Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Just got new update: 2017.42 a88c8d5

This site may earn commission on affiliate links.
When you get your car please let us know what are in the New Release Notes.
I'll bet it has the AEB/Mirrors release notes and nothing more. :D

And we're getting 17.42.1 because 17.42 broke a lot of AP2.5 cars.
We got 17.42 because there was a firmware issue with some AP.2.5 cars. (Like mine.)
We got 17.41 because there was a Homelink bug....

All of this was besides the features that were in these releases (AEB, Homelink, Camper, Mirror Folding)....

Either way, Tesla can't make people happy, releases or no releases. :D
 
I haven't received an update in months. I am still stuck on 2017.28. Anyone else on the same boat ?
I've been on 2017.34 for a very long time.

Looking at firmware.teslafi.com - I think your system has a stuck update and will require a call or the service center to catch it back up.

I'm going to stay on 34 for a while :) So you may want to wait for the current mess to be resolved before asking. You might be in the best possible situation!
 
I'll bet it has the AEB/Mirrors release notes and nothing more. :D

And we're getting 17.42.1 because 17.42 broke a lot of AP2.5 cars.
We got 17.42 because there was a firmware issue with some AP.2.5 cars. (Like mine.)
We got 17.41 because there was a Homelink bug....
Ah, the dilemma of the software release... When to rush out a release when the previous release you rushed out had even more new problems?... Software companies that go down this path invariably find that they have to slow down this "rush it out" mentality and go back to basics, which is better development methodology and more thorough alpha and beta testing. At this point, the unwitting beta testers (basically US!) are doing a better job than their internal alpha testers, since these new bugs are discovered pretty quickly by us... But it shouldn't be this way. Software developers should be held to the fire and be held to a higher standard before being allowed to release it, none of this toss it over the wall and let the testers find it.

In my previous life in the carrier telecoms equipment space, I saw this clear shift from reliable fail-safe software to the 'just reboot it' mentality. Early telecoms equipment never had reset buttons, because they were designed to be powered on and keep on running without ever being reset or rebooted. As more off-the-self computing equipment made their way into these equipment, reset buttons started to appear and engineers thought nothing about just resetting the unit. Of course, Microsoft needs to accept a lot of the blame for this, as that was always their answer to something not working correctly. In the case of software for cars, that same mentality should apply. Software should never lock up and require a reboot. And because of that, the testing requirements should be more stringent and thorough. Yes, it does lengthen the release times, but I would argue that is better than having to fight fires all the time with hot fix releases.
 
Ah, the dilemma of the software release... When to rush out a release when the previous release you rushed out had even more new problems?... Software companies that go down this path invariably find that they have to slow down this "rush it out" mentality and go back to basics, which is better development methodology and more thorough alpha and beta testing. At this point, the unwitting beta testers (basically US!) are doing a better job than their internal alpha testers, since these new bugs are discovered pretty quickly by us... But it shouldn't be this way. Software developers should be held to the fire and be held to a higher standard before being allowed to release it, none of this toss it over the wall and let the testers find it.

In my previous life in the carrier telecoms equipment space, I saw this clear shift from reliable fail-safe software to the 'just reboot it' mentality. Early telecoms equipment never had reset buttons, because they were designed to be powered on and keep on running without ever being reset or rebooted. As more off-the-self computing equipment made their way into these equipment, reset buttons started to appear and engineers thought nothing about just resetting the unit. Of course, Microsoft needs to accept a lot of the blame for this, as that was always their answer to something not working correctly. In the case of software for cars, that same mentality should apply. Software should never lock up and require a reboot. And because of that, the testing requirements should be more stringent and thorough. Yes, it does lengthen the release times, but I would argue that is better than having to fight fires all the time with hot fix releases.

Exactly! Could not agree more.
 
Ah, the dilemma of the software release... When to rush out a release when the previous release you rushed out had even more new problems?... Software companies that go down this path invariably find that they have to slow down this "rush it out" mentality and go back to basics, which is better development methodology and more thorough alpha and beta testing. At this point, the unwitting beta testers (basically US!) are doing a better job than their internal alpha testers, since these new bugs are discovered pretty quickly by us... But it shouldn't be this way. Software developers should be held to the fire and be held to a higher standard before being allowed to release it, none of this toss it over the wall and let the testers find it.

In my previous life in the carrier telecoms equipment space, I saw this clear shift from reliable fail-safe software to the 'just reboot it' mentality. Early telecoms equipment never had reset buttons, because they were designed to be powered on and keep on running without ever being reset or rebooted. As more off-the-self computing equipment made their way into these equipment, reset buttons started to appear and engineers thought nothing about just resetting the unit. Of course, Microsoft needs to accept a lot of the blame for this, as that was always their answer to something not working correctly. In the case of software for cars, that same mentality should apply. Software should never lock up and require a reboot. And because of that, the testing requirements should be more stringent and thorough. Yes, it does lengthen the release times, but I would argue that is better than having to fight fires all the time with hot fix releases.
Amen.. there was a post somewhere here about what Tesla should do about testing software and perhaps allowing a 'roll back one step' option. In the cases of this weekend, they do need to get the fix out quickly. But who rolls out new code on a Friday night?! Roadside was slammed Saturday, and it was a good thing I knew/figured out how to restart my car, or I might have been sitting there until late that night. I'm sure some others weren't so lucky.
 
Ah, the dilemma of the software release... When to rush out a release when the previous release you rushed out had even more new problems?... Software companies that go down this path invariably find that they have to slow down this "rush it out" mentality and go back to basics, which is better development methodology and more thorough alpha and beta testing. At this point, the unwitting beta testers (basically US!) are doing a better job than their internal alpha testers, since these new bugs are discovered pretty quickly by us... But it shouldn't be this way. Software developers should be held to the fire and be held to a higher standard before being allowed to release it, none of this toss it over the wall and let the testers find it.

In my previous life in the carrier telecoms equipment space, I saw this clear shift from reliable fail-safe software to the 'just reboot it' mentality. Early telecoms equipment never had reset buttons, because they were designed to be powered on and keep on running without ever being reset or rebooted. As more off-the-self computing equipment made their way into these equipment, reset buttons started to appear and engineers thought nothing about just resetting the unit. Of course, Microsoft needs to accept a lot of the blame for this, as that was always their answer to something not working correctly. In the case of software for cars, that same mentality should apply. Software should never lock up and require a reboot. And because of that, the testing requirements should be more stringent and thorough. Yes, it does lengthen the release times, but I would argue that is better than having to fight fires all the time with hot fix releases.
I actually like what Microsoft is currently doing with Windows Insider (betas). I am actually not part of it now but below explains it. Some of us enjoy doing beta testing and others not so much. By allowing customers to choose their poison that would help to set expectations.

Co-create an even better Windows | Windows Insider Program

Preview with confidence
No beta software is perfect. That’s why we offer Insiders different versions of Insider Preview Builds. We call them Rings. The Slow Ring is for most Insiders who simply want to explore new Preview features with minimal risk to their devices. The Fast Ring is for advanced Insiders who want to explore new features even sooner — and don’t mind dealing with the bugs and other issues that come with these builds.

Insider Previews updates, rings and support | Windows Insider Program
 
I agree totally, but usually *there* you have another PC or a VM (virtual machine, for those not in IT) to run the beta/Insider software on, since you need one stable platform for everyday tasks. Hard to have two Teslas. Not that there's anything wrong with that....

And with the level of communications that Tesla has seen to be capable of, Microsoft looks like a shining beacon.... sigh....
 
What's strange is that the hardware inside a Tesla is a known quantity - i.e Tesla put it all there. Unlike Microsoft, who have to deal with hardware configurations they had no hand in building (and maybe aren't even aware of). I would like to think that Tesla has a garage somewhere with at least one Tesla of every configuration and hardware revision possible, with which they test all the firmware on simultaneously before release. However, I doubt it.

On a side note; don't rush to update past 17.36 - I'm surprised to see regression still, even at this stage... but 17.40 does scary things for me on roads that 17.36 was completely fine and reliable on.
 
Well, a LOT of us are on 17.40 or 41 or 42 and need 42.1 or whatever is the full rollout release to fix the possibility of the Steering and SAS issues.

Now I'm thinking 17.42.1 is another testing release. Only 3 samples on TeslaFi after 12 hours, and that's exactly like 17.41 and 17.41.6 last week....
 
I would like to think that Tesla has a garage somewhere with at least one Tesla of every configuration and hardware revision possible, with which they test all the firmware on simultaneously before release. However, I doubt it.
Former QA team member here on TMC revealed that they have a perfect simulator of the car so they just plug updated code modules into that simulator to make sure everything works fine before committing (aka unit testing) and the simulator tests all possible car configurations and such. Therefore apparently they think integration testing on real cars is unnecessary (and takes too long time anyway).
 
Former QA team member here on TMC revealed that they have a perfect simulator of the car so they just plug updated code modules into that simulator to make sure everything works fine before committing (aka unit testing) and the simulator tests all possible car configurations and such. Therefore apparently they think integration testing on real cars is unnecessary (and takes too long time anyway).
As an IT person for too many decades in too many areas of IT, LOL, I say. I guess something went awry when the software is looking for hardware that doesn't exist any longer (a direct quote from the Diagnostics team).
 
  • Funny
Reactions: lunitiks
Interesting thread... Seriously, I wonder how they deal with the three versions (I assume: 1.0, 2.0, 2.5) of hardware out there now.... Because, obviously, something isn't compensating for it.
Oh someone is testing the crap out of it

1wszn6.gif
 
Interesting thread... Seriously, I wonder how they deal with the three versions (I assume: 1.0, 2.0, 2.5) of hardware out there now.... Because, obviously, something isn't compensating for it.
there are many more versions obviously.

Multiple components for internal versions. Different steering wheels, seats, various sensors, chargers, ...

3 versions of the AP subsystems (the hw2 and hw2.5 being their own computer is a whole different can of worms).

Apparently the answer is in the simulation - just flip a switch in the simulator and test the new hw config in a perfect manner! ;)
And don't you worry if something slips through, they'll add another testcase into the simulator for it!
 
  • Funny
Reactions: lunitiks
Former QA team member here on TMC revealed that they have a perfect simulator of the car so they just plug updated code modules into that simulator to make sure everything works fine before committing (aka unit testing) and the simulator tests all possible car configurations and such. Therefore apparently they think integration testing on real cars is unnecessary (and takes too long time anyway).
Testing via a simulator might be fine for a basic sanity or go/no go test, but is no substitute for full integration testing. After all, a simulator is just software as well and could have bugs too....

If Tesla thinks testing in real cars is unnecessary and takes too long, will they apply that same thinking to their EAP and FSD testing?