Real-life optimization work
Posted Nov 2, 2005 19:56 UTC (Wed) by elanthis
In reply to: Somewhat uncritical article
Parent article: All hail the speed demons (O'Reillynet)
And of course, most of the optimizations being done are low-level micro-optimizations to code, not huge massive code redesign efforts. It's not at all a problem of bloat, it's a problem that a coder of any moderately large application or framework would run into, be in GNOME, KDE, the Linux kernel, glibc, X, etc.
I'm getting really sick of this "bloat" mantra. Bloat would imply that there is excessive and useless code. That isn't the case, at least so far as most of Frederico's work has shown. It's simply a series of the kinds of optimizations that just aren't possible to predict until you have a stable codebase that is in use by a wide array of real-world applications.
No programmer could predict the necessity of many of those optimizations during the initial development of the software. Anyone *trying* to predict those kinds of low-level optimizations during initial development are idiots, and all they are going to do is waste tons of their time writing ineffective optimizations and causing their code to be unmaintainable down the road, making efficacious optimizations quite difficult to implement.
People need to quit with the "bloat" bullshit. When you see a music player that has a built-in file manager and RSS reader, by all means, shout "bloat!" When you see someone micro-tuning an incredibly complex (by real-world necessity) text layout engine, shouting "bloat" just displays your ignorance.
to post comments)