5 ms·

> "...“We faced a number of challenges to improve the performance of MARSIS,” says Carlo Nenna, MARSIS on-board software engineer at Enginium, who is implementi

> "...“We faced a number of challenges to improve the performance of MARSIS,” says Carlo Nenna, MARSIS on-board software engineer at Enginium, who is implementing the upgrade. “Not least because the MARSIS software was originally designed over 20 years ago, using a development environment based on Microsoft Windows 98!”" Not sure if this implies that there was an actual Win98 instance running on-board, or only some custom system/firmware which was developed under Win98... If they will be remotely installing a newer Windows version (which I doubt) on the Mars orbit, that must be added to the Guinness World Records! Ta-da!

karmakaze 4y ago
I take it to mean only the development environment was Windows 98, otherwise they could just have said using Win98. It wouldn't make any sense to put Win98 in MARSIS itself. Sending things into space isn't about how fast a solution can be made, it's about the robustness of what goes out.
- Sharlin 4y ago
  Yes, there's most definitely not anything on the probe that could even run Windows 98. ESA's article is pretty clear that it's just that the software was originally developed on Win98, but of course the usual game of telephone happened and several websites have actually reported that the orbiter itself is running Win98…
ptsneves 4y ago
Proprietary toolchains of that era are a special beast. They are probably talking about proprietary compilers and development tools, which were certified and depended in win98 environments. Perhaps they just ran the toolchain in a win98 qemu instance and developed on a modern computer. That is what I would do and I am an embedded build system guy.
- flatiron 4y ago
  I’ve had problems emulating serial/parallel ports using qemu and usb. I had a legacy system that had some external devices. We just ran it on a hunk of junk on the sneakernet
  - ptsneves 4y ago
    If i recall correctly you can pass through USB ports if needed. Also given that serial/parallel ports are very simple you can often just write a driver or use the super powerful socat to massage things if needed. I once used socat to serve autopilot simulators that spawned on socat's socket accept and piped to actual serials. I could then serve serial over the networks. Really if I was a poet i would write an ode to socat Anyway, peripherals were non the wiser on the nature of the piping. A bit of annoying plumbing but can be done and enables you to have scalable CIs for embedded devices.
  - ncmncm 4y ago
    Why would a DSP assembler need access to a parallel port? Anyway, it probably works fine in Wine, more reliably than on native w98.
    - zinekeller 4y ago
      > Why would a DSP assembler need access to a parallel port? For debugging using twin probes? This isn't far-fetched, for example NASA has MAGGIE (for Curiosity, https://old.reddit.com/r/space/comments/51m2b4/meet_maggie_the_earthbased_copy_of_the_curiosity/ https://old.reddit.com/r/space/comments/51m2b4/meet_maggie_t...) and (Ethernet-based) OPTIMISM (for Perseverance, https://www.jpl.nasa.gov/news/nasa-readies-perseverance-mars-rovers-earthly-twin https://www.jpl.nasa.gov/news/nasa-readies-perseverance-mars...).

karmakaze 4y ago

I take it to mean only the development environment was Windows 98, otherwise they could just have said using Win98. It wouldn't make any sense to put Win98 in MARSIS itself. Sending things into space isn't about how fast a solution can be made, it's about the robustness of what goes out.

Sharlin 4y ago
Yes, there's most definitely not anything on the probe that could even run Windows 98. ESA's article is pretty clear that it's just that the software was originally developed on Win98, but of course the usual game of telephone happened and several websites have actually reported that the orbiter itself is running Win98…

ptsneves 4y ago

Proprietary toolchains of that era are a special beast. They are probably talking about proprietary compilers and development tools, which were certified and depended in win98 environments. Perhaps they just ran the toolchain in a win98 qemu instance and developed on a modern computer. That is what I would do and I am an embedded build system guy.

flatiron 4y ago
I’ve had problems emulating serial/parallel ports using qemu and usb. I had a legacy system that had some external devices. We just ran it on a hunk of junk on the sneakernet
- ptsneves 4y ago
  If i recall correctly you can pass through USB ports if needed. Also given that serial/parallel ports are very simple you can often just write a driver or use the super powerful socat to massage things if needed. I once used socat to serve autopilot simulators that spawned on socat's socket accept and piped to actual serials. I could then serve serial over the networks. Really if I was a poet i would write an ode to socat Anyway, peripherals were non the wiser on the nature of the piping. A bit of annoying plumbing but can be done and enables you to have scalable CIs for embedded devices.
- ncmncm 4y ago
  Why would a DSP assembler need access to a parallel port? Anyway, it probably works fine in Wine, more reliably than on native w98.
  - zinekeller 4y ago
    > Why would a DSP assembler need access to a parallel port? For debugging using twin probes? This isn't far-fetched, for example NASA has MAGGIE (for Curiosity, https://old.reddit.com/r/space/comments/51m2b4/meet_maggie_the_earthbased_copy_of_the_curiosity/ https://old.reddit.com/r/space/comments/51m2b4/meet_maggie_t...) and (Ethernet-based) OPTIMISM (for Perseverance, https://www.jpl.nasa.gov/news/nasa-readies-perseverance-mars-rovers-earthly-twin https://www.jpl.nasa.gov/news/nasa-readies-perseverance-mars...).