Blog‎ > ‎

Building vs Planning, GC housekeeping, PyMongo, DB, ORM, Indexing, Brython

posted May 15, 2014, 9:03 PM by Sami Lehtinen   [ updated May 15, 2014, 9:05 PM ]
  • Why bother building something, instead of only planning it? This discussion started from Pintrest and their huge burn rate. But competition isn't always necessarily bad. At least it tells you someone else is also thinking it might be a good business. No competition at all, could be much worse. It's just very important to have some kind of twist to differentiate from other competition. I'm just writing one web-site as my hobby. I've analyzed over hundred similar kind of sites, and combined all features in a new interesting combination. It might be interesting, or not, but I'll give it a try, just for fun. I'll study a lot of tech stuff, as you might have noticed from my blog. But that only gives you a consult competence level. By actually doing it, you'll achieve something much more valuable. Because you simply have to think and deal with all the minor yet complex issues which might be more important for final success than you generally thought when doing just the high level plan for project.
    Let's say, we'll take a huge steel tube filled with explosives and light it up. It's simple as that. I just don't get it why they call is being as hard as rocket science. - Or maybe in reality, it isn't just so simple.
  • Garbage collectors mark & sweep, versus copy in real world. (G+) At office we're currently using copy gc. We got two fridges and one is purged on even and another on odd weeks. So if you don't move your stuff to the right fridge, it'll be gone in 14 days. Works very well. Before when we had only one fridge, we used mark & sweep gc. So everything in the fridge was marked at the end of the week with small stickers. If there were marked stuff at the end of the next week in the fridge it got purged before placing new markers.
  • Studied: PyMongo, MongoEngine, MongoKit, Minimongo, Ming, Humongolus.
  • Checked out: Query Optimizer functionality, Google Cloud Storage JSON API.
  • Got a new interesting integration, data analytics and reporting project for one major customer. 
  • A nice top 10 list of common Python mistakes.
  • Tails 1.0 released
  • Peewee ORM library didn't support Partial Indexes, so I had to extend it and add the feature to it.
  • Loaded a few more technical books to Kindle. I'll guess it's soon time for high lights post again. Anyway, this reminds me from the fact why Kindle is so awesome. When cycling out, you can just stop and sit down and have wonderful reading session in straight sunshine while listening birds. As well as a bonus, you don't need to carry those huge double brick sized manuals. - Why this is a bonus? - Well, when I were doing my military service, back in 1995, a few friends though that I'm going to be a barista. - Why? - I had three huge Java books with beans on cover. - Lol, right?
  • One guy asked, if Brython project got extensive documentation. This is my answer:
    What do you need extensive tutorial for? I usually like sites which got quite compact but good documentation. That's just one of the reasons why I like Brython in first place. As far as I know, there isn't any extensive tutorials or documentation, let me know if I'm wrong.
    I just somehow loathe projects with extensive verbose documentation, it's one of the things which is kind of putting you off, even before you start. I know it might be required for novices without any programming background to make things easier. I just prefer working examples which are easy starting point and then you can modify those to your needs, which is exactly what Brython already provides.
    I've been working lately with projects like Peewee and Bottle, those provide just perfect documentation level. You'll get what you need to know, but extensive or verbose blah blah nearing several hundred pages.
    As said, there's quite severe performance punch currently, but there are many things that can be done in future to alleviate that problem. One of those is that systems are getting faster and faster all the time, so it doesn't really matter for most of tasks if there's performance loss or not.
    I've also received a very nice email from Brython developer Pierre Quentel, but I'll get back to it bit later.