3 ms·
Yeah, I really dislike the black and white thinking. The python script example the author gives is a perfect example of what doesn’t need to be made any faster.
by opportune 7y ago
Yeah, I really dislike the black and white thinking. The python script example the author gives is a perfect example of what doesn’t need to be made any faster. If you are interested only in execution time, you might as well never write anything in native python.
But on the other hand, a lot of web content does need to be faster. Gmail has somehow gotten so much slower to load over time. And every time I visit a newspaper/magazine website I am aghast at how bloated they are. Does that mean nodejs is inherently bad, no, but it does mean people should try to optimize noticeably terrible performance that actually degrades UX.
- _bxg1 7y agoSure. But even that probably has more to do with businesses prioritizing features over quality, not programmers lacking character. Much of it on the web also has to do with how much browsers can do. The number of CSS properties that can be applied, the ways different elements' sizes can automatically influence the layout of other elements, etc. These traits are what make the web such a powerful and attractive platform for user interfaces, but the complexity of the platform is definitely becoming a real issue that deserves attention. A couple of points: - NodeJS is server-only and usually has nothing to do with perceived performance of web apps - The biggest offender of web performance is ads. They dump piles and piles of crappy JavaScript from dozens of different sources that all include their own copies of common libraries and have no incentive not to slow down the page. - Beyond ads, the bottleneck is usually not even JavaScript, but layout (as in the paragraph above). Web layout is incredibly flexible and incredibly complex. Computing and rendering it all is slow, but it does serve a purpose. Not that it couldn't be improved. Bad ads are a tragedy of the commons and I don't know what can be done about them unless Google or Facebook decided to throw their weight around to force them to be better. I do wonder if a new web standard could be developed for using some constrained subset of the layout vocabulary, that would be cheaper and more straightforward to compute. The current version has to remain for backwards compatibility reasons, but it's trying to serve a bunch of different types of cases at once, and therefore doesn't do a great job at any one of them.