3 ms·
I know that's in the RFCs, but that's the kind of thinking that caused the web standards nightmare everyone has to deal with today. If you accept incorrect inp
by smanek 17y ago
I know that's in the RFCs, but that's the kind of thinking that caused the web standards nightmare everyone has to deal with today.
If you accept incorrect input, then people will expect you to continue to do so into eternity - and if you have significant market share (facebook, IE, etc) you've just created a defacto (ill-conceived, poorly documented) standard that every other implementor now has to try to reverse engineer and try conform to.
- kragen 17y agoThat's also the kind of thinking that made the internet and the Web workable systems, while Hyper-G, OSI, SNA, DECNET, Xanadu, and so on ended up in the dustbin of history. Yeah, it's too bad that at some point someone has to build quirksmode.org and write the HTML5 standard, but that's better than not having widely-used web standards to deal with in the first place.
- wmf 17y agoThe alternative to Postel's Law is usually data loss. When your mail server receives one of these malformed emails from Facebook, what should it do? Delete it? Bounce it? (Facebook will just delete the bounce.) Print it out and mail it to Zuckerberg COD?
- smanek 17y agoProbably alert the user that it can't be parsed (like browsers currently do for XHTML). Maybe do something like firefox does when it encounters a self signed SSL cert - allow the user to accept it - but make that choice explicit and painful. If most parsers actually did this, I doubt people would carelessly violate standards anymore. Of course, it's a bit of a prisoners dilemma to get everyone to actually do this. The one guy who doesn't will likely become immensely more popular than his competitors ... Oh well, it's nice to dream.