|
|
About
/proc/self/jd, News from the Nix Nut.
Subscribe
Subscribe to a syndicated feed of my weblog,
brought to you by the wonders of RSS.
Flavours
There's more than one way to view this weblog; try these flavours on
for size.
|
|
|
Most common commands
For today's presentation
http://www.oreillynet.com/onlamp/blog/2007/01/whats_in_your_bash_history.html
Work OSX Workstation:
john-lightseys-computer:~ jd$ cut -f1 -d" " .bash_history | sort | uniq -c | sort -nr | head -n 20
88 ssh
88 ls
82 exit
60 cd
30 sftp
30 dig
27 vi
20 rm
11 unzip
11 mkdir
7 less
5 man
4 mv
4 cat
3 zip
3 perldoc
3 lp
2 svn
2 sudo
2 ps
Work Development Server (jd account)
jd@jd.dev.cpanel.net:~$ cut -f1 -d" " .bash_history | sort | uniq -c | sort -nr | head -n 20
222 ls
208 cd
99 sudo
96 exit
53 mv
52 rm
37 svn
29 fresh-rpmbuild-tree.pl
28 fakeroot
22 vi
17 rpm
15 nm
14 ps
13 cp
12 grep
10 less
9 rpm2cpio
8 ldd
7 man
5 cat
Work Development Server (root account)
root@jd.dev.cpanel.net:~# cut -f1 -d" " .bash_history | sort | uniq -c | sort -nr | head -n 20
1386 svn
916 vi
776 cd
665 ls
151 cpgrep
107 rm
75 less
62 make
47 man
45 fg
41 dig
40 ps
37 perlcheck
34 exit
32 tar
31 zonec
30 diff
27 cp
24 /scripts/rebuildhttpdconf
23 /scripts/easyapache
Home Desktop (recently reinstalled, user account)
jd@home:~$ cut -f1 -d" " .bash_history | sort | uniq -c | sort -nr | head -n 20
192 ls
46 cd
41 vi
37 rm
26 sudo
19 apt-cache
18 exit
13 du
12 mv
9 ./quick.pl
9 dig
8 tar
8 java
6 ssh
6 sh
6 ps
4 less
3 wc
3 savage2.sh
2 which
Home Desktop (recently reinstalled, root account)
home:~# cut -f1 -d" " .bash_history | sort | uniq -c | sort -nr | head -n 20
251 ls
61 cd
50 rm
22 du
16 cp
13 mv
12 apt-get
10 exit
8 ps
7 mount
5 umount
5 df
5 chown
4 vi
3 less
3 /etc/init.d/network-manager
3 cat
2 /usr/sbin/NetworkManager
2 mkdir
2 man
Laptop (user account)
jd@macbook:~$ cut -f1 -d" " .bash_history | sort | uniq -c | sort -nr | head -n 20
132 ls
47 sudo
40 exit
37 cd
30 svn
23 ssh
23 rm
21 gnomevfs-ls
17 apt-cache
14 vi
14 nd
9 ps
8 man
6 sftp
6 mv
6 dig
6 cat
5 gpg
3 top
3 tar
Laptop (root account)
macbook:~# cut -f1 -d" " .bash_history | sort | uniq -c | sort -nr | head -n 20
66 ls
49 vi
45 chroot
43 cd
36 /etc/init.d/gdm
33 apt-get
17 less
16 xinit
16 exit
15 ps
14 /etc/init.d/flumotion
12 sudo
12 /sbin/ifconfig
12 apt-cache
11 man
11 dpkg
6 kill
5 which
5 ssh
5 reboot
Combined Totals:
1616 ls
1455 svn
1231 cd
1073 vi
296 exit
289 rm
186 sudo
151 cpgrep
122 ssh
116 less
94 ps
88 mv
86 dig
80 man
62 make
56 cp
48 apt-cache
45 apt-get
45 chroot
45 fg
43 tar
37 perlcheck
36 sftp
36 /etc/init.d/gdm
35 du
31 zonec
30 diff
29 fresh-rpmbuild-tree.pl
28 fakeroot
24 /scripts/rebuildhttpdconf
23 /scripts/easyapache
21 gnomevfs-ls
18 cat
17 rpm
16 xinit
15 nm
14 nd
14 /etc/init.d/flumotion
13 mkdir
12 grep
12 /sbin/ifconfig
11 unzip
11 dpkg
9 rpm2cpio
9 ./quick.pl
8 ldd
8 java
7 which
7 mount
6 kill
6 sh
5 df
5 chown
5 umount
5 reboot
5 gpg
3 top
3 wc
3 /etc/init.d/network-manager
3 lp
3 savage2.sh
3 zip
3 perldoc
2 /usr/sbin/NetworkManager
[]
permanent link
Grep Switches
The LugNut isn't going to make it to the HLUG meeting on Saturday May 10th, so Pete asked me to put something together for the presentation. I'm going to do a talk about the most common commands in my bash history file, meaning we'll probably talk quite a bit about grep, svn, my favorite text editor, and file/directory manipulation.
Catch it live at 2:00pm at HAL-PC in Houston, or watch the live video feed on this server.
[]
permanent link
Retooling
I spent last weekend completely reorganizing my AMD64 desktop system. It had 1GB of RAM, 500GB of drive space over four SATA-150 drives (RAID 1,0...I'm not that crazy) and a Debian Sid AMD64 install that predated Debian support for AMD64 by quite some time. It was a big mess and had been driving me to use a laptop for everything.
After a full weekend of work, the machine has a fresh Lenny install, 4GB of RAM, a single 1TB SATA-300 drive and a vastly reorganized home directory. I managed to get rid of 200GB of garbage in the process.
Now I just need to find time to fix all the castle-combat bugs that have accumulated lately.
[]
permanent link
New Server
After much deliberation I finally took the plunge and moved my domains from the dedicated sever at Rackshack/EV1/ThePlanet that I've been using since 2003 to one in the Netherlands at Leaseweb. The price was more affordable, the hardware is significantly better, and Debian was offered. So far I'm quite pleased with the responsiveness of the machine. Considering that the old box was 20 miles away and the new one is halfway around the world, it's quite impressive.
We'll have to see if anyone at HLUG notices the video archive responding slower.
[]
permanent link
Back from Debconf
Lots of fun in Helsinki, but plenty of work left to do. Right now I'm
fighting with these two stupid IDE drives full of video files I brought
back. One keeps powering off for some reason and I'm concerned that it
may give up the ghost. There are about three days worth
of talks left to be done and I have the only copy of this material.
I'll pick up a 500GB drive to make another copy after I get back from the
HLUG meeting tomorrow.
[]
permanent link
The Hell Known as DRI CVS
I've been trying and trying to build a new version of the
dri-trunk-sid packages but each
DRI CVS checkout I've tried has one problem or another. All versions of DRI
after 2005-04-20 have a nasty Radeon bug that makes some textures flash
black and white. CVS versions from 2005-04-17 to 2005-04-20 have a compilation
bug in posix_tty.c that I really don't feel like isolating. I'm working my
way back from that date to find the last good CVS checkout. You might say
"so what if there is a Radeon rendering bug, DRI covers many other video
chips." Well, out of 100-200 different emails commenting on the dri-trunk-sid
packages only a handful were related to anything other than ATI Radeons.
There are a few people using these packages with old ATI Mach64 cards, VIA,
Savage, and Matrox, but Radeon is the favorite by an order of magnitude.
Any new version of these packages that has obvious Radeon rendering problems
will fill my email inbox with an endless stream of complaints.
I'd like to fix drm-build so that dri-trunk-sid will work properly on PPC and
AMD64 machines. I'd also like to package up driconf, but damn it's taking a
long time to find a working CVS checkout date.
[]
permanent link
Kino with Samsung SCD103
I've had frequent problems importing video over firewire from my SCD103 DV
camera. Today I think I found a definitive solution. Using Linux kernel
2.6.11 with the dv1394 module, I can get Kino's AVC controls to turn on and
play, fast-forward, rewind, etc. I can't get Kino to start capturing from
this camera by simply pressing the "capture" button, but it turns out that's
not as big of a problem as I had previously imagined. The simple solution is
to activate the AVC controls, hit the play button in kino, wait for the camera
to start playing, then hit the capture button. No more iMovie for me.
[]
permanent link
More DVD Madness
Well, it turns out that my last batch of DVDs did have audio sync issues.
It's difficult to spot these problems without playing into a disc for an
hour or so, but after waiting and watching the problem was still there.
I finally managed to produce a DVD that looks and sounds right. I had to go
back and export both mpeg video and mpeg audio directly from Cinelerra, then
use tcmplex to mux the streams together, but the end result looks good. This
leads me to conclude that either (a) Cinelerra is exporting a bad Quicktime
DV file, or more likely (b) mplex is messing up the final mpeg file. Since I
have a good iso for the last presentation, I'm not going to mess around with it
any more. For the next presentation I'll try doing everything normally except
using tcmplex for the mux.
[]
permanent link
DVD Audio Sync Issues Resolved
After far too many hours of poking, proding, waiting and searching, I believe
the answer to my DVD problems has appeared. The workflow that seems to result
in FUNCTIONAL DVDs that don't show horrible, seconds long, audio lag after
playing for an hour goes like this:
- Import from DV camera with kino
- Load .mov files into cinelerra
- Edit and export as "Quicktime for Linux" selecting DV video and audio
- Convert the resulting .mov into an avi: mencoder -o [output.avi] -ovc lavc -oac lavc -lavcopts acodec=mp3 -srate 48000 [input.mov]
- Convert the avi into mpeg audio and video: transcode -i [avi-filename] -y ffmpeg --export_prof dvd-ntsc --export_asr 2 -o [output basename] -D0 -s2 -m [output basename].ac3 -J modfps -export_fps 29.97
- use mplex to create the vob files: mplex -f 8 -o [basename].vob [basename].m2v [basename].ac3
- Create a dvdauthor xml input file by hand or using qdvdauthor
- use dvdauthor to create the directory structure: dvdauthor -o [output directory]/ -x [input.xml]
- use mkisofs to create the iso image: mkisofs -dvd-video -o [output.iso] [input directory/]
- use dvd+rw-booktype to mark the DVD disc as a DVD-Rom.
- use growisofs or cdrecord-prodvd to burn the ISO.
I was worried that the conversion to AVI would kill the video quality since a 15GB quicktime file is being squeezed down into a 498MB AVI, the expanded back into a 1.2GB VOB. The picture and audio quality both seem excellent though, and the sync between the two is dead on.
[]
permanent link
More video editing fun
I've been trying to catch up with all of the old HLUG video footage that I've been
sitting on. I believe I have a good system worked out now.
1. Import the video using Kino
2. Export the video to a single quicktime file
3. Use the viewer window of Cinelerra to split the video into clips that start at the same point in time
4. Use keyframes to fade between the two video inputs
5. Render back to a quicktime video file
6. Import the quicktime video back into Kino
7. Create a DVD or VCD.
It's a shame that Cinelerra doesn't have better import/export functionality. Kino is terrific in this regard,
but it's not very powerful when it comes to editing the video.
[]
permanent link
Installing Debian Sarge on a Mac Mini
1. reboot into OSX install CD
2. Hit Installer->Open Disk Utility on menu bar
3. Split the space on the hard drive, lock the second partition, set the type on the first partition to "free space", and Rename the second partition "Macintosh HD". When you're happy, hit the "Partition" button.
4. Go back to the installer and continue installing OSX. Might as well hit "customize" on the installation type screen and add X11.
5. Wait.....forever....and.....ever....
6. Reboot, reregister.
7. Eject the OSX install CD and insert the Debian install CD.
8. Shut down the computer then power it back on while holding down the 'c' key. If your keyboard is connected via a KVM, you'll probably need to unplug it and connect directly.
9. Hit return at the boot prompt to start the installation.
10. The Debian installer is straightforward. When you get to the partitioning stage choose the option for "guided partitioning" using the largest block of free space. You need a properly constructed boot partition for yaboot to work correctly and Debian knows how to set it up for you.
11. All done, another reboot. Load up OSX once to make certain nothing went wrong there then reboot back into Linux and finish the install.
12. Timezone, root password, user password, manual package selection, "q" once aptitude starts, done.
13. Log in as root. If you want to run Unstable rather than Testing, change /etc/apt/sources.list now. "apt-get update ; apt-get -u dist-upgrade" to pull down the latest updates.
14. "apt-get -u install kernel-image-2.6-powerpc kernel-headers-2.6-powerpc" to install the latest kernel.
15. Reboot into Linux again.
16. "apt-get -u install gnome-core x-window-system synaptic gdm".. Let's get a clean deaktop up.
17. Select "ati" as the video driver and leave the bus identifier blank. Say no to the kernel framebuffer.
18. Reboot or "/etc/init.d/gdm start" to bring up X.
Everything but sound works at this point. The snd-powermac module gets loaded without any problems, but /dev/mixer and /dev/dsp are not working. Glxinfo shows that direct rendering in enabled. Glxgears shows around 632 fps. OpenGL performance looks reasonable.
The lack of sound sucks, but others have reported the same problem. Hopefully this will be fixed soon.
[]
permanent link
DRI and more
Every now and then I'll google around for references to the dri-trunk-sid
packages I keep building. I'm mostly interested in seeing if the folks who
mention them indicate that they're building the DRM kernel modules as they
should or if anyone has decided to fork them.
Occasionally, I'm surprised to see something like
this where someone
notices a problem with my web server. In this case the message seems to be
noting that the index at www.wazzim.com mentions that my server was stolen.
This is a note I put up on that domain many moons back. The server was
stolen by the colo I was running it through and all of the content on
that domain was lost in the process. I probably should have put a less
menacing explanation on that domain. The main point was to explain why
the content that was previously there was no longer available.
Lots of links seem to be pointing at the new Xorg DRI packages in
http://www.nixnuts.net/files/experimental/ I really didn't want these to
spread too far until I get all of the kinks worked out of the packaging.
There's no worse feeling than finding out that a package you've created has
caused someone hideous problems. Even when the problem
someone has is completely unrelated to the work you have done.
So....what am I saying here? In a nutshell, please write to me about
problems with the software or websites here before you spread the news
to the masses. The ISO mentioned in that other thread was downloaded over
400 times and only one person had a problem with the download.
As far as dri-trunk-sid goes.... I'd still love to see someone else taking
over these packages. I'm definitely not the best person to be maintaining
them. Until that happens though, I'll keep working on them and trying to
improve them. Over the past few days I've been working on an automated
DRM module build system that will eliminate the need for end users to
know about the DRM build process. I'll just hook it into the postinst script
and the DRM will be built automatically.
[]
permanent link
Screw the RocketRaid
I finally gave up on the RocketRaid 1640 after I updated my kernel and ended up
with an unbootable system. Using a proprietary driver, compiled outside the
kernel tree, to mount the root filesystem is needlessly complex. Since I still
want a RAID 1 array and I don't want to waste CPU cycles on it, I decided to
bite the bullet and purchase a 3ware 8506-4lp card. This device fits the bill
nicely since (1) the driver is GPL and included in the 2.6 kernels, (2) Debian
detects the card and creates the proper initrd on its own, (3) it's a true
hardware RAID array. The only downside of the 8506 is that the card doesn't
let your create a RAID 1 array by duplicating the data from one drive to the
other...when you create the array you kill the contents of both drives.
Anyone in the market for a slightly used RocketRaid 1640?
[]
permanent link
Bar Business
Last week I spent a ton of time up at
The Speakeasy Bar trying to
put the place into good working order. This week I'll try to get a mailing
list up for the bar and see what I can do about adding some signs to
promote the bands that will be playing there. It's a great place, but we need
to attract more customers.
On a different subject, here's a picture of my wife Shazia and I from a few
months back.
I'm also working on two more HLUG videos,
new DRI packages, and a new XMMS Infinity package.
[]
permanent link
Damn You Maxtor!
My 240GB Maxtor drive on my primary workstation bit the big one last
thursday. It seems like the second half of the drive is gone and the
first half is unaffected. Part of /home is still useable, but all of
/usr/local is gone. Since this is the 3rd time I've lost vast amounts
of data due to a disk failure in as many years, I decided to set up a
RAID 1 array with 2 massive drives. Oddly enough, my wife felt that $500
for a crash-proof RAID array was reasonable (I may have lost a few hundred
pictures from our last few vacations) so I picked up a RocketRaid 1640 card
and two 300GB SATA drives.
The RocketRaid isn't a perfect solution, but it seems to be working reasonably
well. My biggest complaint is that the drivers and RAID management software
are proprietary. Aside from that, I'll have to make a new initrd with the
hpt374 module every time the kernel gets updated. Installation was reasonably
straight forward though. Hook up both drives, install Debian onto the first
drive, create the new initrd with the hpt374 module, sd_mod, and scsi_mod,
edit /etc/fstab to look for the partitions on /dev/sda instead of /dev/hde,
reboot and have the card mirror the first drive over the second drive,
install the management software, done.
Without a doubt there are better choices for a SATA RAID card. I didn't want
to go too crazy on price though.
[]
permanent link
DVD Notes
Creating mpeg2 video and audio suitable for a vob file:
transcode -V --export_fps 0,4 --export_asr 2 -x mplayer -y mpeg2enc,mp2enc -F "8,-q 4" -w 9000 --encode_fields b -Z 720x480 -E 48000 -i input_file -o output_file
Joining mpeg2 video and audio:
mplex -f 8 -o output.mpg input.m2v input.mpa
Split the file into block with vobview using "i". (creates /tmp/image.iso)
Join it back together and make menus with qdvdauthor
In unrelated news, it's election day, the DRI CVS checkout I was trying to build didn't work, and Duilio Protti the upstream author of xmms-infinity agrees that getting rid of the infinite_states binary data file is a good idea altogether.
[]
permanent link
Infinity Done and More
Reading binary data from infinite_states was the main problem with xmms-infinity. I went ahead and swapped all of the int, short, etc with gint32, gint16, etc just to cover all of the possible problems with different architectures. After submitting the package it looks like xmms-infinity is working across all Debian supported architectures now. The upstream author liked the patch and says he'll integrate it into the next release.
Right now I'm compiling a new version of the Debian packages of DRI CVS. I tried building packages of this code several times over the last week, but DRI has refused to compile for quite a while now. From discussion on DRI-Devel it looks like this version might have problems with R100 video cards. Hopefully I'll get some free time to test my Radeon 7000 and Radeon 7500 with the new version.
[]
permanent link
XMMS-Infinity
I think I've tracked down the bug in XMMS-Infinity on big endian machines. It
looks like effects_load_effects is swapping the byte ordering of the data
read from infinite_states. I'll play around with this some more after getting some sleep.
[]
permanent link
DRI Benchmarks
It's been several months since I did the initial run of benchmarks on DRI compatible
video cards. I started another run of benchmarks but stopped after a only a few
cards. Too little had changed at that point to make the effort worthwhile.
Since then there has been a good deal of improvement on the Savage and R200 drivers.
I'm not certain what improvements have been made on other chipsets, but I assume it's
not much. For the next iteration of benchmarks though I'm going to make some substantial
changes.
1. Adding additional video cards:
- Radeon 7500 64MB
- Radeon 8500LE 64MB
- Radeon 8800 FireGL 128MB
- Radeon 8500 128MB (Standard, not AIW)
- Nvidia Geforce 256 128MB
- Radeon 9200 256MB
- Voodoo 3
- Intel 740? Does it work?
2. Changing the demos used:
- massive1 Quake2 demo instead of demo1
- UCGuideDemo1 and UCGuideDemo2 Quake3 demos instead of demo four
- adding RevShogo demo
- maybee adding UT2K4 and Doom3 for the Radeons?
3. Adding the Summit drivers to the comparison?
4. Fixing some problems I caused:
- The Mach64 driver does work when set up properly
- The sis and r128 drivers might be more reliable with a bit of tweaking
It will be quite a bit of work, but the feedback on the original benchmarks was very
encouraging. I'm not the only geek wondering whether a 256MB Radeon 9200 is better or
worse than a 64MB Radeon 8500LE.
[]
permanent link
DRI Developments
A newer version of my DRI packages for Debian
is in the works. A great deal of progress has been made with the r200 (Radeon 8500 and
better) branch of DRI. For months Unreal Tournament 2003 and 2004 have been unplayable
due to frequent lockups. For months Descent 3, HeavyGear 2, and Shogo have been unplayable
due to problems with projected textures. Though I haven't tested recent versions of the
DRI CVS yet, talk on DRI-Devel indicates that these problems should be mostly, if not
completely, fixed.
I've also filed an ITA (Intent To Adopt) on the xmms-infinity package in Debian. I'm
already maintaining four other XMMS visualization plugin packages. Infinity is another
good visualization that needs a bit of work. The current version is completely broken.
Hopefully I can get a new one in before Debian Sarge is released.
[]
permanent link
First Post!
Every blog has to start somewhere. This is the first post of mine.
I intend to use this area to voice random thoughts, news about the different
projects I'm working on, and political nonsense that my wife doesn't want
to hear about.
If you have comments about anything written here, send them to john _@_ this domain.
Have fun everyone.
[]
permanent link
|
|