[HN Gopher] We Diagnosed and Fixed the 2023 Voyager 1 Anomaly fr...
___________________________________________________________________
We Diagnosed and Fixed the 2023 Voyager 1 Anomaly from 15B Miles
Away [video]
Author : noselasd
Score : 157 points
Date : 2025-04-18 22:59 UTC (4 days ago)
(HTM) web link (www.youtube.com)
(TXT) w3m dump (www.youtube.com)
| thadk wrote:
| "Hello world" takes on new dimensions in this context.
| metalman wrote:
| and serious latency
| RamRodification wrote:
| void explore()
| freefaler wrote:
| Pff... and I can debug a stupid bug from 0.00001 miles for the
| 3rd day.
| ordu wrote:
| I'm a little surprised by their approach. I mean, it did work, it
| is cool, and it is the most important thing. Still I can't stop
| thinking that I wouldn't sleep before I wrote an assembler and a
| disassembler. Judging by the presentation they had no assembler
| and disassembler for several months and just lived with that.
|
| asm/disasm can help to find typos in listings, they can help to
| find xrefs or even to do some static analysis to check for
| mistake classes they knew they could make. It wouldn't replace
| any of the manual work they've done, but still it can add some
| confidence on top of it. Maybe they wouldn't end with priors
| 50/50 for the success, but with something like 90/10.
|
| Strange. Do I underestimate the complexity of writing an asm and
| disasm pair?
| october8140 wrote:
| Yes.
| mystified5016 wrote:
| Well, it was a totally bespoke CPU, and we don't have any
| working models on earth.
|
| Writing an assembler for a bespoke CPU is one thing, many of us
| have done it as a toy project, but stakes are a bit different
| here. You'd have to mathematically _prove_ your assembler and
| disassembler are absolutely 100% correct. When your only
| working model is utterly irreplaceable and irrecoverable upon
| error, it probably takes a _lot_ more resources to develop.
| jstanley wrote:
| And if you can't mathematically prove it correct, you're
| better off doing it in your head?
| chuckadams wrote:
| No but the time it would take to build the assembler and
| validate its output would take more time than just writing
| the patch by hand. It's for a craft that isn't going to
| last more than 5 more years tops anyway.
| gherkinnn wrote:
| https://danluu.com/cocktail-ideas/
|
| Yes, I have strong reason you underestimate the complexity
| here.
| bityard wrote:
| He mentioned a few times that writing an assembler was a no-go.
|
| It would have taken much more time than they had available, and
| since an assembler would be a new tool, it would have required
| certification. (So, even more time and paperwork.) Plus, they
| had incomplete docs and there is no working copy or simulator
| of Voyager here on Earth. So any assembler written would by
| definition be incomplete or inaccurate.
| jebarker wrote:
| Puts things into perspective. I often wonder how so many people
| survive without a UI debugger because cmdline debugging seems too
| clunky.
| kabdib wrote:
| henry s f cooper's book _the evening star_ is a great description
| of the Magellan probe (the venus orbiter), and how they were
| debugging what turned out to be OS race conditions on a
| spacecraft millions of miles away
| bityard wrote:
| I think what fascinates me the most about all of this is how
| there are wide gaps in how much design and engineering
| documentation from that time period has survived to present day.
| For a long time, I just assumed that NASA owned and archived
| every design spec, revision, research paper, memo and napkin
| doodle related to their public-facing missions. I learned
| recently that even a lot of the original Gemini and Apollo
| program code (let alone source code) and docs are apparently gone
| forever.
___________________________________________________________________
(page generated 2025-04-22 23:01 UTC)