HTC Ozone

My first "smart" phone was the HTC Ozone, a Windows 6.1 phone. It's build quality impressed me, so much so that I can't think of getting something other than an HTC phone now (current one is an HTC Droid Incredible 2). The nice thing about the Ozone is that it is a world phone; GSM and CDMA. I tried it in Australia with another sim card, but it didn't seem to work. I think there is some Verizon bullshit you have to do before it works in another country. But I did use the wifi in Europe when it was still my main phone. A good phone either way in places where you wouldn't mind losing it or having it stolen. As these old smart phones of ours become more common, i'm starting to think of what to do with them besides making phone calls. I'm always trying to add more cameras to our home surveillance network, and a smart phone as webcam/surveillance actually makes alot of sense. There is even some software for crusty Windows 6.x phones I might try.

Ode to the e815

I have been a lucky man to have my wife give up her old Motorola e815. This phone rocks for a number of reasons. While it's no smart phone, it's does enough for my purposes that because I didn't pay for it, it's smart to me. It's one of the few free-with-contract phones with EVDO I have ever seen, so using it as modem is extremely bearable. But even using the old-school Openwave browser on it with EV as compared to my old free samsung is noticeably snappier. It also has a microSD slot, and copying photos, movies, and music to it is relatively easy. Making ringtones for it is a snap too.

Browsing free models from Verizon these days with their goofy rooty-tooty-new-every-two promotions, I don't think I'd be as lucky to find a comparable phone, especially in terms of reliability. While the e815 is noticebly beaten, Eva's shiny black Samsung's camera phone broke within three months. Who cares, it's only VGA.

Even if I found something that matched on technology numbers and perhaps reliability, one thing scares me the most about a new phone; a shitty UI. Moving back to a Motorola phone, even though my last one was from character lcd, the UI is leaps and bounds better beyond many technologically superior phones i've had a chance to interact with. While you say you might want a smart phone, what people really want is a smart UI.

Google Custom Domain Headaches

Ack, it's been while since I've posted, I know, but what luck, i have something relevant to actually report.

After initially setting up my wife's blog, tucsonquerido.blogspot.com to use a custom GoDaddy domain back in June, I found what everyone had seemed to find out; it's not perfect. The details are at The Real Blogger Status - Beta: Google Custom Domain - Case Study # I tried Google's way initially, but that left tucsonquerido.com users out in the cold, and after all, www is deprecated. So after a bit of tweaking, I got just plain tucsonquerido.com to work with Blogger by using the IP address associated with ghs.google.com. But I was back at square one. Old school web users could not use www.tucsonquerido.com. Still, google usually gave them a search page instead. I left it like this for a long time and google indexed it accordingly.

I decided recently to see if Google had fixed this, and they did! Well, sort of. They added a checkbox that flip-flops between "forward www traffic to tucsonquerido.com" and vice versa if your publishing url already started with www.

"Great," I thought, "I'll just change everything back and flip this bit!" So I did.

48 hours later, it still didn't seem to work. I followed the directions above and tried deleting the A record, but that still didn't seem to work. I called GoDaddy and they didn't quite know what to do either, but they did suggest at least using their A record IP.

However, I seem to have stumbled upon a possible fix. Before this recent mess, I simply changed the existing A record to point to the ip address for ghs.google.com. Like I mentioned above, this worked for a long time, I just left out the www people. So in order to get my wife's search links back, I changed it back, but this time, added a CNAME record www that points to @, or the root, as well as leaving the the checkbox checked in the publishing settings.

So far, it seems to have worked. But give it 48 hours...

The Rube Goldberg-ian WMA to MP3 Converter

Since we've started using our new home NAS to consolidate and backup our media, I've also undertaken the task to convert my fiance's previous WMA media collection she had accumulated since we started dating. Some of our media devices already supported WMA, such as our Blaupunkt car stereo, and Panasonic DVD player, but the newest, an Ipod 5th gen, obviously wouldn't (well not out of the box at least). Even in the car stereo, however, WMA files never fully worked; tags were never read. So even though MP3s are as much proprietary as WMAs, and WMA has many advantages of its own, MP3's ubiquitous support has won my heart and hard disks. I'd like to believe that keeping OGG media around and supporting its file format by buying a player that supports it will change matters, but the winds of time are behind MP3. Soon enough, patents will expire on MP3.

I initially thought the task would be much easier; surely Google will save me by returning results for "WMA to MP3". Simple inexpensive commercial app links were returned, and they all seemed capable of doing the job, but I felt "icky" knowing that there had to be a GPL/GNU solution to easily do what I needed (or at least a "free as in beer" solution). I had faith I could find success using GPL tools such as mplayer, mencoder, and lame. And I was partly right; I found several examples of mplayer usage with lame in thier own Rube Goldberg-ian *nix bash scripts. So I had hope at least I'd be able to simplify or adapt these to a Windows environment.

I started using an all mplayer solution; pipe mplayer to mplayer or mencoder and be done with it! Eh, not exactly. I could get mplayer to output wav files though... or... to stdout... which lame can read! Well, that would be an awesome one line commando transcode, but I could never get it to work on Windows. I did find a patch by Slingbox that gave better support for stdin in mplayer or lame (I forget which now) which lead me down the path of trying to compile my own solution, but that got to be too much work. Plus, i realized that there doesn't seem to be much support for tags. And that's the rub; I'm an ID3 tag and filename freak. If I didn't appreciate metadata so much, I would probably wouldn't be writing this post, but there's just something I enjoy about a good ID3 tag.

So what did I end up with? Actually a couple of reasonable solutions. Now Winamp isn't free as in speech, but I am quite familiar with it and it is free as in beer-ish. The paid for pro version could probably do all of what I needed quite easily, and support my ipod, but here comes the icky feeling again. The free version does have a couple of great features though. It does have a built in transcode, and it does support a recent out_lame plugin. The transcode feature doesn't let you transcode to mp3 without paying for Winamp Pro, but there is a rather elegant hack of installing the lame ACM codec to add support to Winamp WAV ripping to convert to mp3. But after a couple of passes, it would crash, and I could only seem to get ABR support, not the full VBR I desired.

However, the out_lame plugin is quite nice. It will let you take whatever Winamp will play back and convert it to mp3, all according to what tags and filename features you setup in Winamp.
Now that alone is not very Rube Goldberg-ian, and I still had to at least drag and drop both the WMA files and output files, so I developed a couple of .bat scripts that when used with another Winamp command-line add-on called CLEveR, would allow me to batch convert every wma it found, using whatever lame VBR settings, tags and filenames of my choosing.

Would Rube Goldberg even know what I was talking about by now?

Openfiler at Home

A few months ago, as I was anticipating my next move for my home NAS, i came across a slashdot poster's link to Openfiler . While I was originally going to use FreeNAS, i'm glad I saved that bookmark. The FreeNAS installer would not boot on my machine, and if it wasn't for Openfiler, I really wouldn't have a free web-based alternative that I could try turn-key.

Openfiler seems to have come a long way from it's 1.x days, where, if i'm not mistaken, it was a package to be installed on a standard linux install. At version 2.0, it was based on Cent OS, and now at version 2.1+, based on rpath. However, the features it now offers are way beyond Freenas. To me, that means it's getting leaner and meaner, and because it's used in production environments, by paying customers, reassuring.

Although it will take more disk space for an Openfiler install, I really think that's moot, at least in my case. A 32Mb install is great, but really, the cost difference in between a 32Mb compact flash and a 2Gb compact flash is not that inhibitive, especially when you have the ability for local authentication, raid 6, and lots of supported hardware.

Although my installation of Openfiler did have some hiccups, the support forums and irc channel had plenty of people willing to help. The web gui became very easy to navigate after a short period.

As for performance, I have only done cursory tests, but I have been very surprised at the speed. Both my workstation and the nas use run-of-the-mill rosewill gigabit PCI cards, and I have been able to make dvd isos move back and forth very fast. One thing I've found, is that any bottleneck that may exists (besides no jumbo packet support on my D-Link 4300 router), is probably due to my slower pata disks in my workstation, which steadily light their equivelent leds, while the server only blinks occasionally.

I still have lots of things to tweak on the server, such as the hardware monitor, and all the leds, which all blink in unison for some reason, but i'm very happy to have gotten this far, at the performance I was shooting for.

Gamble gone wrong

After a long debate on buying the Thecus 5200, an impressively built NAS bassed around a Celeron M, I decided I would instead use it as inspiration to build my own. Although it's plug and play, has gigabit ethernet performance, supports 5 discs and Raid 6, features high on my list, it is also a tad more expensive than what I believed I could spec out on my own. Plus, I wanted the experience of building my own always-on server.

I started out with an Icy Doc 4 bay backplane and a black Mapower 4 x5.25 bay external raid enclosure. One of my goals was to approach teh same size as the 5200, as well as create a look of a "just a bunch of disks" front facia. My first motherboard, an ASUS P5RD2-VM, was just a tad too big to fit in the enclosure without some heavy modification to both it and the backplane. I then exchanged it for a different Asus board, the K8N-VM, which is the smallest, least expensive microatx motherboard I could find. It didn't have the same exact features performance wise, but it was still workable. Or so I thought.

I tried several combinations of orientation, but nothing layed out well still. The motherboard would fit without modification, but because the case was never meant to mount a motherboard, it would have not mounted well on the little bit of metal to hang it on. Plus, the end of the enclosure would end up looking very ragged. So, I upgraded to the smallest, most reasonable looking microatx case that could hold 3 5.25 bays. The Athenatech A3603BB.400 is a good looking case, that sets itself appart in that sense. It takes the Icy Dock 4 in 3 backplane well, but as reviewers have noted, you can't use just any microatx motherboard, as the lowest 5.25 bay will probably hit. No problem, i have the smallest microatx board that appears to exist. Or so I thought.

I wouldn't be writing this now if everything was up and running. Instead, i'd probably be ripping my entire cd and dvd collection to my new fileserver. I used to think the Asus brand name was revered and sought after (I exaggerate), but I have come to realize I can't trust what even their own manual says, or the standards they say they support. This board does not have enough sata2 ports for my needs, but that shouldn't have stopped me, as it does have two pci-express slots, a 16x and a 1x. Using two run of the mill Rosewill sata2 pci-e 2port cards, I thought surely I could overcome this, and have extra sata2 ports to boot, especially since I could just use the onboard video. Not so, says Asus tech support, whom I called after the machine would beep with a 1x board in the 16x slot, regardless of bios settings. "Only a video card is supported in the 16x slot."

This seems to go against the pci-express standard, but that doesn't seem to be a concern with Asus, as they have other shoddy implementations on the board. Since I was also trying to boot from a CF ide adapter, I had to disable the udma options in the bios. Except that even if I disabled them, various flavors of linux insisted they were still on. I don't know who to trust, but i'm leaning towards linux gurus. It's already known that the linux acpi does not work well with Asus.

The tech support rep, although stumbling his way through an explanation, also led me to believe that their sata2 support does not support any type of switching used for port multipliers. I confirmed this looking up Nvidias 6100 chipset. If I had at least 1 sata2 port that did, i could use an add on switch that supported 5 disks, all at full speed. Simultaneously. My pci-express 1x card, however, does support port multipliers, so I still have that avenue.

So, now I'm fighting my way through any decent install of linux that I could at least test with. My original intent was FreeNAS, but I couldn't even get it installed, and even the developer suggests not to trust it's software raid. Openfiler is next, and if that installs successfully, I'll at least have a test bed to calcualte how far I may need to backtrack in design.

Mega Hurts

After multiple attempts to install Fedora Core 3 on the thinclient, including upgrading from a 1gb card(Transcend) to a 2gb card (riData), frying a cdrom trying to use the onboard molex, and an install bug that would send me in a loop if I was just a bit over the space limit, I finally got it installed(Yay!).

Then I found that there really is no margin to the processor requirements for skype and SkypeMate. Even though fc3 on the 266mhz Pentium Netvista with 256mb of ram is actually quite usable, Skypemate seems to take most of the cycles. Skype alone seems to do ok when up and running. I imagine that the overhead to use the B2k adapter as an audio device is just too much. It is also possible that if WikiPedia is correct about Skype doing constant writes to the hard drive, which in my case is a CompactFlash card, that could be slowing it down as well. I managed to call my cell phone and roughly get connected, but the sound was more than choppy. Either way, the results kept me quite bummed.

I still like the idea, and if nothing else, it satisfies my curiosity temporarily. I had never used Fedora or RedHat before, and it's a simple, if not too simple, setup (minus recurring bugs). I would like next to see if it is possible to bump up the systems processor, using an AMD K6. I just happened to find someone who has already done this and recorded it. Back to eBay!