The plan, such as it is, is to do the following.
Well, I realised late today, that my open-haggis project site was down. Some misconfiguration of varnish and supervisord had resulted in the site not 'bouncing' when I reset the server a while ago. So, it's fixed now but it has prompted me to get back on track with making changes to open-haggis. I need to make a triage list of things to fix/enhance and get cracking. I have been letting open-haggis languish while I got on with my dayjob. I can't let that continue. If you are interested, you can see what I am talking about at www.open-haggis.org
I started to write this from the perspective of "Ruby has this and Django don't" but I realised that was wrong, and started again.
Django implements Multiple Table Inheritance, and to a degree that works nicely, BUT, because the data for a single object is strewn over several tables, it can get VERY slow to iterate across a large collection of objects and resolve exactly which object we have.
Depending on the hierarchy, it can take many 10s of milliseconds to find the real object we want. That is not so bad, but when
you are iterating across a collection of a couple of hundred items, those 10ms add up to considerable delay.
As a consequence, you try to do everything you can to avoid having to resolve to the 'real object'. This leads you into sin.
The real issue is that once you have started down the path of OO sinfulness, you start trying to be clever about it, and the code gets not just sinful, but fiendish. You start off with a nice clean OO structure, that requires you to resolve to the real object, and when you
hit situations where you are iterating over a large set of objects and so you get clobbered by 100s of 10ms piling up, you take a shortcut, changing your OO structure to try to remove the requirement to resolve the objects. This good intention gets layed down in
front of you like a paving slab, making it look like you are making progress and keeping things neat, but all around you the monsters are gathering, sharpening their knives as your path becomes ever narrower, darker and deeper.
Now, I know that all the Django heads out there are going to say, 'well, you should have done it THIS way'. Well, maybe next time I will, but right now, I am down here fighting monsters and trying desperately to stay afloat.
Now, I don’t know about you, but I like SPAM.
Yes, yes, I do. specially pressed American meats.. Salty unidentifiable slabs of animal protein and…
Pork, Salt, Starch, Water, Ham, Sugar, Trisodium Diphosphate and of course, those yummy YUMMY flavourings, antioxidants and preservatives (Sodium Nitrite), De-lish.
Hey, I am a geek, so I had to go get the real ingredients. I am a fatgeek so I happened to have a can (or three) handy.
Now, Nigerian 419 spam? not so much.
In fact, I just received one.
What makes these things possible is that NOBODY responds to them. Well, almost nobody. And the people that do, are the ones they WANT to have respond to them.
But…
What if.
What if, everyone who received a 419 spam, instead of binning it and getting on with their lives, ANSWERED it.
It would take less than a minute a day. Just respond to one a day (if you get one).
What would happen at the other end?
Suddenly the spammer is getting back 100s (thousands, tens of thousands?) of responses instead of one or two.
And to do his business, he would have to READ them.
This makes the work on their end that much harder.
How much would they put up with before the job gets too expensive?
I dunno, but I am willing to find out.
Are you?
Well, I have finished the coding for the radio capture system.
I can now click on a radio .RAM link of the BBC website, and a request will be queued for download. A cron job will pick up the request and download the raw radio as a WAV. Another cron will take this wav, and encode it as an MP3 with the appropriate ID3 tags. Finally, another task will pick up the MP3, and use the tags to move it over to my MP3 archive directory and insert records into the MusicBox database so that it can be picked up by the Shoutcast server.
The only bit left is to increase the power of the requester script to allow scheduling of periodic tasks so that I will never miss an episode of “The Archers” again!
I have been informed by my current client that the project I have been working on has been shelved indefinitely.
Scrapped.
This is the first time in my career in IT that a project I have worked on has not succeeded in its objectives.
Previously I have always taken pride in the fact that when I finish on a project, my product as part of that project will continue (probably) long after I have left.
To not have that feeling, to have the exact opposite.
To know that all the creative work put in, was deemed to not be satisfactory for purpose. To know that as I leave the project, nothing of my effort will go forward.
Its a profoundly depressing experience.
I am a bit of a music buff.
My CD collection runs to some 400odd CDs.
As you can imagine, 400 CDs take up some significant amount of space.
I really need to be able to play my CDs from any room in the house.
I’m a coder.
So, combining these, I came up with the idea of using a Rails app to generate an index to my CD collection (ripped into MP3 files) that can be served via shoutcast.
Heres the project page.
Written by