This is an interesting concept I was just thinking about. Now, I don’t know how effective it can be, but it seems to be an interesting idea. I’d be happy to know your thoughts on this.
Today, frameworks like rails, django and even php frameworks are extremely popular. But with the power of these frameworks, comes more responsibility (Whoa, that’s philosophical).
In Web Hosting forums, a typical hot topic is not how to use these frameworks, since they’re very well documented in the first place, and are quite self-explanatory in their working. The main worry is performance. Can my server handle these many users? Will it crash?
These questions mainly come from shared type of hosting, which have lately started supporting python and ruby platforms. Coming from a PHP background, not many people might have experience on optimizing persistant memory usage of rails or python frameworks. Heck, what could look like a possible memory leak could actually be a feature which you might want to turn off. And the “apparent” solution of throwing more processing power and memory won’t be feasible for many.
Lets say you create an django application, and one dynamic page receives a lot of traffic. Normally, if the page was static, you’d use caching of course. But this one is something like say Digg’s spy tool. So what this page does is simply fetches information, constantly, directly proportional to the amount of traffic it receives, well more or less.
What this page also has is the django overhead loading every time along with it. Now, since I don’t know django very well, I don’t know if a possible solution to such problems might exist within django itself. But if you know of one, you could post a comment below about it.
Now, you could explicitly code this page in framework-less python (i.e. plain python), or if you’re not comfortable with that, even something like PHP (with apc) could do the trick. Now I’m not comparing languages, but simply saying that if you can save the overhead of the framework, you could at least attempt to see if it runs any faster. Now, I’m not telling to code the entire stuff without framework, because that defeats the purpose of using a framework in the first place. Less custom coding means less maintenance trouble.
All the above that I’m saying is with an assumption that there are overheads to frameworks, and they can increase with more feature additions, since this is what I read from various sources about various frameworks.
Your opinion on this topic could be valuable. If you’ve too blogged about something similar, you could post a link to it too
.
Yes, that’s an interesting point of view, but i find it only valid for really small, discrete parts of functionality. I always prefer the overhead of a framework versus the time of developing a web app from scratch which in the end, most of that code would result in a framework also.