I’ve accumulated many notes (2000+) in Evernote over the years, and love that it can store binary attachments such as images or other media files. My favorite feature is the Evernote Web Clipper browser extension; it does a fantastic job at saving the parts of an article I want to save while keeping the styling intact.
Evernote has a free plan which I’ve enjoyed for a long time, but recently the financial status of the company has come into question, and they restricted syncing to only two devices. Also, the last thing I want to happen is another kind of Google Reader shutdown fiasco. I doubt that a shutdown would make my existing notes disappear, but it’s better to be prepared ahead of time. To that extent, I’ve been looking for a viable option to migrate my notes into another platform. Continue reading “Moving Evernote notes into WordPress”
I would really like to rid myself of Dropbox, but all the alternatives I’ve tried are too bloated, beta- or alpha-quality stage, too complicated to set up, or just plain don’t do what Dropbox does (minus the sharing stuff, which I don’t care about). I don’t want btsync, it’s closed-source. Seafile is too complicated, and makes dubious security claims. Owncloud is a cool project, but their file sync is slow, error prone, and has other limitations. There are some good services, but they don’t run on all the platforms I need, including Mac OSX, Linux x86 (32 and 64-bit), Linux ARMv6 (my Raspberry Pi B) and Android. I ran Syncthing for a while, but the continuous memory usage is pretty steep for the Pi, and I’ve experienced random silent file truncation in my shared directories with it. So I needed something else. Continue reading “Unison dependency hell”
Yes, the Raspberry Pi can do fast video encoding. Of course you normally wouldn’t want to re-encode any video with an ARM processor, but that’s not what we’re going to do here. We’re going to leverage the GPU. I should point out before proceeding that the input formats for re-encoding are limited in this method, more about that below.
In order to do this, I’m using a proof-of-concept tool called omxtx, which I think is supposed to be a shortened form of “OpenMAX Transcoding”. Off the top of my head, here are the prerequisites for building the binary from source:
Raspbian. It will probably work on other RPi distros, but I haven’t tried them.
The build-essential package installed, which you normally need to build anything.
Memory split of 64MB for video. I previously had this all the way down to 16 since I don’t use a display on my Pi, but bumping it to only 32MB caused runtime errors from the omxtx binary. You need to give the GPU some breathing room to encode video.
There’s probably some libraries you may or may not have installed that the build wants to link in. When I run ldd on my finished binary, it loads all kinds of media libs like libav, libvorbis, libvpx, etc. YMMV.
While trying to move an older code base to a newer system and thus a newer version of PHP (5.3 -> 5.5), I knew that some of the code would need to be changed to avoid using some removed features. Specifically, I mean call-time pass by references. For those who don’t know, this is kind of a weird feature of earlier versions of PHP that allows one to call a function and pass any of the arguments by reference rather than the usual call by value if the caller prepends an argument variable with the “reference to” operator &.
So, to illustrate, normally this code won’t have side effects because of call by value:
$var="I'm doing it: ".$var;
However there will be side effects if the caller chooses pass by reference:
I thought a regex might be in order to find these guys and fix them:
It only took three years for the frustration to build up sufficiently. :-) Finally I went a little deeper to try to solve the mystery of the 30 second hang time before requests to my Linux guest VM would return. I love doing testing on a Linux guest VM via VirtualBox while working on my Mac, but the occasional slow down (with no load on the VM) drove me to the brink of madness.
I had assumed for a while that it was some complicated (potentially Mac-related) DNS asshattery. No. It’s the routing table, stupid! I finally noticed that this wasn’t right at all. $ route -n Kernel IP routing table Destination Gateway Genmask Flags Metric Ref Use Iface 0.0.0.0 192.168.1.1 0.0.0.0 UG 100 0 0 eth2 10.0.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth3 169.254.0.0 0.0.0.0 255.255.0.0 U 1000 0 0 eth3 192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth2 192.168.56.0 192.168.56.1 255.255.255.0 UG 0 0 0 eth4 192.168.56.0 0.0.0.0 255.255.255.0 U 0 0 0 eth4
So after some fiddling, I added this to my /etc/rc.local: route -v add -host 192.168.56.2 gw 192.168.56.1 eth4 route -v del -net 192.168.56.0/24 gw 0.0.0.0 eth4
It is a host-only interface after all. Now the table looks far more sensible. $ route -n Kernel IP routing table Destination Gateway Genmask Flags Metric Ref Use Iface 0.0.0.0 192.168.1.1 0.0.0.0 UG 100 0 0 eth2 10.0.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth3 169.254.0.0 0.0.0.0 255.255.0.0 U 1000 0 0 eth3 192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth2 192.168.56.0 0.0.0.0 255.255.255.0 U 0 0 0 eth4 192.168.56.2 192.168.56.1 255.255.255.255 UGH 0 0 0 eth4