Trapped at Berkeley

I just upgraded one of my home computers to Microsoft Windows Vista Business and discovered that my Canon PowerShot G2 camera was unsupported. I hadn’t bothered to check whether my camera was supported; it seems most digital camera drivers are just glorified USB Mass Storage wrappers and support for those is generally a given.

Unfortunately, Canon’s drivers are a bit more complex, and the lack of Windows Vista compatibility was confirmed on Canon’s technical support page. Their recommended solution was to use a Compact Flash reader to access the memory card directly.

Unmoved and lacking a CF reader, I tried changing various settings and messing with the latest version of Canon’s WIA driver, but I couldn’t get it to work. On a hunch, though, I downloaded Canon’s TWAIN driver (Canon provides the WIA driver for Windows 2000 and XP systems, and the TWAIN driver for older systems) and gave that a go. Lo and behold, I was able to access and download the photos on my camera, albeit through a outdated GUI interface.

So, what’s the lesson to be learned? Don’t trust hardware manufacturers when they make a blanket statement that hardware isn’t ‘compatible’. You’ll often find that there are work-arounds for getting your hardware to work. Of course, also don’t expect them to come to your rescue when something goes wrong.

Penguin Meets A Huge Leech, originally uploaded by zeroion.

Perhaps the penguin represents Linux and the leech represents Microsoft, who has been “leeching” ideas from Apple and Linux. Unfortunately I didn’t have any apples in my fridge…

An article posted to Linux Journal today reminded me of a recent incident where Linux saved the day. “How Linux Saved My Files and My Job” tells the story of a Windows user who used a Linux boot floppy to recover data from a damaged NTFS partition. My dad and I did the same thing a few months ago.

It was the end of my Thanksgiving break, and I was preparing to fly back to Berkeley when I received a call from my father. His laptop’s hard drive had failed, and he had no backups. His company’s data was on his hard drive, and being unable to recover it was not an option.

Before he called me, my dad contacted a few data recovery firms, and each one had provided quotes in the thousands of dollars. Because the cost was prohibitive, my dad decided to give me a call. I agreed to help, and he brought his laptop home immediately.

On the first boot, the laptop’s BIOS failed to recognize the drive. After a little bit of shaking (really, it works!) and a few more reboots, I was able to get the BIOS to recognize the hard drive, but Windows XP was unable to completely boot. I didn’t have a Windows XP CD, so I couldn’t boot into the Recovery Console - not that it would have helped.

I ran to my bags and started unpacking; I needed to get my CD wallet. As a hardcore Linux user, I always carry a copy of Knoppix with me. I quickly extracted my copy of Knoppix and placed it into my dad’s laptop. I plugged my dad’s laptop into the home network, crossed my fingers, and booted the computer. The BIOS didn’t recognize the drive. Drats!

I turned off the laptop and tried again. Another failure. After a few more reboots, I was finally able to get the BIOS to recognize the hard drive. I mounted the NTFS partition in Knoppix, connected to my local FTP server, and copied the entire hard drive to my home directory. As an additional backup, I un-mounted the partition and used dd to dump an image of the drive. I moved to another computer on my network and showed my dad how to access my home directory using FTP. He copied over the files he needed immediately and burned them to a few CD-Rs. Linux had saved the day (and my dad’s data)!

Because of his experiences that day, my dad learned a couple things. He now uses rsync to backup his data on a daily basis. He also witnessed the power of Linux and is thinking about moving the company’s file servers to Linux.

Anil Dash makes an interesting observation: many nerds, at some point in their lives, have made an Excel spreadsheet documenting some aspect of their lives. Dash concludes his post by asking readers to comment about the spreadsheets they’ve made. Truly entertaining stuff.

As for myself, I made an Excel spreadsheet of my daily weight a long time ago.

No, I was not fat or conscious of my weight. In fact, those that knew me at that time would have described me as a really skinny person. So why did I choose to document my weight? The answer is simple: boredom, curiosity, and the joy of statistics. I wanted to see if my weight dramatically changed on a day to day basis, depending on how much I ate. My results were inconclusive. Although I was disappointed, it had given me something to do…

On another note, the topic of spreadsheets and statistics reminds me of another thing: a weakness in open-source spreadsheet programs.

As an engineering student, I often have to perform statistical analysis on tons of raw data. In most cases, I get my raw data in CSV format. Using Microsoft Excel, I can import the CSV data, plot it, and perform a regression upon the data with a few clicks.

Unfortunately, it’s not as easy with open-source spreadsheet programs. For example, take OpenOffice’s spreadsheet program, Calc. With Calc, I can import the CSV data and plot it with a regression curve, but I can’t get the equation of the regression curve to show up on the graph. I’m also unable to modify the appearance of the regression line.

It’s even worse with Gnumeric. The last time I tried, Gnumeric was only able to import the CSV data and plot it, but it was unable to plot a regression curve. After much Google-ing, I realized that the only way to plot a regression curve was to use Gnumeric’s statistics tools to calculate the regression coefficients and then manually plot the curve. Very annoying (especially when you want to add more data points; everything has to be done all over).

As a result of these issues, I’ve decided to stick with Excel for my data analysis. And yes, I know gnuplot is powerful and that I could do statistical wonders with Octave, but everything is so much easier with Excel.

A few days ago, Mark Pilgrim blogged about the interoperability of OpenOffice and Microsoft Office. He applauded OpenOffice for being able to open a PowerPoint document that could not be opened in its native application.

While I’ve heard many such stories, I’ve had the unfortunate luck of always being stuck in the position where OpenOffice will NOT open a Microsoft Office document, with the exception of the simplest memos. Sometimes, when I’m lucky, the document will open, but it’ll be rendered illegibly.

It just might be me, or it just might be the people that insist upon sending me documents in proprietary, closed, formats, but, even then, I have two other complaints about OpenOffice: it’s bloated and organized in an awkward matter. I’m sure most users would agree with me on the first complaint. In comparison to other Linux programs (AbiWord, anyone?), OpenOffice takes an incredibly long time to load and, once loaded, it makes my system sluggish. As to the second complaint, it’s a matter of preference. You can call me evil, but I’m used to the Microsoft way (as are most computer users). It would be nice if the developers of OpenOffice took that into account when designing their UI.

On a more important note, though, I believe that documents meant for distribution should not be provided in application-specific formats. I’m tired of having to install numerous applications just so that I can open documents when they are sent to me; I have utter hatred for viewers. In my opinion, documents meant for distribution should be provided as a PDF or as a Postscript file. These two formats are open and highly extensible, and their appearance is usually independent of platform, thus establishing consistency.

Of course, if you want recipients to be able to edit the files, that’s an entirely different issue…

Wednesday…the day I normally get a 4 hour break between my two large blocks of class…

However, today I’ve had to spend my otherwise normally peaceful break doing chemistry work, reading physics, studying for math, trying to decode quantum theory, and searching for a ::real:: calculus book (linear algebra based). Ok, I’m a nerd, but I’ll be spending a lot more time in the Math/Stats library pretty soon.

—

The GTK+ runtime for Windows is pretty decent. I installed Gaim and X-Chat last night, and they definitely beat their closed-source rivals. If only more people would begin to use open-source (or at least open-standards) programs.

—

I crashed my workstation in my EECS lab today (ok, I was in there using Matlab during lecture). I think I need to find new ways to vectorize functions…

—

Fluxbox is going to be a staple on my desktop.

Wow! crucial.com just cut the price of 1GB of notebook DDR PC2100 to $295.99 from $330. I’m waiting for the price to reach $200 so I can load up my ThinkPad T40. I’m seriously in need of more memory!

—

Just under 12 hours till my EECS 20N midterm. I haven’t started studying yet (#)($#($ MATH HW), so I’m just about to hit the panic button. Let’s hope I make it.

—

I’ve gotten back two midterms, Chem 4B and ESPM 50AC. I did pretty well on both, so I’m hoping my luck will carry over to Physics 7A (ha!). Hope is good…

—

Why am I on Windows XP? Oh yeah, because OpenOffice Calc sucks and I don’t have enough RAM to reasonably use MS Excel in a VMware on Linux! If only those OpenOffice programmers could get their act together and add some more ::simple/practical:: features to OO Calc. BTW, Gnumeric needs work too. I’ll post up a rant/review of OO Calc soon (after these midterms are over).

—

dilbert is cool. Of course, I didn’t need to say that. You already knew it.

—

On another note, although I should be saving up cash to buy my 1GB stick of RAM, I would really like to see Mel Gibson’s The Passion of Christ. Being a student of the bible and a person who wants to learn Aramaic (I think it’s occasionally offered here at Berkeley), I’m practically compelled to watch it. Maybe I’ll find some time next week…

—

I haven’t been to San Francisco since the week before Winter Break. Hence, I haven’t eaten In-N-Out for three months. For a Southern Californian, that’s pain =P

If you think I mistyped the title of this post, go watch Hackers sometime soon.

I’ve survived an all-nighter + an 8-5 day. However, I have to do physics now. ::sigh:: 3 hours and counting…

—

Concentrated HCl burns. Immediately rinse in case of exposure. Cytosine + uracil can be a new way to get high.

—

Cigarette smoke contains concentrated amounts of carbon dioxide, carbon monoxide, water, arsenic, formaldehyde, phenol, ammonia, and benzene. It’s an awesome cocktail.

—

With the unexpected public release of Windows 2000 and Windows NT 4.0 source code, I feel relieved to be running Linux.

Yay! I’m back in Berkeley. However, I already miss home again. I guess I’ll have to wait another 2-3 weeks before I come back. I really can’t wait to get back too, since I’m going to Europe for winter break! But first, I must survive these last few weeks (and tests) of the Fall semester…

I almost missed my flight on Tuesday. My dad’s notebook had a critical problem: the hard drive failed, and my dad did not have a backup. My dad came home from work, and I played around with the hard drive to get it working again. In the end, I discovered that if I shoke the hard drive a little bit and held the notebook sideways, the hard drive would work. Since I didn’t want to push my luck, I hurriedly plugged the notebook into my home network and mirrored most of my dad’s files to rand (my Debian GNU/Linux server). Although transferring files over the network was relatively fast (100 mbps and proftpd), it took a long time to get the whole process setup, and before I knew it, I was running out of time to get to Ontario Airport. Lucky though, I got to the airport with a some minutes to spare =)

Oh yeah, on afterthought, my method of mirroring files from my dad’s notebook was rather crude. I used Windows XP’s web folders to open an ftp session on rand and just copy-pasted the Program Files and Document and Settings folders. Had I found my Knoppix CD, I would have used dd to mirror the hard drive byte by byte (dd if=/dev/hda bs=10k | ssh $TARGET_IPADDRESS dd of=/dev/hda bs=10k). Of course, dd would have only worked if the disk geometries were identical (rarely the case). But, in that case, I could have tried Partimage (boo for experimental NTFS support)…

And now for some fun. With the topic of backup in the air, my dad is currently investigating backup methods for his company’s Windows 2000 storage server, since the backup utility included with Windows 2000 Server is hardly dependable. Since I’m an open source freak, I suggested swapping the entire backend to Linux, since the company doesn’t even use any Windows services (ie. the file server could be replaced transparently), but my dad’s boss is hesistant to try new things. Consequently, I have to prove the merits of Linux AND design a solution that works without disrupting their current setup. With these requirements in mind, I decided to setup a Debian test-server on their network and slowly migrate everything to it.

The more I thought about it, though, the more I realized how the server needs to act as a network appliance. Since nobody at the company knows how to maintain a Linux server, much less use Linux, I needed to design a simple management interface. As these and other necessities were realized, I abandoned the idea of building a custom server. Instead, I have come to the conclusion that either a new special-purpose distribution of Linux or package suite needs to be created. And here, I announce my commencement of the (yet to be named) “Drop-In Windows File Server” Linux project.

Mainly, I am looking to design a wrapper for Samba that will be easy enough for non-techies to install and setup. The file server will be able to exist in (most) Windows networks and will be accessible by (most) Windows clients. I will not be able to guarantee support for Active Directory and Windows 2003 networks, for reasons I will later post. My target audience is small-business networks that work in a primarily workgroup environment and that depend upon a single central file server.

I plan on writing some Webmin modules to facilitate management by those unfamiliar with Linux. I will also include ftp and NFS support for mixed (ie. networks with some computers not running Windows) environments.

Since remote access is a much desired feature (at least by my dad’s company), I’m investigating methods by which to provide file server access to those not directly connected to the LAN. However, security and usability are two of my greatest (and conflicting) concerns, so until I figure everything out, remote access will be vaporware.

I haven’t decided how authentification will be managed. Ideally, I would like to work with LDAP, but since my dad’s company works with a really simple network (peer-to-peer), I might make the Samba server simple and just give everyone equal access.

More details will be forthcoming. Development will begin when I have most of the details finalized. The project will be released under the GPL license (possibly LGPL). Although my desire is to have the project be a bunch of platform independent scripts, I just might end up making a mini-distro based upon Debian. Time will tell…