Well, they didn' t like that much I can tell you.

[venta]

A quick question for Visual Studio users (with an ancillary part for anyone who writes C/C++/possibly Java/possibly others). Throughout the following, I mean {} or () by "brackets" :)

Is there an option anywhere in Visual Studio which I can set which means that when I highlight/hover over a bracket, it will in some way highlight or indicate its matching friend ? Yes, I know when you first close your brackets they're in bold text, but that goes away as soon as you start typing something else.

I just asked a VS user here, and he (a) had no idea and (b) seemed baffled I'd want that. Now, up until this recent enforced flirtation with VS I've written C/C++ exclusively in Emacs. Which allows you to set all kinds of different manners of highlighting for brackets. I find it phenomenally useful and am currently boggled that any serious text editor intended for programming doesn't offer this.

So... is this a reasonable thing to want to, or a strange quirk of mine bred by too much use of Emacs ? And, more importantly, can I make VS do it ?

Irrelevantly, just because I want to know:

[Poll #1028846]

Comments

[hjalfi.livejournal.com]

I have just been doing float-point implementation on PowerPC.

Arrgh.

However, if you want a chunk of BSD-licensed code that calculates 4 or 8 byte IEEE floating point constants from a string representation, I know where to get one...

[venta.livejournal.com]

Today, I encountered the scariest bit of code I've ever seen (which relied on IEEE float format). It was a quick square-root function for 16.16 fixed point input, and utilised this:

http://blog.int6.net/wp-content/uploads/2006/10/FastInverseSqrt.pdf

Which (until I googled for the magic number and found the above) was wildly impenetrable.

[hjalfi.livejournal.com]

That is deeply impressive. I'll have to remember that.

Oddly enough, the PowerPC (yes, I've been spending far too much time immersed in that sodding "what do you mean, the R in RISC stands for 'reduced'?" abomination of a "of *course* it makes sense to number bits with 0 as the MSB and 31... er, 63... er, n as the LSB" processor) has an instruction, frsqrte, that calculates exactly this:

http://www.nersc.gov/vendor_docs/ibm/asm/frsqrte.htm

It could well use that algorithm...

[two-nukes.livejournal.com]

Ah, that old chestnut, it's been in the 3d games community for yonks as a fast square root used to normalise vectors, though don't think it's used as much now since the vector instructions of a lot of modern processors generally implement a fast approximate reciprocal square root for this sort of thing these days.

Good old Newton Raphson root finder with a sneeky initial approximation using the fact that IEEE FP have exponent and mantissa, halving the exponent is always a good way of starting to head towards the sqrare root.