ralph

As I write this, 4.55% "of those that answered" are female. That seems surprisingly high to me, so I think you could be right. It will be interesting to see whether the relatively high number of women will help keep the conversation civil and respectful.

I'm wondering this because in the UK it's understood that mixed schools are better for boys because they behave better in the company of girls, however, it's worse for girls because the boys still muck around enough to distract from education; girls on their own do better.

Another case where a computer manufacturer ended up designing their own CPUs was the UK company Acorn, developer of the BBC computers. A couple of existing in-house staff, one the author of BBC BASIC, sat down and came up with the first RISC design intended for home computers; the Acorn RISC Machine, now known as ARM.

After a while, Apple came on board and the technology was spun out of Acorn into the new Advance RISC Machines Ltd. with Apple as a shareholder.

http://en.wikipedia.org/wiki/ARM_processor#History

Using the heap to store stack traces when TCO is in use and a "debugging" flag is set won't work. It could cause memory to be exhausted. You're effectively causing each recursion to consume memory; something TCO is trying to avoid. If the coder assumes TCO then they could have written code which will exhaust memory once the "debugging flag" starts storing stack trace information.

I think having ~/.mozilla/firefox/$foo.default/chrome/userContent.css containing

    @-moz-document url-prefix(http://news.ycombinator.com/) {
        td { color: #000000 !important; }
        a { color: #000055 !important; }
    }

helps an awful lot with news.yc's appearance. The main improvement being the banishing of that awful low-contrast gray on gray text.

It's a crap title. A title shouldn't just grab attention; that's easy. I'm here today on this thread, but I haven't been on many others with similarly poor RSS appearance.

Can people putting up stories please consider those of us that use the site's RSS feed? All you get for this one is

    <item>
    <title>You&#39;re calling *who* a cult leader?</title>
    <link>http://lesswrong.com/lw/4d/youre_calling_who_a_cult_leader/</link>
    <comments>http://news.ycombinator.com/item?id=527064</comments>
    <description>
    <![CDATA[<a href="http://news.ycombinator.com/item?id=527064">Comments</a>]]>
    </description>
    </item>

which makes it a bit hard to judge whether to read it or not. Improving the title would be good. Especially since any explanatory text given at the same time as the title doesn't make it through to RSS.

I'd like to think if it's worth posting to news.yc, taking up the time of all those that consider reading it, it's worth putting some effort into a good title. If it isn't, perhaps it's not worth posting?

With hindsight, do you still feel fnids are a good way to go? They present an obvious bottleneck in that the server has to remember many closures for a period of time as opposed to having each client remember their share. I know you drastically reduced the use of fnids some time ago. I wonder if the remainder are a vestige that's OK to live with because life's too short, or there's fundamentally a strong argument for them I'm missing.

Remember my last settings when I return to the site. Provide a `reset to defaults' button if I want sanity to return.

Slide the old button image vertically and put the new one in its place. I'd like to be able to compare the before and after. Also, while it's sliding, I'm distracted from waiting for the new one, but don't make me wait for the slide to finish if the new image is ready to display.

Either wrap long text to fit in fixed-width buttons, or let me specify \n and \\ to break the lines manually.

Does anyone else look at the web page's source to see what bits of an essay are there but commented out?

By your reasoning, you think it shouldn't be possible to upvote comments after a couple of days either then?

What really annoys me is that pg took away the ability to downvote a comment after a day or two. Given I use RSS to visit news.yc every few days, that means I can't fully participate; I can upvote people's comments, but not downvote them.

And all because someone went around downvoting all the comments for a particular user they fell out with. It's a sledgehammer, pg! Come up with something better.

Interesting. Looks like Berkeley DB will be good enough for me to prove the concept then, and if, or seemingly when, it becomes the bottleneck I'll know it's possible to improve on it. Thanks again; it's great having first-hand access to those that have done it instead of just theorising about it, like me. :-)

Thanks for the detailed answer. I've a start-up idea that needs a big dag offline for the production of a smaller (in bytes, not nodes) one used online. My intention was to use Berkeley DB. It's good to see you say it's the fastest of the off the shelf options, but it only reaches 5% of your own code! Any thoughts on what BDB is doing "wrong"? Were you using hash or B-tree with BDB?

Thanks for taking the effort. Are tags implemented as just references from a tag "foo" node to the tagged item?

Still seems to be too many of them at 2009-02-27 12:40:10 +0000. Also, playback gradually slows done here on Firefox 3. I gave up after about 1160 on Michelle's replay.

Was it Ian Fleming that said you should never start writing a sentence unless you know how it's going to end? It seems we're now in an age where sentences are started without knowing how they're going to start.

Hi Colin, I referred to tarsnap and said it had a "good pedigree" on JWZ's blog. Yes, I was referring to your background and domain knowledge as FreeBSD Secuirity Officer. http://jeremy.zawodny.com/blog/archives/007641.html#comment-...

There was a syndicate at work when the UK lottery started. I persuaded them, well some of them, the rest went along with it without comprehending, that it's better to buy N tickets in one game than one ticket in each of N games. Further, we used a set of 210 tickets that guaranteed a `match 3 of 6', i.e. GBP10 win. (Tickets are GBP1 each.) So the syndicate saved instead of playing and once we had enough we queued and bought the tickets, causing a few grumbles from the Sainsbury's customers behind me in the queue.

We had just one ticket win anything. It matched 5 of 6, and our 6th was one away from the `bonus' ball. So we won about GBP3,500. If we'd have got the bonus, it would have been 100-ish times more!

I tried to convince my fellows that we were lucky, but, instead of saving gradually for the next play, they put some of their winnings back in so they could play the 210 tickets again the following week. They won GBP10.

Actually, 211 tickets were played, because quality checks showed one has been miss-filled but naturally we didn't want to drop the erroneous one. :-)

They hate D&D on the systems you've seen them use. I've seen them use it for many years on an OS called RISC OS, written by Acorn, designers of the ARM chip. There, users loved it. Everything supported it on the desktop. You had `filer' windows, and to save an application would just pop up a little window with a file icon and a text box for the leafname. Enter `foo' and drop it on the filer window where you wanted it.

But better than that, you could drop it on another application. Consequently, you could create and edit a pixmap, and drop it into a vector drawing program where it would become an embedded object. Want to edit it again? Drag it from there back to the pixmap editing program's icon on the icon bar. Neither of these touched the disc; it was in memory transfers. And the program code didn't need to know the difference between saving to another app., or saving to disc through a filer window.

Other aspects of the desktop were also novel and worked well. It had a three button mouse. A proper one. No silly scroll wheel. The right button did what the left button did, but altered in some way. So you'd pop up a context sensitive menu with the middle button. Select an item with L and the menu would disappear. With R the menu would stay there. Great for when you want to do several menu items in a row, e.g. adjusting an object's attributes.

Scrollbars worked better. Firstly, the mouse cursor on the screen only moved in one dimension once a drag with L was started. None of this cancel the drag if you wander off orthogonally a bit. But if there were two scrollbars, dragging with R would control both; an instant `pan' widget.

Hold the `up' arrow with L and it would scroll up. Switch to R and it would scroll down. No need to move to the other end of the scroll bar. Same with `paging' by clicking either side of the thumb.

Drag a window by it's title bar with L and it'll come to the front of the stack at the same time. Use R and its depth is unaltered.

Double-click a folder icon in a filer window with L and a new window opens with its contents. Use R, the current window switches to show that directory. Close a filer window with R and it becomes the parent folder.

Note, you didn't need to know all these uses of R to use the system. You could manage with L. But the paper User Guide it was shipped with explained it and you soon got used to the productivity gains.

This was back in the 1980s and 1990s. Today, you can find remnants of it in the ROX (RISC OS on X) Desktop. http://roscidus.com/desktop/

Acorn did some smart stuff. Today, the main vestige is the ARM chip and its instruction set, designed by Sophie Wilson.

The last element of an array is foo[-1] regardless of it's length. :-) And try implementing a circular buffer using an array of N elements with 1-based indexing and then 0-based. Sorry in advance for the loss of hair due to the former.

Sometimes, pv(1) doesn't help. For example, `tar cf - foo | bzip2 -9v >foo.tar.bz2'. You don't know the size of the data that needs to pass down the pipe. But I sometimes find watching tar(1) open the files to read is handy; `strace -e trace=open $(pidof tar)'.

Excellent. Thanks very much. I've made them black and will now set about fixing some of the other poor colour decisions, e.g. the body of the top post of a thread being an unreadable gray which puts me off reading it through no fault of the author.

pg, How about doing something about this complaint? At 31 votes, it's my highest scoring comment. :-) http://news.ycombinator.com/item?id=460859

True. Although are you aware you can select the text of the URL and, in Firefox 3, drag it up to the tab bar and either drop it on an existing tab, e.g. the current one, or drop it slightly to the right of the right-most tab to create a new one. If doing the latter, you should see a downward pointing arrow when you're over the drop zone.

pg, please re-visit the decision to have the post's text in gray-on-gray. It's awful for long initial posts like this one. No doubt I'm not the only one that selects the text to get it the slightly better yellow-on-black. It makes me less likely to bother reading the text through no fault of the writer.

It's .co.uk .de .fr and others too.

In the U.K. we don't have air-con. It can be hard to sleep during the summer days.

So a `DateTime = datetime' in the module would let newer code look more conventional?