Notices

It was gone for 6 days... first time I had screen corruption was upon first run, I uninstalled, reinstalled and everything was fine...
During these six days I randomly used liqbase (most of the time after new packages installation to check if something was broken by new software) to sketch silly things while sick in bed
In the period I rebooted just once for the personal launcher update and charged the NIT twice.
Last app I run before having graphic issues with liqbase was VGBA.
Never fiddled with the screen resolution nor CPU speed under liqbase...
(I just hope my N800 hasn't got the flu from me... )

Its really odd and so transient.
As I said I've seen it only a couple of times since its inception and never reproducible.
You said restarting liqbase cured it, but has it come back since and you have had to restart again, or have you only seen it once today?

Once today, couldn't reproduce it after I saw it the first time...
Can't even say it was because of the device was running hot for a long gaming session, because i restarted liqbase almost immediately, not giving the nit the time to cool off

meh, ill just scratch my head about it and carry on with the other stuff I'm bubbling away at.

I've moved all the style information into one place now and have started to prepare user definable theming.

The graffiti wall has been given a personal boost due to refactoring and upgrading some of the components, it now does kinetic at full speed without jutters whilst still loading in pages from the disk (it simply waits for the kinetic scroll to finish before carrying on loading)

Zooming on the graffiti wall with +- works really nicely (and stores last size) and you can go from fullsize down to hundreds of sketches on screen at one time. This is in preparation for allowing a multiselect tagging mode.

I did some more work on the error handling for images and its a lot more capable at handling errors from libjpeg.

I have had to re-enable slide in menus (oh the hardship!) because when I centralized the styles I removed 99% of the time difference opening up the next menu, it was *too* fast and I needed to do something.
Feedback on this will be required and alternative options formulated for those that aren't happy with it.

A few people have mentioned expanding on the screensaver/clock principle either by displaying a fullscreen clock or more bacon and "stuff".
I will consider a few ideas over the coming days and see if anything comes of it.

In the meantime I still have no png saving and think I need to get it in - I have a lot of things to put in place which require images.

I will upload a new package later on (should be 0.0.4) once its built.

Justnick, one last thought on the corruption.

Did it occur when you first started liqbase or after you had been running a while.
When its happened to me it was right at the start the first time I tried drawing.

Always right after the start of the program, once in the drawing tool I had corruption on stylus contact.

played around a bit, kinda fun

I've also had the screen corruption problem happen. It happens every once in a while, and restarting liqbase always fixes it. The corruption happens on stylus contact, and I've observed it when drawing, and on the grafitti wall. But, as mentioned, it never seems to be reproducable.

ok, the sketching is the fastest part of the system.
Of any area it is the one place where minimal work is done to the canvas.
Simply extend the current drawing by 1 line and refresh.
liqbase always waits for xv to inform it that the frame is completely sent and I do not touch the memory again until I receive this event (overdrawing before this event is a different and leads to tearing).

Since I am not updating the whole display and sending a bitmap which is very similar to current display it appears as though the LCD controller itself is eating the lower portion of the page.

Perhaps it is caused by another process attempting to update its portion of the framebuffer when XV is meant to have it locked?

Has this been noticed anywhere else?



Also, if it occurs, try switching to a different drawing tool, the others work "live" and refresh the whole screen from top to bottom including buttons and stuff.

I tried to recreate what I was doing yesterday before the screen corruption occurred, but nothing happened...
Maybe sometimes a process becomes stuck without being noticed, but it is hard to track down what really happens
I'm aware there's a bug in the GPE summary applet that sometimes can cause the home screen to be reset (it was discussed in the GPE thread), but this was not the case...
From now on I'll try to pay more attention to what I'm doing on the tablet before reporting "bad attitudes" of liqbase