PDA

View Full Version : SlimServer 5.1 final candidate



dean
2004-01-26, 08:10
Hi all,

The latest nightly release of SlimServer here:

http://www.slimdevices.com/downloads/nightly/latest

Is a final candidate for our 5.1 release. Everybody's worked hard to
find and fix the bugs in SlimServer as well as add a bunch of new
features.

Before I pull the plug and post it as the default download, I'd love to
hear about any problems with this release. If things are all clear,
we'll post it as an official release.

Thanks so much,

dean

Caleb Epstein
2004-01-26, 08:34
On Mon, Jan 26, 2004 at 07:10:03AM -0800, dean blackketter wrote:

> Hi all,
>
> The latest nightly release of SlimServer here:
>
> http://www.slimdevices.com/downloads/nightly/latest
>
> Is a final candidate for our 5.1 release. Everybody's worked hard to
> find and fix the bugs in SlimServer as well as add a bunch of new
> features.
>
> Before I pull the plug and post it as the default download, I'd love to
> hear about any problems with this release. If things are all clear,
> we'll post it as an official release.

This is relatively minor, but the elapsed/remaining time
displays for FLAC (and probably anything streamed via
convert.conf) get messed up if you turn off the Squeeze and
turn it back on at a later point. The remain time gets much
larger than the actual song time (I was seeing numbers like
-70:09 for a 10 minute song).

--
Caleb Epstein | bklyn . org | Just go ahead and write your own multitasking
cae at | Brooklyn Dust | multiuser os! Worked for me all the times.
bklyn dot org | Bunny Mfg. | -- Linus Torvalds

Jeff Shanholtz
2004-01-26, 09:19
On Mon, 26 Jan 2004 15:28:15 -0000, Ian Crockford wrote:

> All looks great, except the fast-forward and fast-reverse still doesn't work
> -- the track skips to an undetermined point when normal play resumes. This
> has been the case since the advent of 5.0. Worked ok in 4.x.

I just emailed Dean about the same thing. In addition, the song's time
display can get out of sync with where it actually is in the song. I
reported this problem with the 12/26 nightly build and he said he'd get
back to me when he had something to test but hasn't so far. I would think
this would be a show-stopper because it's such a basic feature to playing
music. I'm actually surpised it's not a more widely complained about bug
(unless it doesn't happen for everyone for some reason).

Jack Coates
2004-01-26, 10:05
On Mon, 2004-01-26 at 07:10, dean blackketter wrote:
> Hi all,
>
> The latest nightly release of SlimServer here:
>
> http://www.slimdevices.com/downloads/nightly/latest
>
> Is a final candidate for our 5.1 release. Everybody's worked hard to
> find and fix the bugs in SlimServer as well as add a bunch of new
> features.
>
> Before I pull the plug and post it as the default download, I'd love to
> hear about any problems with this release. If things are all clear,
> we'll post it as an official release.
>
> Thanks so much,
>
> dean

Seems to have fixed the issue with XMMS hanging up, which makes me
happy. It's played three albums straight through without stopping, and
XMMS versions 1.2.7 through 1.2.9-pre1 would all hang after four or five
songs at most with the 1-19 build.

thanks,
--
Jack at Monkeynoodle Dot Org: It's A Scientific Venture...
************************************************** ********************
* "Stark raving naked in the fornication nation!" *
* -- Debbie Gibson Is Pregnant With My Two-Headed Love Child from *
* Root Hog or Die by Mojo Nixon & Skid Roper *
************************************************** ********************

Mark A. Aiken
2004-01-26, 10:25
I'm still experiencing the following glitch, which I have mentioned
before:

Under WinXP, refreshing the web interface causes very high CPU usage. By
refreshing just a few times quickly, it's possible to peg out my 2GHz P4 and
cause the SB to skip.

In case anyone thinks this is a strange geeky bug, my wife was tinkering
with our playlist from the laptop a few nights ago and remarked on her own
that when she used the web interface it caused the SB to skip. This caused
her to experience the SB as flaky.

I've been avoiding the web interface and using the remote to control the
SB instead, which is quite usable, but a shame, since the web interface is
quite attractive.

I would think this bears some investigation; presumably, nothing this
compute-intensive should be going on while rendering the web interface! I am
happy to help debug however I can.

Mark

----- Original Message -----
From: "dean blackketter" <dean (AT) slimdevices (DOT) com>
To: "Slim Devices Discussion" <discuss (AT) lists (DOT) slimdevices.com>
Sent: Monday, January 26, 2004 7:10 AM
Subject: [slim] SlimServer 5.1 final candidate


> Hi all,
>
> The latest nightly release of SlimServer here:
>
> http://www.slimdevices.com/downloads/nightly/latest
>
> Is a final candidate for our 5.1 release. Everybody's worked hard to
> find and fix the bugs in SlimServer as well as add a bunch of new
> features.
>
> Before I pull the plug and post it as the default download, I'd love to
> hear about any problems with this release. If things are all clear,
> we'll post it as an official release.
>
> Thanks so much,
>
> dean
>
>

Jack Coates
2004-01-26, 10:55
On Mon, 2004-01-26 at 09:25, Mark A. Aiken wrote:
> I'm still experiencing the following glitch, which I have mentioned
> before:
>
> Under WinXP, refreshing the web interface causes very high CPU usage. By
> refreshing just a few times quickly, it's possible to peg out my 2GHz P4 and
> cause the SB to skip.
>
> In case anyone thinks this is a strange geeky bug, my wife was tinkering
> with our playlist from the laptop a few nights ago and remarked on her own
> that when she used the web interface it caused the SB to skip. This caused
> her to experience the SB as flaky.
>
> I've been avoiding the web interface and using the remote to control the
> SB instead, which is quite usable, but a shame, since the web interface is
> quite attractive.
>
> I would think this bears some investigation; presumably, nothing this
> compute-intensive should be going on while rendering the web interface! I am
> happy to help debug however I can.
>
> Mark
....

high CPU usage does occur when the page refreshes; however, I've been
unable to make it actually cause a skip in the music using Linux and
XMMS as a software player on the same machine, while also running Win2k
and Lotus Notes in a VMWare session, Gaim, Cervisia, Mozilla, a handful
of aterms, three gvim sessions and a debug instance of my company's
product. I have 512M of RAM and a 3637 song database.

free -m
total used free shared buffers
cached
Mem: 503 498 4 0 18
296
-/+ buffers/cache: 183 319
Swap: 1176 225 951

model name : Intel(R) Pentium(R) 4 CPU 2.40GHz

--
Jack at Monkeynoodle Dot Org: It's A Scientific Venture...
************************************************** ********************
* "I have counted all the lines on the road between here and Los *
* Angeles that pass straight below, and I am dreaming of a light and *
* it comes from your head." *
* -- The Light From A Cake from Key Lime Pie by Camper van Beethoven *
************************************************** ********************

kdf
2004-01-26, 11:25
Quoting "Mark A. Aiken" <mark_aiken (AT) hotmail (DOT) com>:

> I'm still experiencing the following glitch, which I have mentioned
> before:
>
> Under WinXP, refreshing the web interface causes very high CPU usage. By
> refreshing just a few times quickly, it's possible to peg out my 2GHz P4 and
> cause the SB to skip.
>
> In case anyone thinks this is a strange geeky bug, my wife was tinkering
> with our playlist from the laptop a few nights ago and remarked on her own
> that when she used the web interface it caused the SB to skip. This caused
> her to experience the SB as flaky.
>
> I've been avoiding the web interface and using the remote to control the
> SB instead, which is quite usable, but a shame, since the web interface is
> quite attractive.
>
> I would think this bears some investigation; presumably, nothing this
> compute-intensive should be going on while rendering the web interface! I am
> happy to help debug however I can.

I'm curious, do you have any of the caching settings on? Even with them off and
using the Fishbone skin, I've never managed to skip the music. The CPU usage
will go higher due to the much greater activity to collect and organise the data
for a given page. Turning on the html caching does make a difference. How much
ram you have in your system is also important. The SlimServer consumes a large
chunk of memory, so if you cause a page swap it can get ugly, especially under
Windows.

-kdf

Mark A. Aiken
2004-01-26, 11:30
WinXP's scheduling may be crummier, or whatever the expensive operation
is may be more expensive under XP, but it's interesting to hear that this
happens as well under *nix, though.

Like I said, isn't it pretty suspicious that it takes this kind of
effort to render the web UI?

Mark

----- Original Message -----
From: "Jack Coates" <jack (AT) monkeynoodle (DOT) org>
To: "Slim Devices Discussion" <discuss (AT) lists (DOT) slimdevices.com>
Sent: Monday, January 26, 2004 9:55 AM
Subject: [slim] SlimServer 5.1 final candidate


> On Mon, 2004-01-26 at 09:25, Mark A. Aiken wrote:
> > I'm still experiencing the following glitch, which I have mentioned
> > before:
> >
> > Under WinXP, refreshing the web interface causes very high CPU
usage. By
> > refreshing just a few times quickly, it's possible to peg out my 2GHz P4
and
> > cause the SB to skip.
> >
> > In case anyone thinks this is a strange geeky bug, my wife was
tinkering
> > with our playlist from the laptop a few nights ago and remarked on her
own
> > that when she used the web interface it caused the SB to skip. This
caused
> > her to experience the SB as flaky.
> >
> > I've been avoiding the web interface and using the remote to control
the
> > SB instead, which is quite usable, but a shame, since the web interface
is
> > quite attractive.
> >
> > I would think this bears some investigation; presumably, nothing
this
> > compute-intensive should be going on while rendering the web interface!
I am
> > happy to help debug however I can.
> >
> > Mark
> ...
>
> high CPU usage does occur when the page refreshes; however, I've been
> unable to make it actually cause a skip in the music using Linux and
> XMMS as a software player on the same machine, while also running Win2k
> and Lotus Notes in a VMWare session, Gaim, Cervisia, Mozilla, a handful
> of aterms, three gvim sessions and a debug instance of my company's
> product. I have 512M of RAM and a 3637 song database.
>
> free -m
> total used free shared buffers
> cached
> Mem: 503 498 4 0 18
> 296
> -/+ buffers/cache: 183 319
> Swap: 1176 225 951
>
> model name : Intel(R) Pentium(R) 4 CPU 2.40GHz
>
> --
> Jack at Monkeynoodle Dot Org: It's A Scientific Venture...
> ************************************************** ********************
> * "I have counted all the lines on the road between here and Los *
> * Angeles that pass straight below, and I am dreaming of a light and *
> * it comes from your head." *
> * -- The Light From A Cake from Key Lime Pie by Camper van Beethoven *
> ************************************************** ********************
>
>

Jack Coates
2004-01-26, 12:00
On Mon, 2004-01-26 at 09:05, Jack Coates wrote:
> On Mon, 2004-01-26 at 07:10, dean blackketter wrote:
> > Hi all,
> >
> > The latest nightly release of SlimServer here:
> >
> > http://www.slimdevices.com/downloads/nightly/latest
> >
> > Is a final candidate for our 5.1 release. Everybody's worked hard to
> > find and fix the bugs in SlimServer as well as add a bunch of new
> > features.
> >
> > Before I pull the plug and post it as the default download, I'd love to
> > hear about any problems with this release. If things are all clear,
> > we'll post it as an official release.
> >
> > Thanks so much,
> >
> > dean
>
> Seems to have fixed the issue with XMMS hanging up, which makes me
> happy. It's played three albums straight through without stopping, and
> XMMS versions 1.2.7 through 1.2.9-pre1 would all hang after four or five
> songs at most with the 1-19 build.
>
> thanks,


I spoke too soon, XMMS hung at 3 hours, 19 minutes. The logfile is
empty. This is probably an XMMS bug, but I'm not sure.
--
Jack at Monkeynoodle Dot Org: It's A Scientific Venture...
************************************************** *********************
*"And did you know desire's a terrible thing, the worst that I *
*can find? And did you know desire's a terrible thing, but I rely *
*on mine?" *
*-- Can't Be Sure from Reading, Writing, and Arithmetic by The Sundays*
************************************************** *********************

Mark A. Aiken
2004-01-26, 12:09
I have tag and HTML caching turned on. Are you running under Windows?
Since no one else seems to be complaining about this at the moment, I'm
perfectly willing to believe this is only a skipping-caliber problem under
Windows. Note also that I'm streaming uncompressed audio, making me more
sensitive to skips in the first place.

In case anyone thinks I'm complaining about a little CPU blip when I
refresh a page, consider again: hitting refresh three times quickly on the
web interface will peg my CPU utilization at 100% for a few *seconds*. This
will cause the SB to stop playing for a second or two. That's not Good.
Consider also that casual web interface use by a non-geek causes skips. This
morning I verified the effect after upgrading the server again and my
playlist had about 10 songs in it. The CPU spiking effect happens with
*empty* playlists, too, as I previously posted.

I'm running a 2GHz P4 with 640MB of RAM. My CPU load while, say,
decompressing FLACs, is background noise. I use this machine for Photoshop
editing. Whatever the Slimserver is doing to chew up this much CPU usage
cannot possibly be correct, IMHO.

I know Perl, so I can spend some time debugging this myself if someone
can answer the following:

- Where is the main invocation point for rendering the web UI, so I can
start tracing?

- Under XP, changing the Perl source in the installation directory
doesn't seem to affect the server, even after restarting it. Is the Perl
code actually rolled up into the .exe? If so, is there any easy way for me
to tinker with the server source?

Mark

----- Original Message -----
From: "kdf" <slim-mail (AT) deane-freeman (DOT) com>
To: "Slim Devices Discussion" <discuss (AT) lists (DOT) slimdevices.com>
Sent: Monday, January 26, 2004 10:25 AM
Subject: [slim] SlimServer 5.1 final candidate


> Quoting "Mark A. Aiken" <mark_aiken (AT) hotmail (DOT) com>:
>
> > I'm still experiencing the following glitch, which I have mentioned
> > before:
> >
> > Under WinXP, refreshing the web interface causes very high CPU
usage. By
> > refreshing just a few times quickly, it's possible to peg out my 2GHz P4
and
> > cause the SB to skip.
> >
> > In case anyone thinks this is a strange geeky bug, my wife was
tinkering
> > with our playlist from the laptop a few nights ago and remarked on her
own
> > that when she used the web interface it caused the SB to skip. This
caused
> > her to experience the SB as flaky.
> >
> > I've been avoiding the web interface and using the remote to control
the
> > SB instead, which is quite usable, but a shame, since the web interface
is
> > quite attractive.
> >
> > I would think this bears some investigation; presumably, nothing
this
> > compute-intensive should be going on while rendering the web interface!
I am
> > happy to help debug however I can.
>
> I'm curious, do you have any of the caching settings on? Even with them
off and
> using the Fishbone skin, I've never managed to skip the music. The CPU
usage
> will go higher due to the much greater activity to collect and organise
the data
> for a given page. Turning on the html caching does make a difference.
How much
> ram you have in your system is also important. The SlimServer consumes a
large
> chunk of memory, so if you cause a page swap it can get ugly, especially
under
> Windows.
>
> -kdf
>
>

kdf
2004-01-26, 12:45
Quoting "Mark A. Aiken" <mark_aiken (AT) hotmail (DOT) com>:

>
> I know Perl, so I can spend some time debugging this myself if someone
> can answer the following:
>
> - Where is the main invocation point for rendering the web UI, so I can
> start tracing?
>
> - Under XP, changing the Perl source in the installation directory
> doesn't seem to affect the server, even after restarting it. Is the Perl
> code actually rolled up into the .exe? If so, is there any easy way for me
> to tinker with the server source?
>
There is a windows EXE which compiles all required modules into the executable.
If you have a Perl environment installed, you can use the .zip version which
has complete source. Web UI is handled by Web/HTTP.pm and Web/Pages.pm for the
most part. run slimserver.pl with --d_http for command-line option to track the
debug out put for the web engine.

-kdf