Wild to get a "This project has received too many requests, please try again later." message for what I assume is just a static page. Makes me make a note to self not to use glitch.me
Even when I've got a front page article in the past on HN at like 150 votes I only got about 6000 views in a day. This one has like 10 likes so I doubt it's getting 10s of thousands in a minute. Can only imagine it's some severe rate limiting to push glitch.me users for some paid alternative.
When some of my submissions have hit the front page I got a lot more that 6K views in a day, although probably 10K in some minutes is an overestimate, possibly by a lot.
Agreed that it's probably glitch.me pushing people to a paid option, but there's no real problem with that ... they are a business providing a service.
okay but this was maybe a few hundred thousand requests over about an hour for 536 bytes of html and two js files of respectively 2232 bytes and 2667 bytes
like, an old netbook could serve up all those requests in 100 milliseconds with apache
on a 9600 baud modem (5 kilobytes 300_000 times an hour is under 4 kilobits per second)
are you hosting glitch.me on an amiga 500 on dialup in zimbabwe or something
it sounds like you accidentally set your request rate limit about four orders of magnitude lower than the other limits, because you set them to 4000 requests per hour, which is about a ten-millionth of a laptop
by contrast the 512 megabytes of ram is about a 64th of a laptop and the 712 megabytes of disk is about a thousandth of a laptop
this results in absurd situations like this one, where a purely static web app collapsed under a load you could literally handle on a commodore 64 on dialup
well, i guess with tls you might need a 25-MHz 386 on dialup
To be clear, we’ve got Glitch projects doing a lot more volume than that, especially for static apps. This one was hitting our rate-limiting but that was a bit of an outlier and should be fixed now.
I just had to check whether you were 'the' Colin Wright, and looking at your profile I think you are. Small world!
I am a fellow juggler, and used to run the Internet Juggling Database. We also very briefly bumped into each other in a corridor at Leeds Uni Physics department about 25 years ago ;-)
You don't have contact details in your profile, but feel free to email me. I travel a lot and can put you in my geographical address book so that if I come within reach we can get a coffee.
Off-topic, but what do you use for your "geographical address book"? I would love to have something that showed all my contacts on a map, but haven't run into any software that does this ever.
It will come as a surprise to no one who knows me ... I wrote my own.
More, I created a system wherein people can register, and then whenever someone is travelling they can tell the system and it automatically emails everyone they know who is close to the destination.
I'm the only one who uses it, although there are a few hundred people on it, so I basically just use it as my geographical address book.
I also put landmarks, museums, and other places I might like to visit. When I tell it where I'm going it then sends me email to remind me of them.
That is so amazing! I also attempted creating such a concept but never got anyone of my friends to use it... I ended up heavily using (and abusing) the Google maps share your location feature and I was so excited to see they are adding features related to notifications when people get close - only noticed it a few months ago. Telegram had also released similar features but only when you share your location live witch you can only do for a limited amount of time. Nevertheless interesting to see these kind of features are slowly getting implemented even if easy too late and not close to how I would like and imagine they could be... I also have my live location broadcasted here: https://chagai.website/location.html when it's not broken
I also wrote my own system for pretty much this exact purpose, although instead of emailing me reminders, it just lets me record notes on a map and then I skim that map whenever I am planning a trip to a given country or region.
It's online (but registration is invite only, let me know if you want to play with it) over at https://scoutmaps.io/ – at one point I was going to attempt to commercialize it, but eventually decided I liked it better as hobby software so that I can write features that better serve my current userbase of one :)
Have you watched for more than 5 minutes? Each hand jumps from one Division to the next, increasingly flexing in between jumps. It's working fine for me, and obviously for others, so either there's something odd about your setup, or it's just doing something you don't expect.
Do hands release from the line they're on at the point of the next time increment being reached (meaning, by the time they hit the next line, they'll be late) or does it kindly release earlier, affording time for the hand to travel to and arrive at the next line in time for it to represent the current time?
Just by observing it, when the machine clock clicked to 14:20 (UK time) the second hand snapped to the vertical and the minute hand snapped to the 4 (the 20 minute mark).
Does that answer your question? I don't know what I can do that you couldn't do for yourself, so maybe I'm not understanding your question.
I believe GP's question was: does the hand get released at 14:20 or 100ms before 14:20 to have it reach its position at exactly the right time, accounting for the animation delay.
I don't see how a real life one could be made to look exactly the same (but I could be missing something).
The seems to have something that stops the end of it moving on until enough tension has built up to side underneath it. That much is fair enough. But then that same thing stopping the end would also prevent the same satisfying wiggle when it gets to the next position because it overshoots immediately.
The only thing I can think of is that the restraining pins move in and out is the clock have at just the right time but that would be quite a fiddly implementation and probably affect the visuals
You’d need a mechanism at each 5 minute mark that grabbed the hand and released it at the appropriate time, then you could make the hands a s
Wobbly as you want
A physical implementation could use magnets to hold to the specific spots. Once enough torque is built on trying to move the hand, it would snap out of place and would then find the next magnet and 'hold' onto that spot.
As an aside, I'm not sure that doing this for the hour hand is a "good" idea. The difference between 10:00 am and 11:00 am is significant and needing to check how much the bend is doesn't "feel" right.
(late edit thought) If we're getting to the "this is controlled by some other system that doesn't need to be purely mechanical", having an electromagnet on at the right spot that then toggles off at one 5m increment and on at the next so that there's only one magnet active at a time could reduce possible issues of having the wiggle/snap attach to the wrong magnet.
The thing that I wonder about (with nails) would be "what would happen if the the flexihand was able to wiggle past the peg when it moved from one peg to the next?" In that case, I believe, it would just be a regular clock style hand.
It's possible that I've got the wrong mental model for how that would work.
I think you'd put the nails at N+1 position, not N. (e.g. for the minute mark, you't put the nail at 6, 11, 16, etc. not 5, 10, 15). Then, the hand would press against the nail and progressively bend until it slipped past the nail entirely and jumped to the next 5 spot (and not yet be touching the next nail).
Without over complicating it, I wonder if you had a flexible second hand like a spring, that's just a bit too long, with pins on the 5 second hand, dots on the rest and decent motor to drive the shaft. As the shaft turns the spring builds up tension and eventually bends enough to jump past the pin.
Made me think about the Swiss Railways clock that has the second hand completing the circle in 58.5 seconds; all the clocks on the network then wait for the sync signal from the “master clock".
I love those clocks! But from memory, the second hand gets "stuck" at 12 and then when it moves again, the minute hand advances at the same time. On this one, the minute hand jumps forward at the same time that the second hand goes to 12.
The second hand is supposed to go round the clock in 58.5 seconds, and then 'wait' for a synchronisation signal before starting off for the next minute.
The animated example in https://en.wikipedia.org/wiki/Swiss_railway_clock#/media/Fil... seems to have the first behavior you mention, it get stuck at 12, minute hand increments and then the second hand starts moving again after the minute hand movement.
Both systems could be found. Early to mid 20th-century master clock systems were pretty much all designed around a common theme but there were many variations. Some secondary clocks even used fully free-running second hands that weren't synchronized at all and so had no relation to the minute hand advancement.
The clocks in my elementary, middle, and high schools would do something similar. Not every minute, but every hour. At the top of the hour, when there was 15 seconds left, the second hand would make 3 second jumps until it hit the top, and then hold until the next hour started.
I always thought that was a brilliant way to keep mechanical clocks in sync. And the system had been there since the 1970s, so it was well before digital clocks were feasible for a school.
Our elementary school clocks (70s) made a variety of nearly imperceptile sounds as they got near the top of the hour, or turned it over. We were like dogs when they hear the about-to-eat sounds start.
The details depended on the system. The most common arrangements were either polarity (hour pulse opposite polarity of minute pulse) or a separate wire for hour pulse. Later systems introduced serial digital communications but were never as widely installed as the minute and hour pulse systems in schools.
They had digital communication just not cheap ICs. I don't know how it worked but I imagine it was something like a mechanical switch that held the minute hand until it got a signal from the main office to start moving again.
The clocks could even be set for Daylight time centrally, so I assume it had a way to send a signal that triggered a motor to move the hour hand faster.
Since it's a school I wonder if they could just hold them back 1 or 11 hours on the weekend for DST changes. This would eliminate the need for an additional signal and circuitry on each follower clock.
In the Standard Electric Time Company slave clocks that I've got, the master sends about a 12ms pulse once a minute, which pulls in a solenoid and allows a lever to ratchet one tooth on the minute wheel and advance the movement. Then once every twelve hours, the master sends a longer pulse, which pulls in a stronger second solenoid that allows the wheel to turn freely. There's a weight on the hour wheel which will pull the wheel around until the weight is on the bottom because of gravity, which corresponds with 12 o'clock. So no matter where the clock is, it will fall back to twelve with the longer pulse, and then begin the ratcheting again with the once a minute shorter pulses.
(Sorry for the deprecated master/slave terminology, it's unfortunately what they used at the time)
Thanks for that. We had those as well. But third-grade me never really thought about the purpose. So it's fun to think about it now.
In contrast, I'll sometimes now walk into a lobby or reception area of a business where they will have like five different analog clocks on the wall representing NY, Paris, London, etc, (trying to look "international" I guess) and the minute hands are out of sync. Kind of sad in 2023. :-)
My gym has about 15 different clocks on the walls, and they all tell a different time, some of them egregiously so. Every few months (often after I mention it), someone will go around and fix the worst ones.
My first job was at a Network Operations Center, and it was just like in the movies, with a big projection monitor, two network operators at terminals below, and above us were four analog clocks depicting the four main time zones of these United States. They were electric, and I guess they were battery-powered because of the lack of cords going everywhere.
When Daylight Savings Time began, I sauntered in about 9am and my coworkers were filtering in too. The morning guy, "Matt" had been there since about 6am and he had changed the clocks. So my other coworker, "Jason" congratulated Matt for setting all the clocks correctly and then he asked him, "so did you do it by setting all four clocks ahead one hour, or did you move them all to the left and then move the Pacific clock to the Eastern position?" Facepalm!
That might be a good job interview question, or it would have been if manual, analog clocks were still a thing.
GPS would be tricky to receive in a typical hotel or office lobby (i.e. far away from any windows at the ground floor, possibly in an urban canyon).
In Germany (and in neighboring countries), radio-controlled clocks are quite common for this purpose, listening to DCF77 near Frankfurt [1].
Long wave seems to work a bit better than the L-band for indoor propagation, and it probably also helps that the transmit power is three orders of magnitude higher.
I'm not sure if the US has an analogous service – there's WWV and WWVH, but they are shortwave transmitters, and I don't know if wall clock or even wristwatch size receivers are possible for that.
> I'm not sure if the US has an analogous service – there's WWV and WWVH, but they are shortwave transmitters, and I don't know if wall clock or even wristwatch size receivers are possible for that.
It does. Small WWVB clocks that self set are quite common.
I think you can still buy NTP reference devices for your network that use GPS. So, the lobby of the hotel doesn't have to be within view of the sky, as long as their reference device is.
There are no additional pulses. Only minute pulses. It's a motor driving the second hand. The motor starts by the minute pulse and stops itself when reaching the top.
To deal with any inconsistencies in the speed at which they complete the 58.5 seconds. If a particular defective clock takes 59.9 seconds instead of 58.5 it'll be just fine, as it will start going again when it gets the next minute pulse from the central system.
This was my first thought when I saw it too, I immediately found myself making a sound effect for each tick. It could definitely benefit from a tasteful sproingy sound effect!
Watching it really bothered me, too. After considering it for a while, I think it's because when I see something bending like that IRL my instinct is to make that stop happening so the thing doesn't become worn, permanently bent, or broken.
You need to manually create a canvas at screen resolution and scale it to fit the screen. A default canvas has pixels that are logical pixels, not actual pixels.
I can't tell, since I don't have a Mac, but looking at the code, it implements a note from the Safari documentation for canvas [0]. I assume that if you don't do this, it either is very small or very pixelated.
I saw somewhere on the internet a clock where the hand of the seconds moves at random irregular intervals, but averages out to 60 moves a minute. Extremely frustrating to watch.
Edit: here's a video of it running (somehow the seconds hand runs backwards)
Neat! We built clocks with CAD and a laser cutter in my high school engineering class, and I remember cheekily making mine unqiue by numbering it backwards and/or without noon at the top
> This clock is inspired by a post by ILAMtitan on the 43oh.com forums, which was in turn inspired by Vetinari's clock from the Discworld series of novels. Although the movement of the second hand is quite erratic, it ticks exactly 32 times in 32 seconds, and so keeps quite accurate time overall.
The clock started life as a cheap $3 clock from KMart. It's mechanism was kept, but the driver circuit was removed and replaced with an MSP430 microcontroller. ILAMtitan's code is designed to ensure that every 32 seconds it will tick 32 times, but the exact timing of those 32 ticks is variable. There can be as little as 1/4 second between ticks or as much as 24 seconds. Overall though it will keep quite accurate time.
Reminds me of a wall clock in my high school physics class that would struggle to pass 5 o'clock and ticked backwards and forwards a few times on its rotation through each minute, despite keeping good time. Its struggle through each minute reflected my own struggle through each minute in that class, however.
But the way the minute hand stays pointing to a multiple-of-five-minutes until it Boings! to the next increment is disconcerting.
It looks like it's slow all the time. E.g. 10:29 looks like 10:25 which is more than enough error to look wrong.
Maybe I'd learn to read the bent-hand deflection after a while. But I guess I prefer the clock I have that just points to the right minute all the time.
207 comments
[ 2.9 ms ] story [ 278 ms ] threadEdit: D'aww. Hug of death.
i feel like i am really missing out on something here
I'm getting the author to move it.
Stand by ...
OK, here:
https://somethingorotherwhatever.com/wobble-clock/
worth noting that anyone can remix the project :) https://glitch.com/edit/#!/wobble-clock
Agreed that it's probably glitch.me pushing people to a paid option, but there's no real problem with that ... they are a business providing a service.
We do have some default protections in place to prevent (even accidentally) problematic projects from taking too many shared resources.
We've temporarily removed those barriers for this project since it was getting such a big hug. It should stay up and be Remixable now.
disclaimer: I'm on the engineering team at Glitch.
like, an old netbook could serve up all those requests in 100 milliseconds with apache
on a 9600 baud modem (5 kilobytes 300_000 times an hour is under 4 kilobits per second)
are you hosting glitch.me on an amiga 500 on dialup in zimbabwe or something
by contrast the 512 megabytes of ram is about a 64th of a laptop and the 712 megabytes of disk is about a thousandth of a laptop
this results in absurd situations like this one, where a purely static web app collapsed under a load you could literally handle on a commodore 64 on dialup
well, i guess with tls you might need a 25-MHz 386 on dialup
https://somethingorotherwhatever.com/wobble-clock/
I am a fellow juggler, and used to run the Internet Juggling Database. We also very briefly bumped into each other in a corridor at Leeds Uni Physics department about 25 years ago ;-)
You don't have contact details in your profile, but feel free to email me. I travel a lot and can put you in my geographical address book so that if I come within reach we can get a coffee.
Cheers!
More, I created a system wherein people can register, and then whenever someone is travelling they can tell the system and it automatically emails everyone they know who is close to the destination.
I'm the only one who uses it, although there are a few hundred people on it, so I basically just use it as my geographical address book.
I also put landmarks, museums, and other places I might like to visit. When I tell it where I'm going it then sends me email to remind me of them.
It's online (but registration is invite only, let me know if you want to play with it) over at https://scoutmaps.io/ – at one point I was going to attempt to commercialize it, but eventually decided I liked it better as hobby software so that I can write features that better serve my current userbase of one :)
Does that answer your question? I don't know what I can do that you couldn't do for yourself, so maybe I'm not understanding your question.
The seems to have something that stops the end of it moving on until enough tension has built up to side underneath it. That much is fair enough. But then that same thing stopping the end would also prevent the same satisfying wiggle when it gets to the next position because it overshoots immediately.
The only thing I can think of is that the restraining pins move in and out is the clock have at just the right time but that would be quite a fiddly implementation and probably affect the visuals
As an aside, I'm not sure that doing this for the hour hand is a "good" idea. The difference between 10:00 am and 11:00 am is significant and needing to check how much the bend is doesn't "feel" right.
(late edit thought) If we're getting to the "this is controlled by some other system that doesn't need to be purely mechanical", having an electromagnet on at the right spot that then toggles off at one 5m increment and on at the next so that there's only one magnet active at a time could reduce possible issues of having the wiggle/snap attach to the wrong magnet.
It's possible that I've got the wrong mental model for how that would work.
Currently on exhibit at Chicago MCA - https://visit.mcachicago.org/exhibitions/gregory-bae/
https://en.wikipedia.org/wiki/Swiss_railway_clock
I always thought that was a brilliant way to keep mechanical clocks in sync. And the system had been there since the 1970s, so it was well before digital clocks were feasible for a school.
The clocks could even be set for Daylight time centrally, so I assume it had a way to send a signal that triggered a motor to move the hour hand faster.
(Sorry for the deprecated master/slave terminology, it's unfortunately what they used at the time)
In contrast, I'll sometimes now walk into a lobby or reception area of a business where they will have like five different analog clocks on the wall representing NY, Paris, London, etc, (trying to look "international" I guess) and the minute hands are out of sync. Kind of sad in 2023. :-)
When Daylight Savings Time began, I sauntered in about 9am and my coworkers were filtering in too. The morning guy, "Matt" had been there since about 6am and he had changed the clocks. So my other coworker, "Jason" congratulated Matt for setting all the clocks correctly and then he asked him, "so did you do it by setting all four clocks ahead one hour, or did you move them all to the left and then move the Pacific clock to the Eastern position?" Facepalm!
That might be a good job interview question, or it would have been if manual, analog clocks were still a thing.
In Germany (and in neighboring countries), radio-controlled clocks are quite common for this purpose, listening to DCF77 near Frankfurt [1].
Long wave seems to work a bit better than the L-band for indoor propagation, and it probably also helps that the transmit power is three orders of magnitude higher.
I'm not sure if the US has an analogous service – there's WWV and WWVH, but they are shortwave transmitters, and I don't know if wall clock or even wristwatch size receivers are possible for that.
[1] https://en.wikipedia.org/wiki/DCF77
It does. Small WWVB clocks that self set are quite common.
https://en.m.wikipedia.org/wiki/WWVB
It makes sense that a company might make those, if they also make the NTP servers that those clients might be connected to.
Good stuff. I'm planning on buying some for the house.
Thanks for sharing it!
https://www.youtube.com/shorts/ocvBI_vtJwA
https://www.bbc.co.uk/news/science-environment-42237092
[0] https://developer.apple.com/library/archive/documentation/Au...
Edit: here's a video of it running (somehow the seconds hand runs backwards)
https://www.youtube.com/watch?v=HKU-zmNFSxw
> This clock is inspired by a post by ILAMtitan on the 43oh.com forums, which was in turn inspired by Vetinari's clock from the Discworld series of novels. Although the movement of the second hand is quite erratic, it ticks exactly 32 times in 32 seconds, and so keeps quite accurate time overall. The clock started life as a cheap $3 clock from KMart. It's mechanism was kept, but the driver circuit was removed and replaced with an MSP430 microcontroller. ILAMtitan's code is designed to ensure that every 32 seconds it will tick 32 times, but the exact timing of those 32 ticks is variable. There can be as little as 1/4 second between ticks or as much as 24 seconds. Overall though it will keep quite accurate time.
I will have to buy some cheap clocks and start tinkering.
But the way the minute hand stays pointing to a multiple-of-five-minutes until it Boings! to the next increment is disconcerting.
It looks like it's slow all the time. E.g. 10:29 looks like 10:25 which is more than enough error to look wrong.
Maybe I'd learn to read the bent-hand deflection after a while. But I guess I prefer the clock I have that just points to the right minute all the time.
Yeah, it would be a bit more satisfying if the end moves discontinuously (as it does now) but the stem moved continuously.
Still very nice as it is though.