outbreak

Today I found something in Javascript that just doesn’t make any sense. Ok, maybe some other things don’t make sense either, but I think this one just beats all of them.

Let’s say we have a string and a number:

a = "1";
b = 1;

Now we want to compare these:

alert((a>b) ' : ' (aa) ' : ' (b==a));

Both will give you ‘false : false : true’.

If we change a to “2”

a = "2";
b = 1;

we get a ‘true : false : false’. We’re still OK.

Now let’s change a to “a”

a = "a";
b = 1;

The weird thing is we get ‘false : false : false’ on both alerts.

In typestrict languages you’d get an error. Or at least you should. In php you get a ‘false : true : false’ and ‘true : false : false’ which would mean they decided that all strings are smaller than numbers. This is weird – if you make b a string “1” the return values change – but at least they made a choice.

For what I can comprehend Javascript casts both values to int and I can understand that. The core of the problem is obviously with the NaN value. What I can’t understand is how something’s not smaller, not bigger and not equal at the same time? I’m sure someone can explain this…

The reason is really simple – they’re bloated. Widgets is another word for a huge javascript that does what you want/need and a whole other stuff somebody needed or maybe just the developer thought somebody would someday need. Usually the javascript is so ‘complex’ that you can’t just remove the stuff you don’t need – it’d be faster if you wrote your own.

There is another reason that goes beyond javascript. Most widgets don’t exploit the technology and go about reinventing the wheel which results in crappy/invalid HTML, problems with data persistance and a low level of pluggability. Some actually do use an API but I think that’s overkill.

It’s funny what wikipedia says about widgets – clearly, they represent any purely mercantile commodity that has no artistic or spiritual value. When I first started writing the SortedTable I wanted to accomplish the contrary. I tried to write something that would exploit the advantages of valid XHTML and the fact that you can nowadays rely on DOM functionality of the browser. I did it for me and not for money (and released it under CC license).

The problem is I got many feature requests, that could simply be accomplished via some DOM scripting and using the built-in event handlers. I decided to keep the script simple and display the possibilities through samples. What needs to be done next is the reimplementation of the sort function which is a bit slow – I already have an old implementation of Quicksort in javascript that is going to come in handy for this one.

Anyway, a new version is out. This is the changelist:

• corrected a weird error that broke some sorts in ie6
• added the possibility for a nonsortable column via class=”nosort” on the th cell
• added the possibility to ‘regroup’ tbody elements if there’s more than one (for whatever reason)
• added additional hook to put your custom javascript on (onsort)