PDA

View Full Version : RSS screen saver going a bit nutz



dean
2005-08-24, 20:46
I'm seeing the RSS screensaver running on my Linux box here (nightly
as of last night) and it's spinning through feeds (i.e. the top line
is changing) every couple of seconds, while the bottom line is
scrolling relatively normally (actually a bit jumpy.)

Anybody else seeing anything like this?

kdf
2005-08-24, 21:25
On 24-Aug-05, at 8:46 PM, dean blackketter wrote:

> I'm seeing the RSS screensaver running on my Linux box here (nightly
> as of last night) and it's spinning through feeds (i.e. the top line
> is changing) every couple of seconds, while the bottom line is
> scrolling relatively normally (actually a bit jumpy.)
>
> Anybody else seeing anything like this?
>
seems ok to me. The ticker is definitely finishing the full line
before switching to the next provider.

However....last night, i somehow ended up with two boxes on the same
IP, and THAT was causing the ticker to appear very jumpy. They also
alternated reporting loss of server, so i doubt its the same thing.

-kdf

dean
2005-08-24, 21:58
Hm,

Check the mac addresses on those players, are they unique?

On Aug 24, 2005, at 9:25 PM, kdf wrote:

>
> On 24-Aug-05, at 8:46 PM, dean blackketter wrote:
>
>
>> I'm seeing the RSS screensaver running on my Linux box here
>> (nightly as of last night) and it's spinning through feeds (i.e.
>> the top line is changing) every couple of seconds, while the
>> bottom line is scrolling relatively normally (actually a bit jumpy.)
>>
>> Anybody else seeing anything like this?
>>
>>
> seems ok to me. The ticker is definitely finishing the full line
> before switching to the next provider.
>
> However....last night, i somehow ended up with two boxes on the
> same IP, and THAT was causing the ticker to appear very jumpy.
> They also alternated reporting loss of server, so i doubt its the
> same thing.
>
> -kdf
>
>

kdf
2005-08-24, 22:23
On 24-Aug-05, at 9:58 PM, dean blackketter wrote:

> Hm,
>
> Check the mac addresses on those players, are they unique?
>
>
yup. I just had to go through the settings on the SB2 and it grabbed a
new IP. I guess the lease times got mixed up among the players and the
router.
-kdf

Triode
2005-08-25, 12:06
Not seeing it here. If you can reproduce what happens when you press size? It should clear and restart the ticker - does it find
any text to add to line2?

Adrian

> I'm seeing the RSS screensaver running on my Linux box here (nightly as of last night) and it's spinning through feeds (i.e. the
> top line is changing) every couple of seconds, while the bottom line is scrolling relatively normally (actually a bit jumpy.)
>
> Anybody else seeing anything like this?

dean
2005-08-25, 14:49
I'm not seeing it anymore today. The server process had eaten a ton
of memory last night and was sucking CPU. Also thought I had 260,000
songs or so.

A full rescan and a new nightly and it's feeling better, although I'm
still a bit twitchy. Will keep an eye on it.

On Aug 25, 2005, at 12:06 PM, Triode wrote:

> Not seeing it here. If you can reproduce what happens when you
> press size? It should clear and restart the ticker - does it find
> any text to add to line2?
>
> Adrian
>
>
>> I'm seeing the RSS screensaver running on my Linux box here
>> (nightly as of last night) and it's spinning through feeds (i.e.
>> the top line is changing) every couple of seconds, while the
>> bottom line is scrolling relatively normally (actually a bit jumpy.)
>>
>> Anybody else seeing anything like this?
>>
>
>

Triode
2005-08-25, 14:53
> I'm not seeing it anymore today. The server process had eaten a ton
> of memory last night and was sucking CPU. Also thought I had 260,000
> songs or so.

Its a wonder it was still scrolling! The smoothness of scrolling is actually a visual feedback of server heath....

dean
2005-08-25, 15:01
Well, it was jumpy, that I expected from the CPU hogging bit, but the
strange thing was that it was flipping between feeds every couple of
seconds and not updating the bottom line when it did.

On Aug 25, 2005, at 2:53 PM, Triode wrote:

>> I'm not seeing it anymore today. The server process had eaten a
>> ton of memory last night and was sucking CPU. Also thought I had
>> 260,000 songs or so.
>>
>
> Its a wonder it was still scrolling! The smoothness of scrolling
> is actually a visual feedback of server heath....
>

Triode
2005-08-25, 15:30
> Well, it was jumpy, that I expected from the CPU hogging bit, but the strange thing was that it was flipping between feeds every
> couple of seconds and not updating the bottom line when it did.
>

Hum - this could possibly happen if something is repeatedly calling $client->update() while in Rss screensaver mode.

Currently the Rss screensaver schedules its own updates based on the length of text queued up on the ticker. If update is called
(usually only if the size is changed) it cancels the calculated timer and updates the ticker straight away. Now as it is in ticker
mode, the backgound (line1, overlay1, overlay2) are updated immdediately, but the new line 2 is queued onto the end of the ticker
queue - so you don't see it if the queue extends off the screen already.

If you can think of a valid case when something else repeatedly calls update whilst in screensaver mode I will look at avoiding this
case.

dean
2005-08-25, 15:54
So what happens if we're CPU starved while ticking? We're calling
update then and are behind schedule. Can that cause a race condition
that makes us never catch up?
On Aug 25, 2005, at 3:30 PM, Triode wrote:

>> Well, it was jumpy, that I expected from the CPU hogging bit, but
>> the strange thing was that it was flipping between feeds every
>> couple of seconds and not updating the bottom line when it did.
>>
>>
>
> Hum - this could possibly happen if something is repeatedly calling
> $client->update() while in Rss screensaver mode.
>
> Currently the Rss screensaver schedules its own updates based on
> the length of text queued up on the ticker. If update is called
> (usually only if the size is changed) it cancels the calculated
> timer and updates the ticker straight away. Now as it is in ticker
> mode, the backgound (line1, overlay1, overlay2) are updated
> immdediately, but the new line 2 is queued onto the end of the
> ticker queue - so you don't see it if the queue extends off the
> screen already.
>
> If you can think of a valid case when something else repeatedly
> calls update whilst in screensaver mode I will look at avoiding
> this case.
>

Triode
2005-08-25, 16:08
> So what happens if we're CPU starved while ticking? We're calling update then and are behind schedule. Can that cause a race
> condition that makes us never catch up?
> On Aug 25, 2005, at 3:30 PM, Triode wrote:
>

I don't believe so - it would need to be something else repeatedly calling update. Rss actually has two lines functions - one which
is called by anyone calling update and one that it uses itself. Its only the one called by update which could trigger this case.

Anyway as the code that possibly caused this is only there so that when the font size changes the screen doesn't stay blank for
several seconds, I'll see if I can trap that in another way.

Adrian