Ubuntu has some tools around i18n to make it easier to translate desktop software into many languages. Currently, some of that depends on stuff in CDBS (adding the gettext domain to .desktop files, stripping the translations from gconf schemas, etc.). This is all in /usr/share/cdbs/1/rules/langpack.mk. I would like to use debhelper instead of CDBS in my packages because I really like slide 45, but it doesn’t (yet) support this specific functionality that is mandatory for applications that go into Ubuntu main. I have no idea (yet) how to port this functionality, so I decided to start by writing down the goal. I’ve just started looking into debhelper code which is very nicely factored and documented, and here is the code (doesn’t seem all that impossible) that runs inside langpack.mk:
Now that Jane Silber is becoming the new CEO of Canonical, we need to hire a replacement Director of Online Services (this is mostly Ubuntu One but also some other stuff) to be my boss (I run the engineering team for Ubuntu One). The job description says that London based candidates are preferred, but exceptions will be made for exceptional candidates. I wonder what that implies about all the people living in London…nah, it doesn’t mean that. Londoners are great. I just don’t want to live there.
If you think you’ve got what it takes to oversee Canonical’s Online Services strategy and product development including profit and loss responsibility for the division, and you will make my life less stressful and have lots of zen or improv type skills for making calm plans out of a fast paced startup-like business unit, then please apply! http://webapps.ubuntu.com/employment/canonical_DOS/
Jane says someone with an MBA is preferred, I say someone who uses vim is preferred. If you’ve got both you’re a shoo-in. The perfect candidate would be someone who has run an open source company before.
I’m interested in the RTMP video over http protocol because I’m interested in video streaming. From reading the wikipedia entry, I saw that Adobe published the RTMP spec. Surprisingly, it had a license accompanying it – I’m not used to seeing a license next to a specification, so I read the license rather than downloading the spec. Shockingly, the license contains this statement:
Prohibited Uses
The rights and licenses granted by Adobe in the RTMP Specification, including those granted in
the Patent License, are conditioned upon Your agreement to use the RTMP Specification for only
streaming video, audio and/or data content and not to make, have made, use, sell, offer to sell,
import or distribute: (i) any technology that intercepts streaming video, audio and/or data
content for storage in any device or medium; or (ii) any technology that circumvents
technological measures for the protection of audio, video and/or data content, including any of
Adobe’s secure RTMP measures.
It’s hard to believe there is a software developer on the planet who can promise not to make, have made, or use any technology that intercepts data for storage in any device or medium. How could anyone possibly make use of this specification given such a license?
If you ever need to do backports, I recommend the chdist tool. While working on one.ubuntu.com, I have frequently found myself needing to try out packages on both Ubuntu 8.04 and Ubuntu 9.10 (currently under development), often backporting a package from Karmic to Hardy. I’ve been running 9.10 (Karmic) on my primary laptop since the first alpha in order to work on packages included in the desktop, but we also need to run many of those same packages (Erlang, CouchDB, python-desktopcouch) on our server farms in the data center. Last week Tom Haddon showed me chdist, which makes it considerably simpler to work on backports, especially grabbing sourcepackages from an older distro version. http://packages.ubuntu.com is always nice for checking which version of a package is in the last few versions of Ubuntu, but chdist is even handier, since you can build APT trees for several different releases on the same machine, without requiring much disk space. Now I just need to finally learn how to use kvm, and I’ll be able to test the backports as well as make them.
_why’s disappearance has been all over the net in the last 36 hours. My favorite work from _why was the travelling IRC bot troupe. And when I saw the documentation to shoes in the form of a ransom note, I was inspired to try ever harder at combining art with work, and life with living.
I don’t go to funerals, not even for my own little brother, yet I feel strangely compelled to say goodbye to _why, who I never met or talked to. Maybe I’m sad about both of them. Here’s a song from Daniel Johnston I recently discovered, in memory of _why, and everyone who is gone and used to be here injecting life into our lives:
Worried Shoes – Yip / Jump Music (1983)
I took my lucky break and I broke it in two
Put on my worried shoes
My worried shoes
And my shoes took me so many miles and they never wore out
My worried shoes
My worried shoes
oo oo oo oo oo oo oo oo oo oo
My worried shoes
I made a mistake and I never forgot
I tied knots in the laces of
My worried shoes
And with every step that I’d take I’d remember my mistake
As I marched further and further away
In my worried shoes
oo oo oo oo oo oo oo oo oo oo
My worried shoes
And my shoes took me down a crooked path
Away from all welcome mats
My worried shoes
And then one day I looked around and I found the sun shining down
And I took off my worried shoes
And the feet broke free
I didn’t need to wear
Then I knew the difference between worrying and caring
‘Cause I’ve got a lot of walking to do
And I don’t want to wear
My worried shoes
Working on ubuntuone.com, we have a few different WSGI applications, and some WSGI middleware that we run across all of them. Some of the WSGI applications are Django, some are custom built, and we’re planning to deploy some based on lazr.restful. Currently we run a mildly complicated proxy in front of the WSGI servers made out of a mix of Apache, squid, and haproxy – then we run several instances of each of the WSGI servers behind the proxy. We are using paste.http, but I want to switch to something with some more options. At the top of my list currently is spawning, but there are other interesting options like cogen and of course mod_wsgi. Although there are lots of interesting servers to choose from, it’s hard to tell a clear winner or one that people are definitely having good success with in medium to large sites (sites with proxies, multiple app servers, multiple databases, etc.). I’m working on packaging spawning (and eventlet and greenlet) for Ubuntu now, but if there is another amazing WSGI app server out there I’d love to hear about it.