Notices

It happens to me and other too (including lcuk) mostly on first launch after install, sometimes just randomly...
Anyway quitting the program and then relaunching it fixes the problem...

I've been seeing a bit of corruption now, too. If I can make it happen again, I'll video it and get it to you.

Easiest way: uninstall liqbase, then reinstall.
99% of the time you have corruption on first run.
It's so weird...

lcuk I didn't notice it before... with the 0.0.5 build liqbase crashes on "read a book" and everything related to it. The rest is fine (and looks faster than the 0.0.3)

Well, yeah, I saw some corruption on the first run, at the keyboard screen, but it disappeared and now is back. Not certain, but it seems to have a slightly different visual character, so I'm not thinking it's necessarily the same bug.

yes it is very weird.

I have just taken a silly video of me installing it onto my pristine nokia.
n2 is the one I have seen this on most recently (6th of this month - i noted it), but my primary machine has no glitches.

I'm gonna draw a line under this, moving bugs to their proper place:

project:

https://garage.maemo.org/projects/liqbase/

bugs:

https://garage.maemo.org/tracker/?gr...=716&atid=2684

latest .deb (until it hits extras)

http://liqbase.net/

If anyone has any ideas as to why it would fail only once and then not be reproducable drop me a line.

updated http://liqbase.net/ with 0.0.6 and still cannot reproduce.

This version sends the complete log to a file.
I really didnt want to have to do this but it seems theres enough problems to warrant it.
After I have examined a few logs I will remove it again.

file is saved in ~/.liqbase/liqbase.run.log and simply appends itself.
If you get graphical glitches could you come out of the program and send me this log to liquid@gmail.com please.

Hi lcuk, I filmed for 15 minutes straight... no success
I've updated to 0.0.6, but can't find the log

huzzah!
I was playing last night with a few ideas since its only on first run after an install that people seem to be able to reproduce it.
However, I have no distinction between first run after update and any other run.
The *only* time something different occurs is the first ever run - this is when it detects there is no ~/.liqbase folder and goes about constructing it.
I am at the point of thinking its something in the background occuring when first run happens, whether its the interaction between webbrowser->app.manager->liqbase or something along those lines.
After installing liqbase app manager does a refresh cycle and builds up its list, during this it pops one of those event notification thingies.
What if that is conflicting with the timing required to do first run.

I've attempted to completely overload my system by running everything possible (i had youtube playing and email configured and console windows open and app manager running and as many applets as i could running and the only time I managed anything different (beyond slowing down..) was when i was playing a movie which pre-allocates the xv window and prevents clean startup.

I've made a big note of it on liqbase.net and I'm gonna put it to the back of my mind for a while - like other things something will come to light eventually.

As for the (temporary) log, it is stored inside ~/.liqbase/liqbase.run.log
if you open the console and type something like
cp ~/.liqbase/liqbase.run.log ~/MyDocs/liqbase.copy.run.log

a copy of it should become visible within the file explorer.

I'm gonna stop it logging by default because the few that I have seen do not indicate any kind of issue and its so sparodic and unreproducable (it doesn't feel like a memory leak or sometihng because it can hit on very first welcome screen and remains until reopened)

Time to get back on with actual coding I think
If anyone has any more suggestions please add to the bug report on garage.

Multitouch test: what a very, very cool idea. The main problem right now is that the guessed location of my second finger (ext) is still pretty inaccurate, especially on angles. It looks like the problem is that the touchscreen doesn't report the "false finger" exactly in the middle of the two real fingers.

I bet this is going to require you to write a calibration program; you're going to have to write a data collector that records the location of the first finger (thumb) and second finger (finger), and then you'll lift the first finger (thumb) so the screen can "find" the second finger. When I let go of the first finger (thumb), the second finger marker (finger) slides over under my real finger, so you can use this placement as the calibration...

Just babbling, because this is such a neat idea, and I want it so badly...