Present Perfect

I have seen some pretty badly named projects out there, but when I first heard about gNewSense some time ago it immediately shot into the top 5.

I don't know what irks me so much about it. Is it the camelCase carbonCopy of iPod nAming tHat oVerwhelms mY aEsthetic sEnses ? Is it the clever wordplay on GNU ? Is it the fact that the name calls up something vague and nebulous in my mind, instead of either perfectly describing what it is (youtube) or being mysterious (google) ? Is it the fact that everything you read about the project itself makes it sound as if it is trying to fill all the wrong holes, is a few years behind the times, and is bottled from sour grapes ?

I just can't put my finger on what it is that really irks me to the point where I turn into a critizing idiot about some project I have only a vague knowledge of.

Gee. What a nuisance.

Before I need to make some new year's resolutions and try to be a better man, one more thing to make fun of. I don't remember where I saw it, but at some point I came across ideas for a new video container, of all things, that was named YOY.

Is it just pure coincidence or blind sarcasm on the part of the creator that YOY is read out "Why, oh, why ?". Exactly the thoughts that come to my mind when someone plans to write a new video container. Google isn't helpful right now with my vague recollection, so let's just hope I dreamt it up.

I have seen some pretty badly named projects out there, but when I first heard about gNewSense some time ago it immediately shot into the top 5. I don't know...

Haven't said much about work for a while now. Since Andy is shedding, now's as good a time as any.

Today was a bit of an adrenalin rush. Spain has a gambling mentality, and today was the annual Christmas lottery draw. It's affectively called "El Gordo" (the fat one). This year, the total prize money totalled over 2 billion Euros. The whole ticketing system is an elaborate scheme of 180 series of five digit tickets costing 200 euro, but you can buy a "decimo", which is a tenth of the ticket. If your ticket wins, you win one tenth of the prize money for the ticket.

If I follow correctly, it means that there are possible 180 first prize winning tickets, and thus potentially 1800 people getting part of one of those first prizes, worth 3 million euros each.

The lottery is broadcast on national TV - a 5 hour extravaganza of introductions then the drawing of small wooden balls, where the number on each ball is sang out loud by an orphan, while the value is sang out by another orphan (I kid you not. This is serious business).

Normally I would not be watching or even consider partaking, but it just so happens that we were supposed to be streaming this extravaganza. So if I was going to sit through watching it anyway, I might as well just buy me a ticket.

So it started this morning at 8, and traffic to our platform slowly started rising, then quickly ramped up as the streaming link was published. The stream peaked at close to 600 Mbit/sec, with about 2800 people at the time where the first prize was drawn.

Sadly, due to a peering problem between our provider and Telefonica, we were stuck around that limit. So I guess the next thing to work on for us as a company is to start improving our connectivity situation.

But on the software side, we've now set ourselves a new bandwidth record for a single stream. The last time I felt that adrenalin rush of having something you helped make achieve something beyond my expectation was our first public streaming at GUADEC two and a half years ago - where at the last minute we switched out a Python-based byte pushing system to a new GStreamer element, multifdsink, and thus broke away from a maximum limit of 30 concurrent users :)

So, it looks like everything Andy is saying about our next release is ringing true. Flumotion is going Enterprise.

Now, if next year I could actually win the lottery ...

Haven't said much about work for a while now. Since Andy is shedding, now's as good a time as any. Today was a bit of an adrenalin rush. Spain has...

Err, I don't know what other way to say this, but your plugin doesn't
actually work.

My advice to anyone who doesn't know how to say this: try harder.

Err, I don't know what other way to say this, but your plugin doesn't actually work. My advice to anyone who doesn't know how to say this: try harder.

Sometimes I wonder what we could have done better with all the collective brain energy expended over shell tests and booleanity. Imagine solving an important problem with it ?

Instead, every so often we get to wrestle with the impications of


$ true; echo $?
0
$ false; echo $?
1
$ python
Python 2.4.4 (#1, Oct 23 2006, 13:58:00)
[GCC 4.1.1 20061011 (Red Hat 4.1.1-30)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> int(True)
1
>>> int(False)
0


It's only two values. Maybe the problem was shell authors didn't know C, only shell ? You'd think this would have been worth standardizing on, where was Mr. Posix back then ?

Yes I realize the usefulness of being able to return different exit values from a program and the accepted practice of exit value 0 meaning "everything's fine, move along, nothing to see".

Sometimes I wonder what we could have done better with all the collective brain energy expended over shell tests and booleanity. Imagine solving an important problem with it ? Instead,...

So, after sitting on my own code for more than six months, I finally got round to knocking off one of those items off my TODO list and I did a first release of MOAP.

moap aims to be a swiss army knife for project maintainers and developers, collecting small bits of functionality related to project maintenance, which in the BM (before moap) era I used to do with various random scripts and hacks.

I've been using it myself for a bunch of projects, including GStreamer, Flumotion, mach, Savon and of course moap itself.

In fact, the release notes were sent out with a simple moap command, and the update to freshmeat was done with a slightly more simple moap command.

MOAP currently does stuff like checking in only files based on the latest ChangeLog entry (much like cicl), update VCS ignore files, send out release mails and submit to Freshmeat based on a DOAP file. In the future, I am planning to add features like the prepare-ChangeLog.pl (my days are just too short to be hacking PERL) script, get tickets fixed by milestone from trac, announce on more websites, ...

I've also added DOAP files to all these projects mentioned before, I hope DOAP sees more and more uptake in the future.

I don't know why it took me so long to put a first release out - I was locking myself into an interdependent cycle between all these projects I've been wanting to release. Ironically working on a release tool caused me to release less in the end. In the end it was probably a simple mental barrier I had to talk myself out of.

I was close to a release at some point, and after that weekend Martyn released his maintainer script. It was so close to what I was working on that I thought I should contact him first or take a look at maintainer.py, and I just never got round to it. I lost track of the "release early, release often" mantra. Anyway, this weekend I decided I would leave myself no excuses anymore and just get it out. And anyway, the more people who take an interest in working on ways to automate release and maintenance work the better.

Now I can focus on putting out a release of mach and savon and go back to business as usual...

So, after sitting on my own code for more than six months, I finally got round to knocking off one of those items off my TODO list and I did...