Font, ScreenSaver & USBNetwork Hacks for Kindle 2.x, 3.x & 4.x

[eschwartz]

KUAL allows you to take action -- any buttons by convention (IIRC) designate something you can do but haven't yet.

[dhdurgee]

Originally Posted by NiLuJe

Keep in mind that the buttons in my KUAL settings are an idiosyncratic sort of radio/action buttons: they show what will be enabled *after* you click on them (i.e., if the button is showing "Do not parse periodicals (Default)", then parsing periodicals *is* currently enabled) .

That, and also that KUAL might take a while to refresh on a K3, and thus that the buttons currently shown might not be in sync with the current settings if you're in a hurry .

I am still seeing some unexpected behavior on occasion with the cover mode screen saver. For example, I opened a new magazine for the first time a few days ago. I was surprised that the magazine cover wasn't displayed when the K3W went to sleep. After noticing this happen a few times I used launchpad to restart the framework. This still failed to get this working. I finally used launchpad to restart the kindle. That did correct the problem.

Is there a cleaner way to restart the cover mode if it stops working? Is there a reason it stops working? Did I miss something?

Dave

[NiLuJe]

The framework restart should be fixed in the latest snapshots.

[dhdurgee]

Originally Posted by NiLuJe

The framework restart should be fixed in the latest snapshots.

So then the framework restart should have fixed this and it is a known issue. Would anything short of restarting the framework correct this? If so, perhaps another button could be added to the KUAL menu for screensavers or a launchpad shortcut could be added to do it.

Dave

[NiLuJe]

@dhdurgee:

The fix I referred to was the fact that the "restart framework" button in KUAL didn't actually restart the ScreenSavers hack too. That's what's been fixed .
And yes, on FW 5.x, that's the only way to make the framework happy when it gets its panties in a bunch about a screensaver.


As for the actual root cause, since it pretty much should never happen, and the hack takes great pains to avoid upsetting the framework if ever something goes wrong for some strange reason, beats me.

Just to double-check, because I just realized you're on a K3 and not something running FW 5.x: what was actually shown as a screensaver when you deemed it broken? A white screen or an outdated cover? An outdated cover is the expected behavior, but a white screen certainly isn't, and, IIRC, the framework is much more forgiving on legacy devices, and will pick up a working image on the next cycle (if the underlying image is actually switched to a working one, that is ^^).

That it would keep showing a white screen would point to a real bug. Which is definitely not out of the question, since cover mode on legacy devices is pretty much a house of cards waiting to be blown out ^^.

(Yup, it's been the day of "Oh god, the FW 2.x/3.x/4.x branch of the code is terrible" ^^).

[dhdurgee]

Originally Posted by NiLuJe

@dhdurgee:

The fix I referred to was the fact that the "restart framework" button in KUAL didn't actually restart the ScreenSavers hack too. That's what's been fixed .
And yes, on FW 5.x, that's the only way to make the framework happy when it gets its panties in a bunch about a screensaver.


As for the actual root cause, since it pretty much should never happen, and the hack takes great pains to avoid upsetting the framework if ever something goes wrong for some strange reason, beats me.

Just to double-check, because I just realized you're on a K3 and not something running FW 5.x: what was actually shown as a screensaver when you deemed it broken? A white screen or an outdated cover? An outdated cover is the expected behavior, but a white screen certainly isn't, and, IIRC, the framework is much more forgiving on legacy devices, and will pick up a working image on the next cycle (if the underlying image is actually switched to a working one, that is ^^).

That it would keep showing a white screen would point to a real bug. Which is definitely not out of the question, since cover mode on legacy devices is pretty much a house of cards waiting to be blown out ^^.

(Yup, it's been the day of "Oh god, the FW 2.x/3.x/4.x branch of the code is terrible" ^^).

The problem I was encountering was an outdated cover, not a white screen. After several sleep/wake cycles the cover was still outdated, so I used launchpad to restart the framework in hopes of getting it fixed. When that failed I used launchpad to restart the entire kindle, which did fix the problem.

If a restart is all that will fix this, I guess I will have to live with it.

Dave

[NiLuJe]

@dhdurgee: Oh, launchpad. Yep, that'll never do the trick . The KUAL button will, though (in the snapshots, that is).

As for the root cause of the issue, if switching between books (and waiting the appropriate amount of time) doesn't help, then it's still probably a bug. Do you happen to have a weird USB setup? Or can pinpoint a significant (or not) detail before the desync happens?
If you can, when that happens, try to check the logs and/or check that the cover watchdog is alive (remember, house of cards, I might have missed a corner case).

Also keep in mind that there's a known race condition (in this branch of the code) where rarely, the first sleep cycle on a new cover will either use the previous cover, or a mangled mix of the two. There's a KUAL button to enable a workaround, which isn't the default because it has worse cosmetic downsides (IMHO) . That race condition *might* have been fixed in the latest snapshots, but I wouldn't hold my breath.

[dhdurgee]

Originally Posted by NiLuJe

@dhdurgee: Oh, launchpad. Yep, that'll never do the trick . The KUAL button will, though (in the snapshots, that is).

As for the root cause of the issue, if switching between books (and waiting the appropriate amount of time) doesn't help, then it's still probably a bug. Do you happen to have a weird USB setup? Or can pinpoint a significant (or not) detail before the desync happens?
If you can, when that happens, try to check the logs and/or check that the cover watchdog is alive (remember, house of cards, I might have missed a corner case).

Also keep in mind that there's a known race condition (in this branch of the code) where rarely, the first sleep cycle on a new cover will either use the previous cover, or a mangled mix of the two. There's a KUAL button to enable a workaround, which isn't the default because it has worse cosmetic downsides (IMHO) . That race condition *might* have been fixed in the latest snapshots, but I wouldn't hold my breath.

I don't have anything wierd about my K3W that I am aware of. I did enable verbose mode and I seem to stop seeing messages from the watchdog when it stops working. Is there a way to check the logs and status of the watchdog from a myts session?

I had assumed that the framework restart in KUAL and launchpad were equally functional, I will start using the KUAL one next time.

Let me know of anything further I ca do to help you sort this out. I guess K3 support is of lesser importance at this point as they are no longer made, but there are still quite a few out here. In terms of function I can do more with my K3W than with the new basic kindle I received as a Christmas gift, which has no audio support and requires major work to enhance. In many ways the new kindles are a step backwards.

Dave

[NiLuJe]

ps faux | grep watchdog should give you a hint of what's alive, although that won't tell you if it's hung or deadlocked (which was usually what happened).

That could only be diagnosed by watching the logs live (showlog -f), which pretty much means over SSH, since the UI needs to be free to be used 'normally' .

[Glorfindel]

I thought periodical covers don't show regardless?

[dhdurgee]

Originally Posted by Glorfindel

I thought periodical covers don't show regardless?

They don't by default, but there is a toggle that KUAL can flip. From comments earlier in the thread there are also some periodicals that don't follow the convention the watchdog expects and then it fails to work. In my case both periodicals follow the expected guidelines and the cover mode works as designed with them.

Dave

[Glorfindel]

I see, why not leave it off? Not much of a fix to turn off what doesn't work I know, but still

[NiLuJe]

@Glorfindel: leave what off?

[Glorfindel]

The on/off toggle for periodical covers.

[NiLuJe]

Because it technically works when the input data is adequate .

The issue is that, for most of the test files I had on hand at the time (especially newspapers), there's no proper cover, but there *is* something: the masthead. Which is clearly not made for that, and when it ends up being used as a screensaver by the hack, it's terrible.