Wednesday, April 3, 2013
So, what's holding 64-bit arch from replacing everything 32-bit? There are many reasons; the two most important ones are - 1) there are many 32-bit hardware still scattered around, and 2) very crucial of all, people have developed an adamant inertia for 32-bit. They'll keep sticking to it. That's why some popular software come in 32-bit only, such as Skype, Teamviewer (I know there's a 64-bit version, but inside it's still 32-bit, 64-bit is only the wrapper) and AnyConnect. Good news is - on Windows 64-bit one can easily enjoy the mix of 64-bit and 32-bit applications.
#1 Windows is targeted squarely by all the ISVs at its vast userbase, and
#2 Microsoft through its skilled developer base is able to feed and keep alive an old dinosaur of 32-bit libraries and compatibility layers inside 64-bit OS versions to support both the arches ensuring that point-n-click usability.
On Linux desktop, 64-bit architecture has it's own set of irritants. And when you try to run a mix of 64- and 32-bit apps it misbehaves. There are some endeavors like multi-arch in Debian, but it's still not there.
On my CentOS-Windows (both 64-bit) dual-boot home pc, it was an uphill task to install some 32-bit-only applications on CentOS, whereas on Windows it was like a breeze.
32-Bit (Un)Success story and some How-To's
Showstoppers: Cisco AnyConnect, Teamviewer, Skype
Unresolved: Cisco AnyConnect. It installs quite well, but doesn't work at all. Looking for a workaround using OpenConnect. You may suggest your fixes, if any.
Installed/Fully resolved: Teamviewer (yum localinstall Teamviewer.rpm or whatever the name of that 32-bit rpm, it will pull in a few 32-bit dependencies)
Partially resolved: An old version of Skype (thanks CentOS Wiki, resolved partially, because you never know when the next update will break it all.
Skype installation procedure:
The latest version doesn't work. Please download a bit older static build (mentioned below) that's still available.
yum install glibc.i686 nss-softokn-freebl.i686 alsa-lib.i686 libXv.i686 libXScrnSaver.i686 libtiff.i686 glib2.i686 libSM.i686 libXrender.i686 fontconfig.i686 pulseaudio-libs.i686 alsa-plugins-pulseaudio.i686 libv4l.i686
tar xjvf /tmp/skype_static-184.108.40.206.tar.bz2
ln -s skype_staticQT-220.127.116.11 skype
ln -s /opt/skype /usr/share/skype
ln -s /usr/lib/libtiff.so.3 /usr/lib/libtiff.so.4
ln -s /opt/skype/skype /usr/bin/skype
Wednesday, February 27, 2013
Question: Why do I like Debian?
Short Answer: It gives me a "Debbie Does Dallas" experience. Addictive like pristine porn - no nonsense and predictive. The rest gazillion other lady-like distributions invigorate you with too much Viagra but don't show up at THAT moment.
Long Answer: Read this review and/or comb the web for so many other reviews.
From early days, sitting on version 4.0, that's right Etch, it flirted me. It was neither too easy, nor too beautiful, but always rock solid. Then followed Lenny and Squeeze only to reaffirm and sometimes spice up that experience.
With that prelude follow the run.
Wheezy from it's very alpha and beta phases put so many Debian lovers into suspicion. Whether it'll default to XFCE or Gnome; why it jumped into that gnome 3 mess; and if it'll stick to the proclaimed "2-years - one release" schedule. With the release of RC1 it certainly proved that it sticks to the schedule. Also confirmed that gnome3 is not that much of a holly mess the jerks make it out to be. Little obsolete as per Fedora and Ubuntu standards, but the latest Debian is very modern and polished.
After cd/dvd drive started behaving mad while still under warranty I didn't go for a replacement. Not wise to add to plastic waste. Better use USB drives - for sharing data and creating bootable media. Debian is infamous for creating USB bootable media. Unetbootin didn't work as expected. But now you don't need it anymore. Bytecopy procedure works like a charm with netinstall images. I picked up a 64bit all-nonfree-firmware loaded netinstall image and created a usb install media with dd. In my case, it was:
dd if=firmware-wheezy-DI-rc1-amd64-netinst.iso of=/dev/sdc bs=4M; sync
Change the device options as per your setup.
Butter smooth install routine, as expected from a Debian installer! I chose my first SSD, an OCZ Octane SIII as / partition and dedicated a donkey 2TB WD HDD for /home. No swap (who needs swap when you've 16GB of RAM). Went for some SSD options - noatime and discard. Then followed the sane default package selection and user account creation, MBR selection, et al. Boring, it may sound, if you're accustomed to ugly surprises during installation.
Install routine completed. Poweroff. Night set in. I went to bed.
Next morning I woke up to a nice Debian Wheezy on gnome-shell. Seductive as they say, but not slutty. Never experienced any crash, gnome-shell nuisance or kernel-panic stuff. There were some minor annoyances which I hope will be fixed with the Final Release. Mind it, it's still an RC1!
Everything worked. Network sharing with Windows 7 on my Asus EEE 1215b, data sharing with my ntfs partition on Windows 7, mac filesystems on a hackintos (never tell mac people), on that same machine. Then some more, bluetooth, WiFi and all.
For the first time, multimedia experience was very smooth. Just install libavcodecs-extra package and you're there. Throw anything at it, it plays. And for those abominable porno stuff with some esoteric format there's always the venerable VLC. In short, the days of bad multimedia experience is over. Basement dwellers and FOSS purists may complain though, for this non-free move.
Very frugal. No sudden clock ups and downs. CPU stays at ~1-5% throughout till you don't burn it with intensive gaming, program compiling and media encoding.
Very low footprint of memory. Freshly booted system uses ~320MB, with all effects - bells and whistles on and mysql, apache services (and what not) still enabled (for my wife) on runtime.
Do that foreplay. Fix bugs and bad things!
Wheezy at RC1 has its share of bugs and problems. However, none is a showstopper. You can live with them without any fix, cos most of them will be fixed automatically in the Final Release. If you like you can fix them very soon with a few minutes of dirtying your hands in the command line. First fire up terminal and be a root user - use su or sudo, whatever you're used to.
#1 - "cannot set freq 16000 to ep 0x86"
During booting the screen fills with "cannot set freq 16000 to ep 0x86" error messages and stops for around 4 secs (it might vary in your case) and then the login windows appears. This bug is related to Logitech HD quickcam. The easiest fix is to update kernel from Unstable Repository. Enable unstable repo in /etc/apt/source.list. Upgrade kernel and disable, or better delete that entry from the sources.list. Never fiddle with testing and unstable.
#2 - no sound
While Intel Azalia HD sound card worked nice on first boot, consequent boots were without any sound while the audio mixer button was still showing active.
Permanent fix for this annoyance is to put:
options snd-hda-intel model=generic
at the end of /etc/modprobe.d/alsa-base.conf
#3 - enable poweroff option:
Default gnome-shell 3.4 doesn't show poweroff option if you don't press Alt key with still point the mouse to the suspend option.
The easy fix is to edit /usr/share/gnome-shell/js/ui/userMenu.js
Look for "this._haveSuspend = this._upClient.get_can_suspend();" and replace the same with "this._haveSuspend = false;" (don't use those quotes, they're here to distinquish commands/config text from human English)
#4 - [warn] PulseAudio configured for per-user sessions ... (warning)
This pulse audio message is not fatal. But you can fix it too:
and change it to
That pretty much ends this review.
Tuesday, January 29, 2013
KDE 4.9 (and above) workspace comes as a great relief when the major distributions resort to some fu**ing desktop environments (such as gnome and unity) and/or spin some lame reiterations (such as cinnamon, mate and a ton others). The latest KDE workspace is way too much polished than unity and gnome3 - highly usable, customizable and not so resource-hungry. But you can still make it lighter and faster if you don't rely much on desktop search function and KDEPIM stuff like kaddressbook, kmail and the rest. Amaik, very few people use them anymore, and can be done without them. Just a working set of thunderbird, pidgin and firefox, and you can very well dump kdepim + akonadi + nepumok + strigi.
By "Dump" I mean "Disable", cos you can't remove those packages as they are deeply integrated into the workspace. Here's how you can disable:
#1 Add these lines (if they don't exist, else make necessary changes in them) to ~/.kde/share/config/nepomukserverrc
#2 Then add these lines (if they don't exist, else make necessary changes in them) to ~/.config/akonadi/akonadiserverrc
#3 Open ~/.kde/share/config/kdedrc and set "autoload" to false for nepomuksearchmodule
#4 Finally, remove the file nepomukcontroller.desktop and akonaditray.desktop from /usr/share/autostart/, if they exist.
Log out or restart your system to experience the lightness and speed of your sexy KDE workspace.
Saturday, December 8, 2012
"Year of Linux Desktop" has always been a delusion of grandeur, or the so called pipe dream at best. It'll remain so for the future n-number of years, cos Linux is free and it shows, feels and works like everything free, with a connotation of "Cheap".
There's no problem in its being free and open source as long as it works for the majority of userbase, which it certainly doesn't. Why? Because it lacks the commercial gloss people are used to. Against popular belief, you won't have to practice closed/proprietary means to achieve that polish and gloss. All that's important is certain orderly and cohesive approach to how software work, in both users' and developers' point of view, and a clear direction. For example, Android is both opensource and free, yet it brims that much desired commercial gloss, courtesy the big G's focus on user experience. The Big G knows what to flaunt, what to hide, and so on.. and on..
On Android you don't bother which bootloader to load - grub or lilo, which DE to choose from - KDE, Gnome, LXDE, Blackbox (there're a dozen others), how to set system initiation - systemd, sysvinit, innserv... how the sound and audio subsystems talk to the rest of the system, bla..bla..bla... Here these ugly system software work under the hood, users are unaware of it for a lot of good reasons. This is how the big G establishes order in an otherwise chaotic open source model of software development.
The prevailing model of linux and the software built around it suffer from three perennial diseases due to its overtly free nature of "Fork everything and Break everything" development:
#1 Shoddy quality of software: Forking model of development that leads to bad testing, rapid release cycle and unnecessary overdo. Users are often forced to fix certain problems where they should not happen in the first place. For example, a certain sound card works out of the box on a certain distribution, needs to do some manual editing of certain config file on some other in order the sound to work, wherein still others the users are forced to remove a certain component and install/build something on their own. Whereas the kernel versions, alsa/pulse components, DEs and the usual software stack are the same across all these examples. Reason? Bad integration! Morever, these type of user annoyances are not restricted to sound only, they are there in network, graphics, power usage, and then some more, in other application software too. The situations/annoyances aggravate further when every distribution puts together certain types and versions of the system and application software different from every other and in a different way, where most of them seem similar on the surface.
#2 Overkill: Forking and freedom to do everything leads to unnecessary overkill of time and efforts and further contributes to shoddiness in software quality. For example, what KDE or Gnome should do? Develop DEs and a few DE-specific tools/appls, right? But that's not the case, they keep constant focus (and ignore the core activities) on things they should not do in the first place. First of all, application software should not be DE-specific, yet every DE, be it KDE or Gnome, has its own set of office suits, browsers, messengers, media players/managers and system utilities and tools. This, more often than not, results in poor duplicates of certain industry standard software. Intrigued? Just compare KOffice with LibreOffice, K3B with Brasero or XCDroast...? Everybody loves to skin a frontend, cos its' tough and boring to squash the bugs out of standard libraries such as gtk and qt, and backends such as cdrdao or growisofs.
#3 Always in transition: Lastly, this forking and chaotic nature of development creates an environment where the platform is always in a state of transition. Consider the period of transition such as hal >> udev, alsa >> pulse, xserver >> wayland, module-init-tool >> kmod. Sadly, it has been so from the very beginning, and it will remain so till eternity. It is a major contributor to user annoyance. How about consolidating various forks, putting enough manhours and creating something that will just work, in stead of forking every which way possible and breaking things on the way.
I'm sure only very few will agree with me. Cos most of the so called open source community members consider free software as a cult than a tool for general well being. For them choice (read it Chaos) is everything. And yet they expect a placebo effect to these perennial diseases of bad quality and performance. They'll poop, fork here and there, make badly integrated incohesive software tank to drive to the nearest supermall, whereas a lean street bike would have been a better choice.
Monday, August 27, 2012
But I feel, sooner or later, Linux will start dominating the consumer desktops with google power, in a form similar to Android. For quite long I've been following android-x86.org, a community effort to port Android to x86 platform. So far it's been non-yielding. None of the release is feature-complete. However, the good thing is - this project is less fanatic about open-source. I'm sure google will bet on it and make it big by building a consistent development environment wherein the new desktop will consist of highly-integrated subsystems backed by a simple application distribution system. Google will strip as much ugly stuff as possible off the OS, hide the rest from the users, and put a bright layer on top of the customized linux kernel. Of course, the hard, the hobbyist, and the raw fragmented-linux desktop with full of choices will be there for the basement dwellers.
Friday, July 13, 2012
What's value of linux being modular if it needs so many duplicate and different versions of the same libraries.
I am running debian squeeze which has almost 50% packages from wheezy such as xorg stuff, kernel, drm, mesa and some application software. However, accidentally I found that I've bewildering number of duplicate libraries. It's insane, can't even remove those duplicates as they are dependencies for a variety of packages. In the normal world a later package should override an older version, but it's different in linux, sadly. Even if I've deborphan and debfoster I see tons of different versions of libraries such as: libavcodec, libavdevice, libavfilter, libavformat, libavutils, libdb, libjpeg, liblzma, libmatroska, libnspr, libntfs, libssl, libswscale, libusb and libx264. Please have a look at the youtube video to know what I mean.
Wednesday, June 20, 2012
I was very much at home with my mighty Debian untill I was lured by GIMP 2.8.
I knew Debian Squeeze repo will never have it. So, looked for backports. Nothing there too. Then tried backporting it myself. It is not as easy as:
mv gimp-2.8.0.tar.bz2 /tmp/
tar xf gimp-2.8.0.tar.bz2
make all && make install
there's much more. Compiling it is really tough, you'll be flooded with errors related to missing packages, flags and lots more. Then I took the easy path of pulling it down from testing repository by adding:
deb http://ftp.de.debian.org/debian/ testing main
then configuring apt pin priorities to make testing version as additional to prevent distribution wide updrade by putting:
Pin: release a=stable
Pin: release a=testing
apt-get -t testing install gimp
Holy shit! It almost installed truck loads of packages from testing. Finally I fucked my Debian Stable. It now feels slow, sound stopped working, it shows tons of boot error messages.
Dear linux distribution enthusiasts please bring out some api/abi that doesn't change every fortnight. 99% users love stability more than the shiny non-working stuff that doesn't bring any perceivable difference to their lives.
Tuesday, June 12, 2012
I'm also a speed freak. Here's my take for speed.
Hardware: Intel H61 board, Pentium G620 processor, 8GB Corsair DDR3 1333MHz RAM, Atheros LAN Card, Intel HD 2000 Graphics on the same die of the CPU, 32GB Kingston SATA II SSD as / partition + Western Digital Blue SATA II TB HDD as /home partition.
OS: Debian Squeeze 64Bit, backported 3.2 kernel, mesa, drm and xorg manually backported from Debian Wheezy repository.
Package Repos: Debian squeeze, backports, mozilla-debian, debian-multimedia, google and debian wheeze.
Desktop Environment: Gnome 2.30
Software: LibreOffice 3.5, Gimp 2.6, Mplayer and Totem with all codecs and browser plugins, Iceweasel (Firefox) 13, Icedove (Thunderbird) , Sun Java and Netbeans full suite, GCC + G++ along with Code:Blocks, Full Wine suite, and the fat repertoire of media players, audio/video editors, remote access/desktop sharing tools, mysql server, client and admin, a plethora of games and tons of system utilities and recovery tools.
Speed benchmarks: Booting 8 secs, libreoffice startup 1 sec, firefox 1 sec, gimp startup 2 secs, and surprisingly netbeans startup 3 secs. Initial memory footprint is just 130MB. You can't expect more.
Tweaks I did: Optimized SSD to use trim (noatime,discard option in the /etc/fstab of the ssd drive), disabled ipv6, changed /etc/hosts to look for localhost whenever possible instead of searching and reaching there, removed swap partition, removed initram disk, put all my driver modules directly into the kernel so as to avoid seeking initram. Finally cleaned all the unnecessary locale files, symlinks, orphan files and then a few tweaks on gconf-editor. Finally cleaned the gnome config residues (after installing all the software I needed) with gconf-cleaner.
Here's the testimony video, watch it in HD 720p format full screen:
Friday, May 11, 2012
Only this morning I saw a post by Caitlyn Martin on tuxmachines.org. About "Linux Hardware Support Myths and Legends"
Come on Caitlyn, admit, Linux is still not ready for consumers. It's never going to be considering the control, complexity and its nature of development. We all love linux. We can tinker our way to use it at our homes and offices. We'll keep on using linux, and we are not sad or sorry if linux doesn't make it to the mass adoption. But your adamant post is laughable at the best. Numbers tell, if there was a pinch of truth in your statement linux desktop would have enjoyed comparable marketshare like the one from Redmond. But 1-2%!
You wrote, "Linux is compatible with more hardware than any other OS bar none. That certainly includes Windows. Try installing Windows 7 on some random laptop from scratch and see how much is missing or unsupported without third party drivers. My experience doing Linux installs for my customers is that a lot of off the shelf hardware "just works" and the rest needs proprietary drivers downloaded to make it work, just like Windows. There is, indeed, some hardware that doesn't work with Linux and years ago that was a real issue. The fact is that more and more manufacturers are supporting Linux well and other drivers have been adequately reverse engineered."
Partly right, linux does support maximum hardware out of the box. But that other OS supports all of the hardware, though not all out of the box. Besides, the maximum number of hardware that linux supports is not as good or simple as it is in that other OS. Compare the installation procedure and performance of modern ATI and NVidia graphics cards on linux with that other OS. Linux fails from a long margin. Moreover, the support is both complicated and buggy most of the time, and not feature-complete.
Caitlyn, please try any of the recent Northern or Southern Island ATi Radeon graphics card on your favorite linux box. Do the same on that other OS. Compare the performance and installation procedure. On that other OS, all you've to do is to pull in the software (just one executable file) from AMD and install it. Your favorite distribution with modern kernel and gallium drivers will readily support the Northern Island ATi chip out of the box. But the performance will be shoddy at best. Then you will search your distros repo or browse the manufacturer's website for proprietary drivers. Download and install the packages. Graphics still won't work the way it does on that other OS. Then you will look for vaapi wrappers, configure the media player settings, bla..bla..bla.. In linux it's not just a driver package, you've to fight with xorg, kernel, drivers, wrappers and decoders. After this big fight your distribution may play your 1080p hd stuff. But again, not as good as on that other OS.
Audio driver scenario is no different. The point is, Sound and Graphics support in linux is very complicated and is always in a catch-22 stage. Admit it!
Here's a copy of the discussion thread on her post. Just imagine how mis-informed and adamant Caitlyn is in comparision to the mature, practical and realistic Jack.
By Jack on May 10, 2012 11:21 AM
Having used Linux since the end of the 90'ies i certainly agree that hardware drivers are not really a big problem. But for some, it really is.
Therefore one should never claim "pick this or that distro - it works out of the box" unless one in fact has experience with the exact combination.
If you do recommend where you should not, you are likely to be the creator of a user very hostile towards Linux, thus a true credibility killer.
Another wrong thing to do is recommending Linux with the ambition to make a Linux enthusiast out of the victim. Recommend (and if possible assist in installing) Linux on basis of the potential users needs and capabilities.
Forget your own preferences - it's all about the potential user, as long as you are sure that you are able to assist.
Now, I've got some shocking news!!
There are no perfect Linux desktop environment, and there are no perfect Linux distribution. There are probably 20-30 distros and 4-5-6 desktop environments that are technically suitable and has the potential, but none of them are really there.
We, as enthusiasts, are able to tweak a distro or a desktop environment into submission. The regular users who just want a neat piece of equipment to work well are not.
Where are the distros for consumers? They don't exist.
By Caitlyn Martin in reply to comment from Jack on May 10, 2012 12:58 PM
Jack: I have some shocking news for you. Any of the major Linux distributions are consumer-ready. That's been true for at least a decade. Any of the major Linux desktop environments are consumer ready. That's been true for at least a decade as well. The current generation of desktops on Windows and even on MacOS are based on ideas that first appeared in Linux. See my recent article on the subject. If those Linux distributions aren't "really there" then certainly Windows and MacOS are even less ready for consumers. You are certainly right that there is no perfect Linux distribution or desktop. There is no perfect OS, period. Linux is in no way more flawed or less ready for consumers than the other options out there.
I see the second part of your comment the same way I see the articles proclaiming that Linux hardware support is terrible: as spreading more fear, uncertainty and doubt for no good reason.
Seems, evangelists like Caitlyn are more of problem than solution to the badly integrated and ever-forking open source OS.
Tuesday, May 8, 2012
- "DONE" means that it is implemented and relatively bug-free.
- "MOSTLY" means that it is implemented but has some known bugs.
- "WIP" means that someone has started on the initial implementation.
- "N/A" means that the feature is not supported by the hardware.
- "N/N" means that the feature will not be implemented, because a better alternative is or will be available.
- "TODO" means that someone needs to write the code.
- "UNKNOWN" means that the current status of this item isn't known.
|Max Supported Display||2-6||4-6||6|
|Decode on 3D engine||WIP||WIP||WIP|
|Decode on UVD||TODO||TODO||TODO|
|Mesa 3D features||Evergreen||N.Islands||S.Islands|
|Anti-Aliasing||DONE||not tested||not tested|
Request to all those who love Linux and PCLinuxOS
Please share your experience and experiments with PCLinuxOS. Hurry up, mail me at firstname.lastname@example.org