Earlier this week I wiped my work phone (a Motorola running Android) in preparation for installing the new mandatory corporate spyware. One of the policies the software enforces is locking the screen after two minutes. I spend a lot of time driving and glance at the map or messages while I'm waiting at a traffic light. The screen lock makes that impractical so when the phone offered to enable voice activation I thought I'd give that a go. Step one was choosing an activation phrase. I tried "work phone" but the phone baulked, claiming the phrase had too few syllables. I tried "work telephone" only to received the same error. I'm happy to report that my work phone now answers to "unimaginable piece of crap".
I tend to be ambivalent about AMD. I first became aware of them back when they offered a 40 MHz 386DX (Intel's topped out at 33 MHz). I ran a 450 MHz K6-2+ for a while in my primary desktop PC. It wasn't long after that AMD went off the rails with Athlon, which was fast for its time but hot enough that I wasn't about to build a PC around it. I did run an Athlon XP box for a while that was donated by a kind friend who understood enough to make sure it was properly cooled.
My wife had a single-core Athlon 64 PC that was later upgraded to a quad-core Phenom II X4. If you were careful, it was possible to pick a quad-core Phenom II with 6M L3 cache and 65W TDP. After that AMD started building APUs that had much less cache because of die space that was taken up by the added GPU. I understood the concept and I know it was popular with our local PC shop but for my uses they weren't appealing chips.
Occasionally AMD would release a one-off chip that seemed interesting. I still secretly crave a Kabini chip just to potch with; to see how much use I can wring out of a quad-core 25 Watt chip with a single memory channel. I had my eye on Delhi too: an 8-core single-socket Opteron with 65W TDP. I had almost given up on AMD when they released Ryzen, a series of chips that might be a worthy successor to the Phenom II. Plenty of cores, plenty of cache. Affordable quad-core options with an upgrade path to octo-core if the need arises. I might build one soon if only to stop my daughter complaining about how slow her desktop PC is.
I've been potching recently with a Raspberry Pi 2. It's a promising little board, especially because of its GPIO header. Despite the quad-core processor, it's probably not going to replace my (ancient) work desktop. It boots NetBSD/evbarm and seems to run X11 nicely but the big challenge is storage speed: I'm using it with a microSDHC card that reads at 16 MB/S but only writes at 3 MB/S. A USB flash drive turned out to be even slower. Because my Raspberry Pi only has 1 GB of RAM, any large application is likely to swap and then everything grinds to a halt. That said, rdesktop seems to work nicely so it's plausible as a graphical terminal, despite the 100M Ethernet port.
One of my favourite 16-bit computers was the Atari ST. It shipped in 1985 with a simple single-tasking operating system called TOS, which included lumps of CP/M-68k, DOS Plus, GSX and GEM, all from Digital Research. I liked the consistency and simplicity of TOS but in the late 1990s I tried a multitasking operating system called MiNT. I used it with a GEM program called TOSWIN that let me run multiple text-based programs, each within its own window. If I'd had a hard disk I could probably have added the Minix filesystem and MiNTnet to produce something that resembles BSD. This evening, while I was looking for something else I stumbled across this interview with Eric R. Smith who created MiNT. It taught me some interesting things about the thinking and history behind MiNT.
Every now and again I'm asked to give someone a way to connect to a Windows application server. Usually this is for a visitor but I have one or two other users who find it convenient to have a terminal for this on their desk so that they don't have to break out a laptop and wait for that to boot. We have a shelf full of old desktop PCs that were retired when Windows XP was killed off. Here's what I do:-
- Install NetBSD/i386 6.1.5
- Install rdesktop, an excellent remote desktop client.
- Create a user called "termuser".
- add the following to termuser's .profile:-
startx clear exit
- create a .xinitrc file for termuser:-
rdesktop -fd MYDOMAIN servername
Things to do: have the terminal authenticate against Active Directory or LDAP and then pass those credentials through to the application server, so the user only sees one login screen.
There are times when I find it convenient to run a program or park some files in a place that's out on the Internet, "in the cloud" as the marketing people like to say. Lately I've been using BSDvm.com for that. For just under US$ 10/month I chose a single-core NetBSD VM with 256M RAM and 5G disk. This may sound tiny but it's perfect for the text-based applications I run on it and, unlike many other VPS services, includes unmetered bandwidth.
I had expected to have to install NetBSD on the VPS but it was delivered as a freshly-installed NetBSD/i386 instance that looks and feels just like it does on bare metal. One surprise benefit was that when I "experimentally" erased the VM's hard disk, I was able to contact tech support who had it re-imaged almost immediately and at no cost. Because it has a public IP address, I can ssh into it and even run things like a Web server that I've used as a convenient way to get binary files to people without having to explain sftp. OpenBSD and FreeBSD are also available but I used NetBSD because I'm already familiar with it. I find the BSDvm a convenient thing to have around.
The Android smartphone that I used every day for work eventually failed. It was fine as a tiny tablet but no longer worked for phone calls. When word of this reached our service director he called me into his office and opened a desk drawer full of "gently used" Apple iPhones and dug out an iPhone 4s. One or two people in my department use iPhones so it seemed likely that I'd be able to find the apps to do my job and I thought it would be an opportunity to try out an iPhone and iOS.
The phone itself seemed nice enough, though the glass back seemed ridiculous because the phone slid around like Bambi on a frozen pond until I found a nice orange case for it. The single button at the bottom of the screen confused me a bit: I'd poke at it like a caveman, wondering why it didn't behave like the back button on Android. A colleague took pity on me and explained that this was the "home button" that would (logically enough) take me back to the home screen. A double-press takes me to a task list that lets me kill things off.
The phone's battery life is abysmal. I have to plug it in every few hours or it craps out, leaving me without a connection to the office. I've heard that iPhones have weak batteries but batteries do age with use, so perhaps it's not fair to judge them all by this one example. I have been pleasantly surprised by how fast the user interface feels and the stability of the apps that I've tried so far. I wonder whether iOS applications are built against cleaner, faster (and fewer?) libraries.
Unlike the Android phone I'd been using, the iPhone 4s doesn't support "4G" LTE speeds. For most of what I do that's not an issue: I just have to wait a bit longer for downloads when I'm not on WiFi. I was able to find iOS equivalents to most of the apps I use and I was pleasantly surprised that Google apps like gmail and Google Hangouts were available for the iPhone. The calendar app won't sync against the MS Exchange account that tells me where to go every day and nobody seems able to figure out why. Because of this I have to bother a friend in the office to have him read Outlook to me. We're high-tech like that.
Today I have a server to repair or perhaps rebuild. It's an old HP ProLiant ML110 G4 that we bought new over a decade ago. It has a dual-core Intel Xeon processor and just 2G RAM but with NetBSD and Samba that's probably still adequate for a simple file server. I'm told it won't power up, so perhaps it just needs a new power supply. If the problem is more serious I'll have to find another server and reach for the backup tapes.
Recently I've considered OmniOS as a server operating system. Like NetBSD it's deliberately minimal. One possible advantage is its support for ZFS (yes, I know we'd need more RAM). Something that puts me off OmniOS though is an apparent lack of documentation. It seems to be built with the assumption that the server admin already knows Solaris or OpenSolaris and most of my SunOS experience dates back to when SunOS was a BSD.
There are some other alternatives to unix: I could try Linux or even Windows Server. I'll probably test these at home but NetBSD has been very stable for us so I'll stick with that for the production server. I'm a firm believer that "if it aint broke, don't fix it". The hardware's broken though, so let's hope I can fix that.
I would like an electric car, partly because of the rising price of petrol. My previous job was about 30 miles away and a Nissan Leaf would have coped with that commute even if my workplace had no provision for charging. My current job has me working in the field and there are days when I drive 200 miles or more. Today reminded me why I can't buy an electric car: I had no time to stop and charge, even for a half-hour top-up. Another concern is the annual cycle of very cold winters and very hot summers. I suspect that might impact the lifespan of the battery.
Driving home I passed one of the filling stations that sells E-85. I've heard some debate about whether its reduced fuel economy offsets the lower price. I'm told it burns a bit cleaner though and it would reduce the amount of fossil fuel I get through every month. Sadly none of the nice little cars that interest me (Chevy Spark, Nissan Note etc.) have the "Flex Fuel" badge that tells me they can drink it.