geekmaster kindle video player

[knc1]

When you are ready to try an audio track, I stumbled on an interesting one:

Hacker's Heaven
http://steve.savitzky.net/Songs/heaven/

Available as ogg:
http://steve.savitzky.net/Songs/heaven/heaven.ogg
And mp3:
http://steve.savitzky.net/Songs/heaven/heaven.mp3

[geekmaster]

Originally Posted by knc1

When you are ready to try an audio track, I stumbled on an interesting one:

Hacker's Heaven
http://steve.savitzky.net/Songs/heaven/

Available as ogg:
http://steve.savitzky.net/Songs/heaven/heaven.ogg
And mp3:
http://steve.savitzky.net/Songs/heaven/heaven.mp3

Cute. It fits my style.

I suppose you noticed that in my (NON-life-critical and NON-mission-critical) code, I do not waste time or space to check return codes (except for the ioctl() so it can fall back to eips ''). Of course, I *do* write (other) life-critical and mission-critical code, and I was trained young (when most people did not know what a computer was) to write code that I can defend in court to "laypersons" (non-technical consumers of the pre-internet past)...

There is still a lot of my code out there running in factories in many countries around the world, and when you have very heavy equipment moving at high speeds in close proximity to humans, you do not want a heavy piece of your machine flying through a brick wall out into the parking lot because of a "little" coding error. When robotic machinery "crashes", you cannot just press the reset button and start over. Pushing pixels is much safer than pushing metal.

For hacker purposes, who cares if a broken video file makes your video display out of alignment to the screen? Just go download a good video file to replace the bad one...

P.S. There was not enough room on the mobileread tagline to fit my original "Bare-Metal Bit-Basher and Pixel-Pusher" that I wanted, but as you can see from recent posts is a more accurate description of my code. And if you want to know more about "bit-bashing" in a style that earns my respect, you absolutely MUST read these little "pure magic" web pages (the "DeBruijn sequence" code has been especially handy on multiple occasions):

http://aggregate.org/MAGIC/

http://graphics.stanford.edu/~seander/bithacks.html

http://supertech.csail.mit.edu/papers/debruijn.pdf

The DeBruijn (pronounced "DeBroyn") sequence paper is especially magical. I used that method often before I switched to using "Bit Scan Forward" x86 instructions, and ArmV5 has "Count Leading Zeros", both of which obsolete the DeBruijn method, but I still have fond memories of it and it is worth learning just to stimulate your thought processes.

Also, the DeBruijn paper mentions "perfect hash". My original reason for using tinycc (tcc) was to implement "minimal perfect hash" code that generates C source code to fit your data, which is then "jit-compiled" with tcc from within my code. You should read up on mph too (be sure to read the comments at the top of the C programs at this link):

http://burtleburtle.net/bob/hash/perfect.html.

[knc1]

I thought you might like that one.
Hmm... Now if a fitting video can be located for the sound track...

[ixtab]

So, I finally got around to trying it, and guess what:

[root@kindle us]# zcat gmvid.gmv.gz |./gmplay 
Segmentation fault

KT 5.1.0 in main mode, using the latest precompiled binary from your first post... It seems to do something to the screen though, because it blanks it, and afterwards the framework draws on a distorted screen until it's completely restarted.

[knc1]

Originally Posted by geekmaster

http://aggregate.org/MAGIC/

http://aggregate.org/MAGIC/#Dual-Lin...ointer%20Field

1970's, DataGeneral RDOS, "Sequential Read" disk files.
Although it is probably not original with D.G.

[geekmaster]

Originally Posted by ixtab

So, I finally got around to trying it, and guess what:

[root@kindle us]# zcat gmvid.gmv.gz |./gmplay 
Segmentation fault

KT 5.1.0 in main mode, using the latest precompiled binary from your first post... It seems to do something to the screen though, because it blanks it, and afterwards the framework draws on a distorted screen until it's completely restarted.

It works on MY K5 5.1.0 (upgrade)... I will check to see if I uploaded the right version.

[ixtab]

Interestingly enough, it does work in diags. I'll reboot to main and retry, and update this post in a few minutes.

[geekmaster]

I just downloaded it from the first post and put it on my K5 (5.1.0 upgrade). It ran fine booted from main (starting USBnet from the launcher menu), using this command:

zcat sintel.gmv.gz|./gmplay

What command did you run to launch it?

You are the first person to report this problem and I cannot duplicate it, so perhaps you should download it again and try again. You can even compile it with tcc from another of my threads.

[ixtab]

I can't duplicate it either. After a reboot, it now works in main as well. Same files, same command.

The Microsoft solution: if it doesn't work, reboot.

PS: It's indeed impressive! Good quality and fluid playback.

[geekmaster]

I just noticed the "distorted screen" part of your post. Are you decompressing it (zcat|gmplay)? The program assumes that it is just raw bits, and the .gz file would give different bits.

[thatworkshop]

Originally Posted by geekmaster

It works on MY K5 5.1.0 (upgrade)... I will check to see if I uploaded the right version.

I do remember to receive Segmentation fault too, though I'm not sure if it was before me updating to 5.1.0. Now it runs smoothly.

And by the way, each time I run gmplay, I should restart or do a killall pillowd, because KT becomes unresponsive to Home key. It doesn't matter if I let the animation complete or interrupt it with Ctrl+C. I wonder why! It used to work Geekmaster, in the very first day you released this code (early version of gmplay).

[geekmaster]

Originally Posted by ixtab

I can't duplicate it either. After a reboot, it now works in main as well. Same files, same command.

The Microsoft solution: if it doesn't work, reboot.

To keep my code small and simple I do not check or change the screen parameters (bit depth, etc.). Perhaps you ran a program that set your kindle to a "non-default" video mode and did not put it back the way it found it... Bad (other person's) program!

[geekmaster]

Originally Posted by cscat

I do remember to receive Segmentation fault too, though I'm not sure if it was before me updating to 5.1.0. Now it runs smoothly.

And by the way, each time I run gmplay, I should restart or do a killall pillowd, because KT becomes unresponsive to Home key. It doesn't matter if I let the animation complete or interrupt it with Ctrl+C. I wonder why! It used to work Geekmaster, in the very first day you released this code (early version of gmplay).

The segfault is usually caused by a buffer overflow or such. The code is very simple. Perhaps I have an "off by one" error that crept into it. If somebody finds a fix please let me know.

Regarding the framework locking up -- I think that happens because I am "dominating" the display and the framework checks for display "not busy" and just "gives up" if it goes to long with eink busy... But I only saw the framework go dead like that once on my K3, and an SSH "framework restart" command fixed it...

[geekmaster]

Perhaps I should do in gmplay like I did in my "algorithmic art" scripts: pause and resume cvm and Xorg so they do not get frustrated and go comatose...

[ixtab]

Originally Posted by geekmaster

To keep my code small and simple I do not check or change the screen parameters (bit depth, etc.). Perhaps you ran a program that set your kindle to a "non-default" video mode and did not put it back the way it found it... Bad (other person's) program!

Yep, I think I can reproduce it. Try this in main:

1. Open any book, switch to landscape mode.
2.

[root@kindle us]# zcat harvey.gmv.gz  |./gmplay 
Segmentation fault

3. Switch back to portrait mode.
4. gmplay works again.

So the rotation setting seems to "physically" rotate the underlying device (if you know what I mean). You can even verify this by setting the device to portrait inverted -- gmplay will also play "upside down".