the work you have done with this new beta.

Love the addition of the pubdate metadata field. I have created a ticket #2745 that includes a patch resolving a minor oversight in the Library Database2.set_metadata method not including this field. This prevented the date field on epub books from being read on import, which was annoying.

I'm not much of a python programmer (need to learn it however, as my mad perl skills are making me feel like an old man).

Thanks,
Edward

[kovidgoyal]

@ldolse: Swapping the DB files should be all that's needed

@riemann42: thanks for the patch. I'll look at it when I get a moment (I'm kinda swamped at the moment moving house)

[cartman]

Good morning everyone!
I have been testing b10 and i think it is the first version of calibre 0.6 that partially solves the collection problem i have been having (basically collections were either not transferred to my 505 at all, or were just empty).
Now i get collections on the device, and they have books inside. If they are news collections, they work ok. But all the other collections have sometimes random and sometimes the correct books inside, and all of the books are empty(they show as 1 page books but when i open them i get the last page i was reading on a previous book).

In fact, all books that belong to collections are seen by the 505 as empty books, except for books books that belong to news collections(by the way these books are also the only epub books on my library). Books that do not belong to collections seem to transfer ok.

[kovidgoyal]

What format are these 1 page books in?

[JSWolf]

I have just found yet another ePub conversion bug. toc.ncx is now being generated as ONE line. It used to be generated so it was in a nicely readable form.

Also, can we have the anchors used in toc.ncx removed? All they do is slow down the ToC on the 505 & 700.

[kacir]

I would like to install 0.6 Beta in Mint Linux.
Mint Linux is based on Ubuntu Jaunty.

How do I do it?

[cartman]

Originally Posted by kovidgoyal

What format are these 1 page books in?

LRF. PDF books seem to work fine.
Edit: When I connected my device today to transfer news, all collections were finally lost. Also when sorting the books by date i can only find one out of the 11 epubs that were sent to the device.

[sherman]

Originally Posted by JSWolf

I have just found yet another ePub conversion bug. toc.ncx is now being generated as ONE line. It used to be generated so it was in a nicely readable form.

Unfortunately, the toc.ncx has been generated as one line since the latest few versions of 0.5.x . It's a simple matter of running it through tidy or similar to get a readable form.

Also, can we have the anchors used in toc.ncx removed? All they do is slow down the ToC on the 505 & 700.

Depending on how the source file and/or calibre is setup, this may not be the best thing to do.
If calibre splits the xhtml stream at TOC items (eg: Chapter headings) then removing anchors will have no negative effects, and indeed speeds up navigation. I realise that this is what happens for the majority of conversions.
However, if calibre ends up splitting the xhtml stream purely by the 300K size limit, then the anchors are necessary for correct navigation.
And then of course, there is level 2 and 3 TOC elements, which often have several occurrence in one xhtml split. Again anchors are required for correct navigation.

What I'd like to see is calibre detecting a scenario of one TOC element per xhtml split (eg: when there is a page break at every chapter) and omitting the anchors when this is the case.

[nrapallo]

Originally Posted by sherman

Unfortunately, the toc.ncx has been generated as one line since the latest few versions of 0.5.x . It's a simple matter of running it through tidy or similar to get a readable form.

I use my RegEx capable editor, Textpad, to add newlines ("\n") to the end of tokens that should signify the end of the line. It makes a world of difference when re-editing. IMHO, there's no efficiencies gained by stripping those line endings and making the 'toc.ncx' one long line/text stream.

Depending on how the source file and/or calibre is setup, this may not be the best thing to do.

If calibre splits the xhtml stream at TOC items (eg: Chapter headings) then removing anchors will have no negative effects, and indeed speeds up navigation. I realise that this is what happens for the majority of conversions.

I agree!

However, if calibre ends up splitting the xhtml stream purely by the 300K size limit, then the anchors are necessary for correct navigation.

Not so, since any split file doesn't require an anchor to go to the top of that new file i.e where the 300K split took place. You would only effectively need the anchor links when dealing with illustrations or sub-headings which require linking to a place not at the top of a page or beginning of a file.

And then of course, there is level 2 and 3 TOC elements, which often have several occurrence in one xhtml split. Again anchors are required for correct navigation.

True, and given the frequency that these secondary links may be used, leaving them in should not be a problem.

What I'd like to see is calibre detecting a scenario of one TOC element per xhtml split (eg: when there is a page break at every chapter) and omitting the anchors when this is the case.

[=X=]

Originally Posted by kacir

I would like to install 0.6 Beta in Mint Linux.
Mint Linux is based on Ubuntu Jaunty.

How do I do it?

Downloaded the file with the extension bz2 in the beta downloads here is the (link)

Make a directory calibre like /usr/local/calibre and unzip it there.

You'll have to add a launcher link or run it from the command line.
=X=

[user_none]

Originally Posted by =X=

Downloaded the file with the extension bz2 in the beta downloads here is the (link)

Make a directory calibre like /usr/local/calibre and unzip it there.

You'll have to add a launcher link or run it from the command line.
=X=

Run calibre_postinstall. It will create the launcher, setup bash auto-completion and create the hal fdi files so device detection will work.

[sherman]

Originally Posted by nrapallo

Not so, since any split file doesn't require an anchor to go to the top of that new file i.e where the 300K split took place. You would only effectively need the anchor links when dealing with illustrations or sub-headings which require linking to a place not at the top of a page or beginning of a file.

Which is basically the scenario I was referring to - and level 2-3 TOC elements often falls under this category.

True, and given the frequency that these secondary links may be used, leaving them in should not be a problem.

No, it isn't usually isn't a problem - just so long as the book isn't too large.

An example of a book that is very much affected by anchor tags is The Bible. This is a very large book. with a very large 2 level TOC (at least my version is like this anyway...). If anchors are present in the toc.ncx file, even page turns can get very slow at the end of the book (the New Testament for example). And this is true even when anchors are only present for the secondary TOC elements.
In this situation, the best solution I found was to be very generous with page breaks in the source document/calibre chapter detection. Having a page break at every primary and secondary TOC element eliminated the need for any anchors, and the resulting epub was almost as fast to navigate on the 505 as an LRF. The downside to this is the length of time it takes to do the conversion through calibre - it takes a while to split a file into 1,190 separate files.

[=X=]

Calibre beta 10 seems to be broken for Ubuntu.

=X=

[notsure]

Originally Posted by =X=

Calibre beta 10 seems to be broken for Ubuntu.

=X=

Thanks for that, I have d/l'd but not installed.