But one advantage I see for the Imperial System is that it's mostly based on factors of two and three, like two cups per pint, two pints per quart, three teaspoons per tablespoon, and three barlycorns per inch. The downside of that advantage is the sheer number of units available, like drams, furlongs, pennyweights and gills. I don't think it's that bad though.

For instance, the chocolate ice cream recipie I've been using contains the following ingrediants:

• 1 cup whole milk
• 1 pint heavy cream (or 2 cups)
• 1 teaspoon vanilla extract
• ¾ cup sugar (using 1 cup makes the ice cream too soft in my experience)
• ¾ cup cocoa powder

In the metric system, you get some really weird values though:

• 237mL whole milk
• 474mL heavy cream
• 5mL vanilla extract
• 150 grams sugar (using 200 grams makes the ice cream too soft in my experience)
• 72 grams cocoa powder

I think the amount for cocoa powder is correct, as I found answers from 75?95g of cocoa powder per cup, and as I like chocolate, I used ¾ the upper value.

I suppose one could get by with:

• 250mL whole milk
• 500mL heavy cream
• 5mL vanilla extract
• 150 grams sugar
• 70 grams cocoa powder

for more ?round? amounts in metric. I do wonder if such rounding up (or even down) might affect the results though (probably not). Personally, I find the Imperial version easier to remember, but that might be bias on my part.

An eraser (or as it states on the package, a gomme à effacer) in the shape of a human ear. And it also appears not to be for kids between newly born and 12 years old if I'm reading the small graphic icon in the corner of the package correctly. Maybe not for kids less than a year old? I don't know, it just has a small kid face with the text ?0?12,? encased in a circle with a slash.

I ? don't know what to say, other than ?Thank you!? but aside from that ? I have nothing.

?It's pronounced ?skwim.??

?Pardon??

?It's pronounced ?skwim.??

?So, you're saying the ?E? is silent??

?Yes.?

???

?It's Washington! What can I say??

?No. Don't say that.?

?You don't want me to go to the store??

?No, don't end your sentances with a preposition.?

?Really? How should I have said that??

??That's where I will be.? You don't need the ?at.??

And the next thing you'll tell me is to stop splitting my infinitives.?

?Pththththththththt.?

Of course it exhibited different behavior with different blocks. Most of the blocks it would work for just under five minutes, then fail. I found this out by writing some very simple programs?one to send some data once per second to a multicast address, and one to receive the data. I would run the sender on two computers, and the listeners also on the two computers. Both the listeners would receive data from both senders, and then as they approached five minutes, they would only receive multicast packets from the sender running on the same computer.

It was only when I switched to using the one non-routable multicast address block, 224.0.0.0/24, did things Just Work?. 224.0.0.0/24 is categorized as ?local subnetwork? and is not routable.

Sigh.

So now I'm able to use the multicast program I was using before, since it was always local to my home network anyway.

Other notes about the ?wireless service unit??it's reporting pages suck. There's the ?event report? page, which dumps data like:

TCP and UDP traffic is being stopped, but what TCP and UDP traffic? No indication, and there's no way to configure what is logged. Lovely.

The other log report is the list of current NAT sessions. It's more useful as it includes source address, destination address, NAT address, protocol, port numbers, and lifetime, but the table itself is capped to a maximum width, so making the browser window wider doesn't show more columns. Horizontal scrolling for the win? I guess? Sigh.

The port-forwarding feature is wonky. On my old router, I could set incoming packets from the Internet to TCP port 22 to be forwarded to my development machine. On the ?wireless service unit,? however, setting that up means all traffic to TCP port 22 gets forwarded to my development machine, even on the local network! I mean ? yeah ? it works, but it's not public traffic that gets forwarded, all traffic gets forwarded. I can work around that but it's annoying.

The ?wireless service unit? has also spontaneously rebooted itself a couple of times. Not enough for a pattern to emerge, but enough to be very annoying. And one time it failed to obtain an IPv6 address (which shouldn't change in my opinion but then again, I don't run the Monopolistic Phone Company) and I had to power cycle it to get IPv6 back.

And I can't shake the feeling that it's doing something to my DNS queries, even though I'm running a local DNS server ?

Developer gets shadowbanned by Hacker News and asks for a real IP ban instead

Sean Conner discovered he's been shadowbanned from Hacker News and would rather just be banned at the IP level if they don't want him there. SEAN'S BLOG POST | HACKER NEWS DISCUSSION

Unsupervised Learning NO. 495

I can see how the author might get that from if they only read the headline on Hacker News, but (to his credit) he spelled both my first name and last name correct, so he must have clicked through to my post (where my name shows up at the bottom of the page) but not read the actual post.

Wow!

But given the author writes about AI, perhaps he had his AI write the summary for him. Given the two errata he mentions about his previous newsletter, if he does use an AI, perhaps he needs a word or two with it ?