On Thu, Dec 11, 2008 at 8:41 PM, Henning P. Schmiedehausen < henn...@schmiedehausen.org> wrote:
> ^ > > b) undefined methods: (snip) These are all standard browser provided methods. jslint assumes that you're writing javascript for arbitrary execution environments and as such complains about anything not explicitly declared. This can be fixed by adding some funny comments to the source files I believe, but I have no strong opinion on the matter. The code is absolutely correct. > > c) unclear breaks: > Yep, that does look pretty ugly. > d) scary for loops > > (snip) Yep, hasOwnProperty checks should always be done in object iteration. Any code that does not do this should be considered a bug. > f) C'tor names > > This is more coding style than anything else. However, most C'tors > adhere to this, only those fail. Yes, I understand that they probably > can't be fixed; however then this should be documented. Jslint is a > standard tool an people will run it on our code base, pointing these > things out again and again. > jslint is wrong here. There is no ctor by those names. These are arguments to the functions in question. This is the only way to ensure proper scoping when constructing a variable object name. jslint assumes that nobody would be silly enough to want to call "constructors" in this fashion, but this is the only way for this code to work. Not that I'm defending the actual code in question here, though. I think this is a fundamentally wrong approach to javascript, but it hasn't caused any major problems so it's OK by me. i) functions in loops > > That code is a) unreadable and b) error prone. Needs fixing > (snip) There's nothing error prone about these, but I'll buy the readability argument for the sake of discussion. Javascript is a functional language and anonymous functions are a perfectly natural feature in the language that any competent javascript programmer would make good use of. jslint is complaining because of the potential memory overhead of these declarations (which are static and don't use any loop data) in a loop. In other words, it's making a performance argument, not an argument for code quality. You can find identical constructs in the javascript implementation of jslint itself, as well as every well-written javascript library in existence. It turns out that the memory overhead is real on IE6, and that's about it. So I'll accept that moving these anonymous functions outside of where they're being used for the memory savings in IE6 is worth it.
