On Tue, Mar 1, 2016 at 5:17 AM, Grant Edwards <invalid@invalid.invalid> wrote: >> A lot of web sites these days load nothing but a script >> that goes and loads everything else, while you gaze at a splash >> screen. IMO that's unideal. > > :) > > "Unideal" is too kind -- I would described it as something more like > "the evil spawn of brain-dead incompetents". I often wonder what > sequence of decisions/accidents get people to "solutions" like that.
I do understand where the thinking comes from, though. 1) Pages should be more responsive than the 1990s-style CGI model, where every action is a link or button click that completely replaces the current page. 2) Many actions therefore must be implemented using JavaScript, an AJAX call (if necessary), and DOM manipulation to change just the parts of the page that need to be changed. 3) This means that the entire content of the page must be under JS control, and the entire "user-accessible" part of the back end should be returning JSON, since that's what #2 requires. 4) If everything has to be under JS control, why not just send out a stub to start with, and begin everything with an AJAX call? That way, we don't need to maintain two of everything! The logic is reasonable. Deduplication is valuable. But the result is a web site that has a stubby splash screen until the back end responds, and that's why I call it merely "unideal" rather than "evil spawn &c". >> However, even that is probably a losing battle. :( > > At least the era of using giant tables full of fragements of a single > large gif image to do page-layout seems to have died a long-overdue > death... Oh yes, I do not miss those days! And it was compounded by browsers that, when they started receiving a <table> tag, drew nothing whatsoever until they received the </table> and could lay the whole thing out. Add to that slow internet connections, so it would likely be some seconds or even a minute or more before the </table> arrives, and you have a perfect recipe for frustration. But no web site these days would ever make you wait for its contents. 'Course not. ChrisA -- https://mail.python.org/mailman/listinfo/python-list
