|
Tuesday, 19 January 2010 23:45 |
|
Today I released my first program for Maemo on the N900.
With www2sms you can send SMS via the webinterface of O2 without actually using the webinterface.
It's designed to easily patch in new providers. Patches are welcome :-)
Currently it will not work on Maemo, because "mechanize" for python is missing on Maemo.
Thought you can run it on any platform that runs Python and QT
Everything except the scheduling works. If someone finds a quick solution, patches are welcome.
or write to '"maemo" at "olausson.de"'
|
|
Maemo SDK running on Gentoo-Linux |
 |
 |
 |
|
Friday, 18 December 2009 11:07 |
|
It's amazing!
 (Click image to enlarge)Now it's time to port some of my favorite CLI applications to Maemo.
The Nokia N900 with Maemo-Linux is, what I was waiting for since I started to use Linux.
Three cheers for OSS (OpenSourceSoftware).
|
|
Thursday, 10 December 2009 20:59 |
|
This describes how to configure the SIP account on Maemo (on Nokia N900) to work with 1&1.
Username: {Your1und1PhoneNumber}@sip.1und1.de (add the "international area code" to your number but without the "+")
Password: {YourVoIPPassword}
"Use for telephone numbers" Selected
"User name" Leave blank
"Transport" Auto (or UDP) works both
"Outbound proxy" Leave blank
"Port" 5060
"Discover public address" Selected
"Loose routing" NOT selected
"Keep-alive mechanism" Auto
"Keep-alive period" Auto
"Autodetect STUN" Auto (or stun.1und1.de)
- Here are some additional information:
realm: 1und1.de
STUN-Server: stun.1und1.de
Sip-Server: sip.1und1.de
Ports (UDP): 5070 - 5079 and 30000-30019
|
|
Sunday, 06 December 2009 22:14 |
|
Some things i came across when I was setting up a new Linux KVM host:
- Which scheduler on the host ist the one to go for?
I was playing around with the available schedulers but thankfully someone else did that before.
See Bugreport #505695 on the RedHat Bugzilla.
So apparently the best suited scheduler for a KVM host is the "deadline" scheduler.
- Which scheduler on the VM ist the on to go for?
It states, that with the "noop" scheduler, the host is able to optimize the I/O request, which is what we want.
Further, the RedHat Bugreport #505695 suggest to "put virtio-blk devices in rotational mode". At least I couldn't find a kernel parameter to enable this at boot time (if someone knows, please let me know),
so this has to be done on every VM by hand (or add the following to a start script like /etc/conf.d/local.start):
# for device in /sys/block/vd*/queue/rotational ; do echo "1" > ${device} ; done
The VMs should be started without a "drive cache". For "qemu-kvm" start the VMs with
# -drive cache=off
And don't forget to enable "KSM" for page merging in Kernel 2.6.32 and later. Why? See here.
|
|
Advertisement can, but must not, be annoying |
|
|
|
|
Monday, 23 November 2009 23:27 |
|
Many Sites use advertisement to get some revenue for the service they provide.
So far this is not bad at all.
If I like the site and the advertising is not to intrusive I tend to click here and there on a banner/link to support the site.
But many sites have those annoying blinking, flickering, shouting, music playing, and and and Ads I can't stand.
Thought I am using Opera, there is noting like " Adblock Plus" for Firefox. Okay, Opera has a build in Ad filter, but manually adding every banner is a lifetime job. Thankfully there are two Sites (I know) which collect hostnames/IPs of servers known to host banner and other advertising stuff.
One system-wide solution would be, to add those servers to the hostfile and redirect the DNS request.
This can easily done (at least on Linux) with a ready to use script on http://hostsfile.mine.nu/.
Another approach I describe here here, is to use the urlfilter.ini file in Opera.
Just download a recent list in Opera "urlfilter.ini" format from http://hostsfile.mine.nu/ and replace the existing one in your Opera directory. To automate this and stay up to date I hacked this little python script together.
It will fetch the list, merge it with existing entries in the urlfilter.ini file and save the new list. You can enter patterns which should be removed from the list.
Just run the script (via cron) weekly and surfing the web is fun again.
From time to time you should check the file cause currently my script only merges in new lines, but does not remove hosts which have been removed from the adserver list.
(It might do in the next version)
Get it here:
opera_blocklist.py
|
|
|
Friday, 13 November 2009 15:21 |
|
I made it a bit more intelligent.
- can distinguish between MD/Mini and Minimization only output.
- switched from fixed column formating to white spaced formating
New version:
|
|
Linux - Sane - Epson V300 Photo |
|
|
|
|
Sunday, 08 November 2009 22:40 |
|
And finally I found one...
To all those scanner seeking Linux users:
It's supported via the sane epkowa backend.
One drawback:
The buttons are not working and for now I had no time to look into this, but a working "copy" Button would be nice :-)
|
|
YOUR BATTERY IS EMPTY - NAG |
|
|
|
|
Saturday, 24 October 2009 08:28 |
|
You know what pisses me really off?
A mobile phone with such a crappy, intrusive, annoying and energy wasting firmware like the one in the Samsung X820.
I am talking about the useless "Your battery is empty" notification.
My phone is set to vibrate in 99.9% of the time cause I don't want it to be a pain in my neck.
(Despite the fact, that the "vibrator" is way to intense and loud it is still better than any sound)
But what happens when your battery starts to loose power?
No, the phone does not silently die, it vibrates for 1 3 second every ~10 5 minutes for 2 hours!
It could run a few hours longer, if it wouldn't try to tell me it is "empty", which it obviously is not, in the described way!
I guess you can imaging many circumstances where this "FEATURE" is more, much more annoying than a dead phone!
Is it so hard to give the user control over it's phone and let him choose on which events he will be notified?
I guess not.
So if any mobile-phone-fimware-developer is reading that:
PLEASE GIVE THE USER THE POSSIBILITY TO CONTROL ALL "NOTIFICATIONS"
|
|
Intel Core i7 920 vs. AMD Istanbul 2427 |
|
|
|
|
Sunday, 18 October 2009 14:10 |
|
Last night I was wondering how the Core i7 architecture compares to the AMD Istanbul architecture when running molecular dynamics (MD) simulations with NAMD.
I compiled a "multicore" version of NAMD 1.7b1 with ICC (11.1.056) for both systems --> Linux-x86_64-icc.multicore-linux64-icc.
| Cores |
Core i7 (2668 MHz) |
Istanbul (2211 MHz) |
Difference |
| seconds/step |
| 1 |
1.40421 |
2.00233 |
30% |
| 2 |
0.736545 |
1.02655 |
28% |
| 3 |
0.500922 |
0.700323 |
28% |
| 4 |
0.369021 |
0.521004 |
29% |
| 5* |
0.370917 |
0.427389 |
13% |
| 6* |
0.358904 |
0.35105 |
-2% |
| 7*+ |
0.340182 |
0.293715 |
-15% |
| 8*+ |
0.305315 |
0.259213 |
-18% |
(Seconds/step are taken from a single run after 500 steps)
*) Virtual Cores (HT) on i7 PC
+) Second CPU on Istanbul node
Intel Core i7 920 beats AMD Istanbul 2427 on 5 cores and comes close on 6 cores while KDE is up and running Amarok (mp3 player), Browser and the usual
other stuff like Kopete, Skype, etc. , whereas the Istanbul is a dedicated compute node. The frequency difference is 17% percent, so even with that in mind,
the i7 comes very close to 5 native Istanbul cores!
Hit "Rerad more..." to see the raw data
|
|
anamo.py and anaout.py updated |
|
|
|
|
Wednesday, 30 September 2009 09:50 |
|
When using anamo.py/anaout.py with wildcards (-f "file*") the file order was as the files are stored on the hard disk.
Fixed it by sorting the resulting list:
for element in chmout : if '*' in element : chmout.extend(glob.glob(element)) chmout.remove(element) chmout.sort()
New version:
anamo.py v1.4
anaout.py v1.2
|
|
