[mcpherrinm]

I'm loving this. It didn't seem like the most useful feature when I first heard about it, but it seems pretty well implemented. (Using the current nightly).

The pagemod and inspect commands seem super useful. I hate having to click around an inspector to find the thing I want -- "inspect #foo > h3" just seems so much easier.

[agumonkey]

You made me investigate HN dom structure and it's quite scary !

[snprbob86]

It works, doesn't it?

If you have to interface with HTML or CSS directly and frequently while writing your application, then you want to make sure your HTML and CSS are super clean and maintainable. But, as PG has said before, HTML can be treated like object code! There was once a time when everyone had to write super clean and maintainable assembly code, but now you rely on a compiler.

Ever look at compiler generated assembly code? It's quite scary!

[icebraining]

I find that a shortsighted attitude by PG. Web pages shouldn't be just a black box to present to the user. They should be structured documents that is flexible and malleable, so that the user or third parties can filter, adapt and enhance the content.

http://userscripts.org/ is the best example of that, but there are many others.

[snprbob86]

http://userscripts.org/scripts/search?q=hacker+news&subm...

Seems to be a non-problem...

[agumonkey]

In this case it's naive redundant and inefficient object code. Tables and BR unless you're targetting pre-css browsers aren't really necessary. But I can be wrong. Anyway it's still more work and output that needed.

[MatthewPhillips]

I can't figure out how to expand nodes in their inspector though. I had to bring up the source to find out what selector I should use. You should be able to do it both ways.