Ontario, Canada
Stephan has an interest in software freedom, human-computer interaction, user interface/experience design, programming, and Linux... but he prefers to leave graphic design to the experts.
Stephan has an interest in software freedom, human-computer interaction, user interface/experience design, programming, and Linux... but he prefers to leave graphic design to the experts.
Authored Comments
I was aware that ICEcoder could be run locally but that still ties into my point about browser-based apps being heavier.
The whole point of using things like Vim and urxvt is that they make it easier for me to have "many programs open". Hence why I think that using tools like Vim is "working smarter". I get the same functionality but waste less CPU and RAM, which means it's easier to do things like having copies of of Firefox Aurora, Firefox Stable, Chrome, Opera, Midori (GTKWebKit), Arora (QtWebKit), and three or four of the http://www.modern.ie/ testing VMs just sitting open in the background for quick-access testing.
Naturally, I couldn't do that on the 2GHz Celeron but, if I just got lazy and let my desktop grow a beer gut, I might have trouble doing it on the dual-core Athlon too... especially if I'm developing and testing something that is, itself, CPU- and RAM-intensive.
(I actually didn't plan to upgrade to a 3.4GHz dual-core Athlon II with 16GiB of RAM. I was perfectly happy with an Athlon64 X2 5000+ (2.6GHz, dual-core, also 65W TDP) with 4GiB of RAM that I bought about 5 years ago but my old motherboard died back in January. Since I was buying new kit anyway, I decided to make room to run more than one IE testing VM at once.)
You also seem to misunderstand my point about Vim. Even if you implemented Vim keybindings, it wouldn't interest me because I don't use Vim keybindings. I use <em>custom keybindings and plugins within Vim</em> and the amount of effort required to port my <em>.vimrc</em> and chosen plugins over to <em>any</em> other application just isn't worth it.
True. I know browser extension load-outs and the design trade-offs between Chrome and Firefox make a big difference.
Chrome is memory-heavy but responsive because of its multiprocess design (though <em>window.opener</em> support causes multiple tabs to share the same process) and I can't stand a lot of the design decisions that went into its extension API.
Firefox's single-process nature helps it to be memory-light and its approach to extensions makes it easy to implement things like HTTPS Everywhere and RefControl, but they push so many things Chrome does naturally (like matching the system's native scroll-wheel behaviour for tab bars) into extensions that said lightness is a bit deceptive... and I'm still trying to track down the last of the slow extension memory leaks
Worse, and more relevant to this case, a messy site in one tab can bog down every other tab.
Given how I find Opera to be unacceptable for being closed-source and Chrome for making various design decisions I dislike and can't override with extensions, I'm stuck with Firefox, which means I'm effectively stuck with Ye Olde Windows 3.1 Cooperative Multitasking.
Basically, I'd have to run a separate Firefox or Chrome profile just as an application runtime environment and doing that just for ICEcoder doesn't make it feasible.
Perhaps in the future, but I just don't see the web runtime environment (or whatever you want to use as a general term) being mature enough to satisfy me right now.