Dailydave mailing list archives
[Fwd: Re: IronPython.]
From: Matt Hargett <matt () use net>
Date: Fri, 10 Sep 2004 10:48:24 +0000
Dave Aitel wrote:
http://ironpython.com/ is a .Net implementation of Python that runs onthe CLR or on Mono.Originally, I thought it was going to be a failed experiment. I assumed, perhaps naively, that the CLR and .Net's architecture in general couldn't handle a language as truly dynamic as Python is without a major performance hit. If you look at his latest slidepack though, you'll see I was wrong. He runs Guido's latest benchmarks (meant to be brutally rigorous) 4% slower (overall) than CPython. In many tests, he was a lot faster than CPython. The amazing thing is that he managed to do it at all. Also, it's amazing that Mono can run it - it doesn't run it as well as the Microsoft CLR, but it runs it well enough to use. This means you could write a pyGTK that wrapped GTK#, and all your pyGTK apps would run on Mono. pyGTK is still, by far, the fastest and best way to write GUI's on any platform, imo. My opinion is that the reason Miguel doesn't like what he sees of Avalon is that Avalon is not as easy to use or conceptually clean as GTK# (which is very similar to pyGTK - libglade, etc). Once you've used the pyGTK/GTK# toolkit, there's really just no going back. We don't want to be professional GUI developers - we just want to look like we are.
Miguel explained his thoughts on Avalon/WinFX in a blog posting a while ago. I think it was here: http://www.go-mono.com/monologue/ . I personally can't figure out what the fuck the point of WinFX is. Are they trying to make it so non-programmers can design UIs and not have to touch code? You get that now with WYSIWYG drag and drop designers like VS.NET already has. If they think XML will make it more accessible, then they are making the same retarded mistake that the XP people made with FIT tests and that Rational made with UML. Most marketroids I have worked with have a kind of invisible barrier as to what they can, or are willing to understand. XML might as well be C#, UML (that is worth anything) might as well be a 20 page flow chart, and a FIT test table might as well be XML. And they shouldn't have to know that stuff -- keeping all the variables of a market in your head is quite enough for anyone to be getting on with, IMO. If you've got a technical project manager (as one should), then dumbing it down is just annoying in the first place. If the point is to displace SVG by making an XML language that is generic enough to make UIs *and* attempt to replace Flash, then that is evil and makes sense. Interesting to see that XUL (from Mozilla) caught on somewhere. Maybe the SVG standard will be extended (or already has) to do this UI stuff. Anyone know? The CLI/CLR in 2.0 has some changes to make dynamic language more.. "native" so they they can interoperate better with non-dynamic languages. IronPython is a really cool example of what can be done, but there are some hacky things on the backend to make it work. I think Brad Abrams (whose blog URL I don't have handy, but it's something like blogs.msdn.com/brada) dropped some hints about further changes in .NET 3.0 for dynamically typed languages. I think someone also mentioned some similar things in a video interview on channel9, but I can't remember who. I'm personally waiting for Tcl.NET :) _______________________________________________ Dailydave mailing list Dailydave () lists immunitysec com http://www.immunitysec.com/mailman/listinfo/dailydave
Current thread:
- [Fwd: Re: IronPython.] Matt Hargett (Sep 10)