KPV: a PDF reader for Kindle, based on muPDF, GPLv3

[dave2008]

Originally Posted by eLiNK

Actually, that's how it works in Duokan. You press ALT + some number, and an input box with the number already pressed pops up.
But it does not matter.

An input box fired up by some shortcut would be fine as well.
Could you also provide some information on current page / total number of pages and incorporate three buttons "GoTo", "Goto Start", "GotoEnd"?

That would be sweet.

No problem, but we need to first add button system to current UI. This may take some time.

[geekmaster]

Originally Posted by knc1

One less process creation.
shebang takes two (shell and shell proc'ing lua) sometimes three, depending on how it got called.

kernel binfmt_misc only has to proc' one process (lua) and that is probably in the kernel's cache buffers (a single process clone call).

Too bad tcc doesn't do ARM code (just x86 and a bit of MIPS).

tcc DOES do arm code. lrizzo posted it on his web page with his first kiterm a long time ago. I also built it from source (static and lib-dependent versions) that work in the kindle without optware, and generate arm code. I have a static nano too. Like lrizzo, I do not have all the headers I need though... hello.c worked fine. To find the latest arm stuff, you need to search for tinycc, not tcc...

The nice thing about static linking is they work on all kindle models. Lib-dependent versions complain about wrong or missing libs when you move them to a different kindle model.

[dpavlin]

I just finished implementation of two-column zoom mode which is very useful when reading articles on kindle (and frequently requested in this thread).

When you press F, you will be zoomed to top-left column of page. This will also switch you to pan mode, so you can use fiveway to move up/down in current column and left/right to switch columns (and change pages if needed).

Pressing left will always position you to bottom of previous column, while pressing right will move you to top of next one which is consistent with reading flow (even when scanning document backwards).

At any time you can press a/s/d keys to get whole page overview and exit pan mode.

Until changes are merged to main repository, you can download new version at:

https://github.com/downloads/dpavlin...er-de1725d.zip

[dave2008]

Originally Posted by dpavlin

I just finished implementation of two-column zoom mode which is very useful when reading articles on kindle (and frequently requested in this thread).

When you press F, you will be zoomed to top-left column of page. This will also switch you to pan mode, so you can use fiveway to move up/down in current column and left/right to switch columns (and change pages if needed).

Pressing left will always position you to bottom of previous column, while pressing right will move you to top of next one which is consistent with reading flow (even when scanning document backwards).

At any time you can press a/s/d keys to get whole page overview and exit pan mode.

Until changes are merged to main repository, you can download new version at:

https://github.com/downloads/dpavlin...er-de1725d.zip

Works great on my K3! This is a killing feature for reading papers :-) Thanks dpavlin.

[knc1]

Originally Posted by geekmaster

tcc DOES do arm code. lrizzo posted it on his web page with his first kiterm a long time ago.

I must have been looking at an old copy I had on hand here.
No, wait, I don't have that excuse. I have the tests of its x86_64 code generation posted and the ARM code generation went in before x86_64 was completed.

Does this mean that Robert Landley and Fabrice Bellard have settled their long running dispute?
I don't see Rob mentioned:
http://savannah.nongnu.org/projects/tinycc/

But since the project has been moved to a git repository and the project lead has changed...
http://bellard.org/tcc/
http://repo.or.cz/w/tinycc.git

Perhaps the long running war just died due to lack of interest on the parts of Landley and Bellard (or they both ran out of ammunition).

[knc1]

Originally Posted by knc1

Perhaps the long running war just died due to lack of interest on the parts of Landley and Bellard (or they both ran out of ammunition).

While on the subject of Robert and building code for ARM -

He has his Aboriginal Linux project to the point that it will build (native, in a QEMU environment) LFS-6.8 for all supported target processors (with the exception of a few remaining problems with Sparc).
http://landley.net/aboriginal/

Just select ARM{v6} as the target in his build system, it will churn out a complete, working, native build system image for QEMU. (or use one of his pre-built binaries).
http://landley.net/aboriginal/about.html

With a decent x86{,_64} system of a few cores, faster than on the real hardware.

The only header files that might be missing from the image would be Kindle I/O devices (eink and touch screen - keyboard is in the FreeScale kernel tree).

[dpavlin]

Originally Posted by dave2008

Works great on my K3! This is a killing feature for reading papers :-) Thanks dpavlin.

I made a small video which tries to explain how to use it:

[geekmaster]

That is MUCH better than rotating to landscape mode for two-column documents. Thanks!

[dave2008]

Originally Posted by eLiNK

2) TOC: nothing big, but some lines contain a special symbol at the end of the line (an empty box), others don't. Maybe it is just me.

Hi eLiNK, I just checked the special symbols in one of my pdf book. It turns out that these symbols are caused by wrong formated title enties in the pdf. So, for my case, it is not a bug in kindlepdfviewer nor muPDF. I add a workaround to fix these boxes for my book, not sure it will work for you though. Your boxes mighty be caused by other reasons.

The fixes will be included in next new version after hw merged pull requests I guess.

[dave2008]

OK, I added a down page in kindlepdfviewer's wiki:
https://github.com/hwhw/kindlepdfviewer/wiki/Download

So now you know what has been changed in every version.

[eLiNK]

Originally Posted by dpavlin

I just finished implementation of two-column zoom mode ...

Nice. I'll give it a try.
Thanks.

[eLiNK]

Originally Posted by dave2008

OK, I added a down page in kindlepdfviewer's wiki:
https://github.com/hwhw/kindlepdfviewer/wiki/Download

So now you know what has been changed in every version.

Looks good.

BTW, you've got a message.

[eLiNK]

Originally Posted by dave2008

Hi eLiNK, I just checked the special symbols in one of my pdf book. It turns out that these symbols are caused by wrong formated title enties in the pdf. So, for my case, it is not a bug in kindlepdfviewer nor muPDF. I add a workaround to fix these boxes for my book, not sure it will work for you though. Your boxes mighty be caused by other reasons.

The fixes will be included in next new version after hw merged pull requests I guess.

No problem,
in the meanwhile I'll check where those special symbols are coming from.
I have created bookmarks with the Pdf & DjVu Bookmarker (http://sourceforge.net/projects/djvubookmarker/).

[Germanc]

This is great, and with djvu support it will be _perfect_. But I'm having a problem. Big pdfs (mostly scanned books) are really slow to open and to turn pages, up to 20 seconds. Is that expected ?
Keep up the good work, if the speed issue is solved I promise to hang up a poster with the developers names on it :P

[dpavlin]

Originally Posted by Germanc

This is great, and with djvu support it will be _perfect_. But I'm having a problem. Big pdfs (mostly scanned books) are really slow to open and to turn pages, up to 20 seconds. Is that expected ?
Keep up the good work, if the speed issue is solved I promise to hang up a poster with the developers names on it :P

It probably is. If you google for mupdf bugs, you will often find same complaint. Complex pages with lot of bitmaps (like Linux Journal pdf files) also take a while to render (and take a lot of memory, but we are much better at that lately).