Produtivity, simplicity, smartness: Replacing the old languages by smarter ones
Looks like that we are not the only ones who truly believe Python can be an excellent, high productive application framework.
Just in quick read in LinuxDevices today, I found 2 more companies, including mizi Research, with python based application frameworks (if no based at least powered)
I know people can start saying “bla bla bla 2 cents in a million” or something else, but the point is, if you have been involved in any aspect of software development, it doesn’t matter if GTK or QT (the nowdays flame) that much… it matters actually if you need to use them using C, C++, C# or Java against the more lean and productive languages like Python, Ruby and why not (Vala*).
I’m not a developer, but I’m surrounded by them enough to see and realize the amount of impact the language can have in the whole team productivity.
In Canola1 we went all C, and even with basic request (I will exclude UI) the basic software took at least 4 times more effort to be build and maintained with a “acceptable” bug rate.
In Canola2 we went with Python, leaving C only for the scanner of media, and the basic software came out not only faster but also more powerful (feature wise) flexible (expansible in several more ways) and with at least 4x time less “lines of code”.
I, thinking now as “product designer” envolved in the whole process, it is REALLY clear the value of creating a product today running from the most powerful languages to enable a incredible fast development process, with very good results in short and mid term that clearly allows you even to stop the project losing less money. Why? Because you will have a more full featured prototype running a lot faster and time is (the old cliche) money.
Does it have a price? For sure! it’s heavier, slower, but when the architecture is well done it can be done quite nicely and the best part : it opens the way for improving the more high level languages even more. Maybe LLVM improvements, maybe some sort of revolutionary JIT, or even just simple optimizations that give power enough to solve people’s problem in the best time x cost x quality equation.
*I think is not a real “language” but does it’s job right?
If you think of C as its intermediate/assembly code, and the glib-gobject thing as its runtime, then of course Vala is a real language.
Btw, Canola 2 rocks big time. (And i’m not “pulling bag” here. :D)
February 9th, 2008 at 6:34 pmThe startup time of Canola2 is too big. I think, Python is one of the contributors to the long startup.
February 19th, 2008 at 9:37 pmFrom another hand, media scanner is a complete failure. It can’t read unicode tags, it can’t read id3v2.4 tags, it miss a half of my media files.
Vt: if GTK was not loaded on the memory it would take more or less the same time to load the media player, so I don’t agree too much, but for sure it’s a matter of optimizing binds.
The startup for canola is too big now mainly because of the database integrity, removing it, it’s showing almost instantly now. This is not only starting python, but also all the EFL libraries in almost no time, so I can for sure post here some benchmark to prove my points.
Th database check is mainly a “security” to the application to avoid crashing with corrupted database, but we already doing a version where users takes responsibilities for taking out and in card in use :) sometimes we try to protect the user too much and it causes things like that.
Yes, about the mediascanner, first version, you know what happens… now André did a great job and it’s find a lot more musics will be in next version, and the unicode problem we will kill soon ( we actually making sure that it will work with some users’s file from japan, china, russia, and so son). In another hand we are not able to tag almost all the formats supported by the platform and mplayer (including flac etc)
Let’s see how it goes, if you have good testing files, please send to us!
Br
Marcelo
February 20th, 2008 at 3:05 am