Humor Random Comments Thread

[dbeachy1]

https://twitter.com/x/status/1354457380935901190

 

[BruceJohnJennerLawso]

is it possible to implement what a time-of-flight chip does in software if you have a triple axis accelerometer?

If you know the acceleration an object undergoes and it starts stationary relative to your reference frame, wouldnt you be able to use that information to reconstruct its position relative to a starting point?

 

[jedidia]

Anything you can do with specialised hardware, you can also do with software. The main issue why you wouldn't is usually performance and reliability. I can't really speak for the use case you propose because I don't know anything about it, but if there's a lot of sensor data to be dealt with and integrity is a top priority, having a chip only dedicated to reading them and processing their data seems like the smart thing to do.

 

[Urwumpe]

is it possible to implement what a time-of-flight chip does in software if you have a triple axis accelerometer?

If you know the acceleration an object undergoes and it starts stationary relative to your reference frame, wouldnt you be able to use that information to reconstruct its position relative to a starting point?

Are you sure, you are maybe not mixing two different things up? I just looked at what a time-of-flight chip can actually be, and found something that is used in laboratories for calibrating experiment results.

The usual way to measure the time of flight of a rocket is to merge multiple sensor data streams in a computer to get as precise information about launch and landing as possible, for example a switch or pull-out wire for detecting lift-off more reliably, while using accelerometers to detect ground contact (unreliable - e.g. when parachutes come into play).

 

[BruceJohnJennerLawso]

Are you sure, you are maybe not mixing two different things up? I just looked at what a time-of-flight chip can actually be, and found something that is used in laboratories for calibrating experiment results.

The usual way to measure the time of flight of a rocket is to merge multiple sensor data streams in a computer to get as precise information about launch and landing as possible, for example a switch or pull-out wire for detecting lift-off more reliably, while using accelerometers to detect ground contact (unreliable - e.g. when parachutes come into play).

Possibly

Im thinking of something like this:

VL53L1X Time of Flight (ToF) Sensor Breakout

VL53L1X Time of Flight (ToF) Sensor Breakout

www.pi-shop.ch

more in relation to setting up a robot to have some sense of its position over time, while say mapping out the boundary walls of a room with a separate rangefinder sensor.

Im not sure exactly how that sensor does what it does, but I think what I want (running estimate of position relative to where we started from) could easily be done by sampling rotation, acceleration and integrating that by time to get velocity, then integrating again to get position. It should all work provided the robot doesnt start with an initial velocity (sliding down a slope for example)

 

[BruceJohnJennerLawso]

Dled Orbiter 2016 for the first time since 2013?

Ran the Checklists/DG-ISS scenario just like I did way back in 2011, got to orbit and couldnt figure out why the hell my orbit kept degrading, even at 400km

Took me almost two orbits to realize I accidentally hit the hover engines key with them running at ~5% the entire time ?

With all of the high res terrain & texture packs my Orbiter install is almost 70Gb, eyewatering

Lord knows I begged martins back in the day for terrain in Orbiter, so cool to see it happen

 

[Urwumpe]

Dled Orbiter 2016 for the first time since 2013?

Ran the Checklists/DG-ISS scenario just like I did way back in 2011, got to orbit and couldnt figure out why the hell my orbit kept degrading, even at 400km

Took me almost two orbits to realize I accidentally hit the hover engines key with them running at ~5% the entire time ?

With all of the high res terrain & texture packs my Orbiter install is almost 70Gb, eyewatering

Lord knows I begged martins back in the day for terrain in Orbiter, so cool to see it happen

This terrain really distracts from the important tasks, right?

 

[BruceJohnJennerLawso]

This terrain really distracts from the important tasks, right?

I havent really seen it yet. Im still stuck in LEO trying to figure out TransX again, Id hoped it would all come back to me again but it looks like Ill have to read the manual.

My new rig built post-school is actually gaming quality now which makes the experience so much smoother. I remember trying to squeeze every last ounce of performance out of my crappy Samsung laptop back in 2010, going into the task manager and setting orbiters process priority to high as if that would make any difference.

It sucks that UMMU/UCGO support seems to have lapsed, I had the idea way back then of how cool it would be to make a spacewalk simulator in what VR is now, but it sounds like VR in Orbiter may not happen for a long time, if at all

 

[BruceJohnJennerLawso]

Did a free-return slingshot around the moon, aerobraked over teh Indian ocean west of Australia and tried to make for Woomera on maybe 20% fuel in the DG. Got close, but it was night and I guessed that the runway would be east-west and it turned out to be North-South

Not to mention the damn Woomera runway has no beacons for some reason

 

[dbeachy1]

https://twitter.com/x/status/1354983091459264513

 

[Matias Saibene]

Hirundinidae bias:

partial:

severe:

 

[dbeachy1]

https://twitter.com/x/status/1358436404863320064

 

[n72.75]

I can't say I've ever seen that before.

That was darn adorable.

 

[jedidia]

Had to waste a whole saturday chasing a critical bug because someone with a university degree seemed to think it's no problem to have the frontend rely on an undocumented naming convention that relied on all data having the prefix "test" in the name that stretched through several libraries and was hardcoded in string literal arguments. Like, would it have to be so difficult to at least make a properly documented static const if it was this important? Because of course I "fixed" that perceived little oversight in one library when looking for another issue, only to have all production feeds of a certain type just stop delivering data.

Like, I didn't even go to bloody highschool, but it's obvious enough to me that this is insane. Do they teach people anything actually useful at these establishements??

 

[Urwumpe]

Had to waste a whole saturday chasing a critical bug because someone with a university degree seemed to think it's no problem to have the frontend rely on an undocumented naming convention that relied on all data having the prefix "test" in the name that stretched through several libraries and was hardcoded in string literal arguments. Like, would it have to be so difficult to at least make a properly documented static const if it was this important? Because of course I "fixed" that perceived little oversight in one library when looking for another issue, only to have all production feeds of a certain type just stop delivering data.

Like, I didn't even go to bloody highschool, but it's obvious enough to me that this is insane. Do they teach people anything actually useful at these establishements??

If you have a university degree, you can program any nonsense you like, as long as it looks cool. And then you read "You are not supposed to understand this" in a comment ahead of 100 lines of code that do something magical with your source files.

 

[n72.75]

Not sure it's just the younger generation. I have plenty experience in my professional life of people far younger and older than me doing stuff like this.

 

[BruceJohnJennerLawso]

Had to waste a whole saturday chasing a critical bug because someone with a university degree seemed to think it's no problem to have the frontend rely on an undocumented naming convention that relied on all data having the prefix "test" in the name that stretched through several libraries and was hardcoded in string literal arguments. Like, would it have to be so difficult to at least make a properly documented static const if it was this important? Because of course I "fixed" that perceived little oversight in one library when looking for another issue, only to have all production feeds of a certain type just stop delivering data.

Like, I didn't even go to bloody highschool, but it's obvious enough to me that this is insane. Do they teach people anything actually useful at these establishements??

I took an intro CS course or two, they seemed to spend a lot of time obsessing on functional programming as a paradigm beyond the point of reason to me.

It did stress the importance of unit tests, documentation, type safety, etc. tho

 

[jedidia]

I took an intro CS course or two, they seemed to spend a lot of time obsessing on functional programming as a paradigm beyond the point of reason to me.

In my humble opinion, there is no point beyond which obsessing over a functional paradigm can reasonably be stressed, because people will then go to write in languages not designed for it and immediately throw out all the concepts that make code so much more safe if you didn't hammer the point home to the max... It's not so much about applying everything dogmatically, as it is about applying the things you can reasonably apply, but for that you have to be very familiar with the concepts as a whole. I never want to write backend code in a purely functional language again, but the 2 years I was forced to write in clojure were the most valuable coding lessons in my entire life.

It did stress the importance of unit tests, documentation, type safety, etc. tho

Sounds like quite a different kind of university than our guy went to. Could I have somebody from that university, please? Because that all sounds very appropriate... (though one can overdo it on the documentation front IMHO. I tell my developers that they'd rather invest time in making their code readable rather than into documentation to make sense of an incomprehensible mess...)

 

[Urwumpe]

Sounds like quite a different kind of university than our guy went to. Could I have somebody from that university, please? Because that all sounds very appropriate... (though one can overdo it on the documentation front IMHO. I tell my developers that they'd rather invest time in making their code readable rather than into documentation to make sense of an incomprehensible mess...)

I agree there mostly. Of course, if you do large business software and complex distributed systems, you have a different idea of what "less documentation" really means.

But if I need to read 5 pages of documentation to understand what the implementation of a business rule does, something is going the wrong way. The principle of the least surprise is a very important one. All I really need as documentation are "cooking recipes". How to setup the build environment on a new machine? How to get started for programming? How do deploy on the test system, who to call for deploying to production? What happens if the new version is rejected from production how do we go on? When I translate a business rule - what is the domain language? What is the domain model that we are working with?

Any kind of documentation you write, is automatically as important as the source code. If your documentation is not that important, why did you write it in first place? Same applies to tests. If you carry with you 2000 unit tests, that all fail for years, why didn't you delete them?

The best tests are tests, that you really care for.

 

[jedidia]

Wife: "Son, I cleaned your bicycle, pumped the tires and oiled the chain!"
Son, excited: "Did you mount rocket thrusters?"
Wife: "Uh... no?"
Son: "Awwww..."

?