?

Log in

No account? Create an account

Previous Entry | Next Entry

The Ubuntu Experience: an interim update

Since I got all out of kilter going 5 weeks without a post, and since taiji_jian was nice enough to help me all day today with Ubuntu, I thought I would post another interim update.


Thanks to all who have helped with previous ones, which I posted on:

Especial thanks to taiji_jian, the great Paganini of TEUNC, for demonstrating the concept of ubuntu amply yesterday and today.

Here's my fortnightly update:

Another update about Ubuntu is in order. After eight months of being left fallow, Telperion, my 14-month old COMPAQ Presario SR1010NX desktop system (Celeron 2.8GHz, 256Mb RAM, 40Gb HD) is now running a 5.04 Hoary Hedgehog installation. All hail taiji_jian!

As I mentioned I was finally able to burn a bootable Ubuntu CD. However, the kernel kept quitting part way into the boot, until taiji_jian found out from another Linux-savvy friend that I needed to put linux acpi=off at the boot prompt. At this point, I had already spent two and a half hours installing WinXP just to do an Ubuntu install from Windows! How's that for perverse?

After this, I found that my 4x CD-RW, which is 6 years old, was burned at too high a rate. I downloaded ISORecorder, and, after much wrangling as gondhir and taiji_jian offered both jeers and encouragement, I reburned the CD-RW and finished the installation. Now, just a few things remain...

And now, the new questions:


Get thee behind me, 640x480: video drivers under Linux

Auugh! 6x4 is teh suxxor, and no mistake. I've just spent a little over 3 hours trying to get the modelines that I need to put into /etc/X11/xorg.conf, to no avail. I can't increase the resolution at all.

Is there an Ubuntu/Debian equivalent to SuperProbe on Red Hat?


Ubuntu setup - initial questions

My reasons for running Ubuntu are:

  • To house my database server (MySQL, Quiddity, etc.) and databases for research: the FIBR project, my new projects on collaborative recommendation (CR), information extraction (IE), machine translation (MT), and others.

  • So that I can have a *nix-based FORTRAN compiler and executables for the ITR project.

  • As a home for my servers: print, mail, Wiki, and IRCd

  • For remote access to applications via X


Is anyone familiar with setup of any of the above under Ubuntu or Debian Linux?


Quick draw McBana: unsafe removal of USB devices

How come hot-unplugging a USB data key (e.g., my 1Gb SanDisk Cruzer) from my ThinkPad and COMPAQ Presario results in a complaint from Windows 2000 and WinXP Home that I forgot to shut it down, and sometimes results in loss of the most recently-written data, but my new Dell Inspiron 6000 seems to force-sync the device?

Is the Dell just not telling me about a problem that also occurs there, or is it really robust to hot-unplugging? If so, is there any way I can get the older systems to automatically force-sync as well?


Thanks very much, as always!

--
Banazir

Comments

( 9 comments — Leave a comment )
scottharmon
Aug. 21st, 2005 05:50 am (UTC)
;-)
"Ubuntu" is an ancient African word, meaning "I can't configure Slackware".
banazir
Aug. 21st, 2005 06:22 am (UTC)
Ancient African Words
Hey, I ran Slackware! I just found my InfoMagic 5-CD set, (c) 1995, with Red Hat and Slackware.
I think it was RH2; I know I've installed 3-6 myself and had students install 7 - Fedora Core.

Ubuntu is an ancient African word, meaning:

scottharmon: "I can't configure Slackware"
zengeneral: "Surrender to closed source, M$ will rule all"
banazir: "Every OS sucks"

--
Banazir
chaosinaskirt
Aug. 21st, 2005 07:37 am (UTC)
at least with my key with WinME and WinXPh, there's an icon in the taskbar that pops up (though it usually hides with the compressed XP tasks). you right click the icon, tell it you want to shut down the drive, and wait a sec for it to say that it's ok to remove and remove it.

i can make screenshots if it'd help.
banazir
Aug. 21st, 2005 07:13 pm (UTC)
Transparency: of the key, by the key, for the key
Transparency is the key!

Please see below. What I am trying to do is to be able to rest easy hot-unplugging the drive as soon as the activity light goes off. You may have seen that many keys light up their LEDs as long as they are plugged in.

That's fine, but when I need to go, I need to be able to grab the USB drive and just yank it out. I don't mind having it go through extra overhead - time-wise, power-wise, etc. When I have to go, I have to go. No time to sync or shut down. Let the drive sync automatically every minute; I don't care.

Anyhow, thanks for your reply.

--
Banazir
zengeneral
Aug. 21st, 2005 02:58 pm (UTC)
my new Dell Inspiron 6000 seems to force-sync the device?
Is it XP-Pro?

insert the usb drive

check, My Computer -> Properties -> Hardware -> Device Manager -> Disk Drives -> Right click the USB Drive -> Policies

If you don't see
"Optimize for Quick Removal"
or
"Optimize for Performance"

then you have a shitty USB drive...

doing Safely Remove Hardware is analogous to doing

File.Flush();
File.Close();

... which, you should always do...
banazir
Aug. 21st, 2005 06:59 pm (UTC)
Mental core dumps and exceptions: lazy users, transparency and low frustration thresholds
You know how you sometimes core dump?
Well, I raised and caught a few exceptions there.

I'm not sure how you get a "Properties" on "My Computer" or "Hardware" from there, but for me it is:

Start -> Control Panel -> System -> Hardware -> Device Manager -> Disk Drives -> SanDisk Cruzer Mini USB Device -> Properties -> Policies

And yes, there is an "Optimize for "Quick Removal" / "Optimize for Performance" radio button. On this machine, it seems to already be set.

doing Safely Remove Hardware is analogous to doing
File.Flush();
File.Close();

Exactly. So, my question is: does "Optimize for Quick Removal" automate that?

... which, you should always do...
Do you mean that the implementor of the driver for the device should, or the user should? I agree with the former and would disagree with the latter.

Facts:

1. I am very lazy, as most users are regardless of skill level.

2. Consequently, and as a side effect of my personality type (ENFP), transparency is very important to me. I can do real work on a computer (code, write, design, etc.), but give me a straight path. Elide details. "Throughput at any cost," as my friend and classmate Mark Orletsky at Johns Hopkins used to say.1

3. I have a low attention span and minimal patiences for docs, as taiji_jian and gondhir discovered. I read docs (or more often, skim them), but I'll flit back to trying things very quickly.

4. I have a relatively low frustration level, but I won't give up on anything entirely. I will table things "for now" and come back in 6 to 18 months. Sometimes I will try again for 2-4 weeks and go back to the wasteful or unreliable way I was doing it before, as the lesser of two headaches. Consequently, disks or volumes 2 through N of software sit on my shelf unopened for months or years. My biggest debacles have been (Debian, Gentoo, and Ubuntu) Linux, CodeWarrior, and MacOS 10.x.

5. Transparency is not that hard to implement in software or hardware, regardless of the impact on "user discipline".

I believe that most users with discipline would benefit from more transparency, and transparency leads to greater reliability in computer systems. I absolutely deplore, despise, and abhor "verbosity by default": every time I see those horrendous boot messages scrolling in Linux I think of that T-Mobile commercial where the Sprint or Verizon users are trying to get the guy to use their overhead-laden service. "Y'all are clowns," he says. I sincerely believe that verbosity for the sake of looking sophisticated or technically powerful is poserhood of the lowest order and one reason that Linux will lose or fail to gain any more market share if its users and developers do not wise up.

1 That's not to say I like every detail hidden. I like a balance between flexibility and transparency. I generally don't like interface changes that merge features.

Take the way Trillian 3.0 messed up Contact List management. That really threw me off. It took me a while to get on board with Trillian 3.1's way, which generalizes Metacontacts into just one more layer in a hierarchy of groups. That's good, but it takes getting used to.

I am ambivalent about defaults. Skins, I like. Pre-loaded sets of features such as the ones in Mozilla Firefox's Tabbrowser Extensions are good, but they should be documented or tweakable via a simple tabular interface (one screenful of feature checkboxes instead of many tabs). GUI design is governed by simple principles that have been elucidated by people from Tufte to Jobs (see Apple's Human Interface Guidelines).


--
Banazir
zengeneral
Aug. 21st, 2005 07:15 pm (UTC)
Re: Mental core dumps and exceptions: lazy users, transparency and low frustration thresholds
Exactly. So, my question is: does "Optimize for Quick Removal" automate that?
I believe so, I don't have problems with it, but I switched to performance mode so my encryptor works faster (my USB drive is encrypted, is yours?). So, I take the extra second to safely remove device

Do you mean that the implementor of the driver for the device should, or the user should? I agree with the former and would disagree with the latter.

The driver implementor should implement according to spec (if there is one); as it is a usb drive and users are careless, the developers should have placed special care on it. Users should be aware of how to treat their devices. Treat them badly, expect loss. Treat them well, expect happiness.

5. Transparency is not that hard to implement in software or hardware, regardless of the impact on "user discipline".
... no, transparency is hard. it is far easier to bring tons of decisions to the user rather than hide them. Automating decisions for transparency sake is what windows does all the time. Does *nix? no, *nix is anti-transparency.

My biggest debacles have been (Debian, Gentoo, and Ubuntu) Linux, CodeWarrior, and MacOS 10.x.
Why do you subject yourself to shitty software? or, better yet I imagine, shitty hardware?



In the case of USB Drives, ALWAYS-ALWAYS-ALWAYS (jumps on desk), Safely remove them... if you don't, then you will lose data. You might lose a little, you might lose a lot... just don't risk it.
banazir
Aug. 21st, 2005 07:52 pm (UTC)
The right to be "careless", part 1 of 2
So, I take the extra second to safely remove device.
If it was an extra second instead of 30, I always would.

Since you tell me that I can't really rely on the implementors of the SanDisk Cruzer's driver to have put in the requisite flush, I will be paranoid and do so anyway. But I consider that a failure of specification on the part of the designers of USB 2.0. (Not necessarily SanDisk's failure, as perhaps they did do it, and I just haven't RTFMed.)

The driver implementor should implement according to spec (if there is one); as it is a usb drive and users are careless, the developers should have placed special care on it. Users should be aware of how to treat their devices. Treat them badly, expect loss. Treat them well, expect happiness.
If you add furthermore that "the specifiers should also take special care to make things transparent for the lay (or careless) user", then I agree.

In an ideal world, users, developers and designers would all be attentive and disciplined, not careless. However, as the world is far from ideal in this particular case, I choose to look out for "my kind", which is first and foremost the user. My typical compatriot may be a more technically qualified user than average, but he or she is a user first and a developer second. (And that's who everyone has to look out for. After all, unless you're going to advocate coding for its own sake, for the sake of the craft, or for the sake of the artifact: the user is the developer's raison d'etre.)

(continued)
banazir
Aug. 21st, 2005 07:59 pm (UTC)
The right to be "careless", part 2 of 2
... no, transparency is hard. it is far easier to bring tons of decisions to the user rather than hide them.
I agree that it's hard; I'm saying that in this case, it wasn't that hard. It would have saved many calories of user effort for the designers to have built this feature in. The few milliwatts of power consumed by the computer are cheap and the few calories of energy consumed by the user are expensive.

Automating decisions for transparency sake is what windows does all the time.
True, but there is a reasonable stopping point. I think you and I both agree that Microsoft Student goes past that point.

Does *nix? no, *nix is anti-transparency.
Well, yes; see my rant above.
I don't think I need to further than "poserhood of the lowest order". OTOH, as I said, there is a tradeoff. Users should be able to sacrifice transparency voluntarily when they want customizability.

Why do you subject yourself to shitty software? or, better yet I imagine, shitty hardware?
Haven't I told you?

  • 1. Linux

    • a) I want my servers: mail (lists, IMAP, SMTP), pritn, Wiki, file (Samba sharing), web.

    • b) I want them all on one machine. Windows generally can't do this. In my experience, the overhead for each of the above services adds up to too high a cumulative load.

    • c) Pursuant to (b) and your question about cheap hardware: I am cheap. I give myself a $2.5K a year budget for hardware. That sounds reasonable, but I am fully committed (i.e., tapped out) for 2004-2005 due to notebooks. (Why notebooks? Convenience, mostly, but that's another story.) Since I wanted to put $6000 for 2006-2008 into one tremendously good desktop system, and since I needed to get scads of peripherals (color inkjet copier/printer, scanner, USB data drives, DVD burners), I had to settle for a cheap Celeron. Because of (b), I decided on two cheap Celerons, which meant that they had to be very cheap indeed. $1200 got me all of the hardware, including two $450 COMPAQ Presario SR1010NX systems. You get what you pay for.


  • 2. CodeWarrior: portability was important (and still is). At the time I wanted Windows, Linux, and MacOS-based binaries, so I needed a single compiler suite I could target at all of them. The fact that I was using Java and the move to Eclipse made CodeWarrior obsolete, but I still want a nice multi-platform installer. NSIS has gone further in this vein, but if I want a carbonized standalone BNJ app, I can still think of no way to get it other than to build using CodeWarrior. (BTW, I don't like Java developers' equation of .jar, or even .class, with .exe. That's irritating, too.)

  • 3. MacOS 10.x: One word - Photoshop machine. Well, Photoshop and Final Cut Pro. Not much more in the way of reasons, for me.


In the case of USB Drives, ALWAYS-ALWAYS-ALWAYS (jumps on desk), Safely remove them... if you don't, then you will lose data. You might lose a little, you might lose a lot... just don't risk it.
I agree that the 20s of utility loss (probability 1 or close to 1) versus the expected 2000s of utility loss to recover data (probability of, say, 0.1 or 0.2) makes the decision clear-cut. I've copied things and not had them before, for a utility loss of maybe 200s; I've deleted things entirely before and had to re-download them from Azureus or Kazaa over weeks. My highest losses have been on the order of 20000s for recovery time. I'm not so stupid as to move the only copy of something of mine to a USB drive with no backup, so I have never had to reconstruct anything, but some minor things have been irrecoverable.

Again, though: why set up such a choice for the user? Quality software not only works, you don't see its guts churning or the intermediate and waste products of its working, like a child opening his mouth while chewing food or showing you where he went potty.

--
Banazir
( 9 comments — Leave a comment )

Latest Month

December 2008
S M T W T F S
 123456
78910111213
14151617181920
21222324252627
28293031   

KSU Genetic and Evolutionary Computation (GEC) Lab

Teunciness

Breakfast

Science, Technology, Engineering, Math (STEM) Communities

Fresh Pages

Tags

Powered by LiveJournal.com
Designed by Naoto Kishi