dlang.org Library Reference

[H. S. Teoh]

On Mon, Dec 10, 2012 at 01:58:30PM -0500, Nick Sabalausky wrote:
> On Mon, 10 Dec 2012 09:45:06 -0800
> "H. S. Teoh" <hsteoh at quickfur.ath.cx> wrote:
> 
> > On Mon, Dec 10, 2012 at 12:34:13PM -0500, Nick Sabalausky wrote:
> > > On Sun, 09 Dec 2012 18:10:42 -0800
> > > Ellery Newcomer <ellery-newcomer at utulsa.edu> wrote:
> > > 
> > > > Is anyone else noticing e.g. std.datetime taking upwards of 30
> > > > seconds to render the blob of links at the top? It's freakin
> > > > freezing my entire browser.
> > > 
> > > It only takes about a second or two for me in FF2. I'm not seeing
> > > any blob of links at the top though.
> > 
> > 'cos you're running with JS turned off. :-)
> > 
> 
> <g>
> 
> It makes the web 10x faster, 10x more practical and 10x less
> obnoxious - it's the "Make the web instantly better" checkbox ;) It's
> almost magical!

Heh. I used to browse with no JS. But due to the annoying bandwagon
jumping trend that is a common pathology in people involved with
computers, more and more websites are starting to depend on JS and are
mostly (or completely) dysfunctional without JS. So I grudgingly turned
it back on.

Mind you, though, what with browser bloat and JS memory hogging plaguing
my browsing experience recently, I've tentatively switched back to no JS
by default, and enable only on a per-site basis (Opera is good for that
kinda thing). There have been websites that outright crash my browser or
soak up all available RAM and then some (causing the browser to be 99%
I/O bound and the pig which is X11 to essentially grind to a halt), and
then I ban JS for that site, and boom, it's instantly better. Just like
you said. :-P


> > Also, I think the actual results depends on your system. If you have
> > a high-powered system you probably wouldn't notice too much lag. It
> > "only" takes about 3-4 seconds for me on a dual-core 3.4GHz Intel
> > machine. But older machines will probably see a bigger lag.
> > 
> 
> Yea, at the same time though, if it takes a dual-core 3.4GHz (with
> two levels of cache, out-of-order execution, pipelineing, branch
> prediction, SIMD, external GPU, etc) 3-4 seconds to render a few pages
> of formatted text, then something in computing has gone very, very
> wrong.

I said the same thing when Java first came out, and the earlier JVMs
used to gobble up 2GB of RAM and 99% CPU upon startup, all just to
display a "hello world" message. :-)


> > But anyway, I think all of this is kinda missing the point. The
> > point is that the blob of links at the top of the page is plain
> > unhelpful.  It is only useful if you already know what you're
> > looking for, in which case you could just use your browser's search
> > function to find the matching text instead. Much more useful is a
> > broken-down categorization (with proper nesting, etc.) of the
> > functions, classes, structs, etc., of the module. Something like
> > what std.algorithm does.
> 
> Looking at it with JS on, yea, I see what you mean now. Not so nice,
> not so helpful, and definitely not worth the rendering lag. The
> new-style baked-in summaries on pages like std.algorithm are much,
> much better.

I would go so far as to propose that we get rid of those unhelpful link
blobs completely. They make the page slow to load, and for no real
benefit. Sounds like a lose-lose proposition to me. But I'd probably get
lynched for saying that. :-P


T

-- 
English has the lovely word "defenestrate", meaning "to execute by
throwing someone out a window", or more recently "to remove Windows from
a computer and replace it with something useful". :-) -- John Cowan