v7.1 beta testing begins (Driver Mode, Self Parking, AP restrictions)

[gsxdsm]

If Tesla had a reasonably big firmware team this would be one thing. But I have it on good authority (from an engineer working at Tesla who really would hate me to give more specifics on how they know) that the firmware team is surprisingly small. From my email exchange with that person I understand that the many different versions are a result of the quite un-orthodox firmware release process that Tesla uses.

Anyway, my point is that it appears that the different releases are branches that are being released (after going through QA) without being merged into a linear stream. So if we start at 2.7.56 and go to 2.7.77 and 2.7.85, then .77 could contain fixes that aren't in .85 - but then .106 may have the combination of the changes that were in both earlier releases.

Again, I don't want to get my source in trouble, so I won't get into more detail, but I was quite perplexed by what I was told.
I consider my source reliable - but who knows, this might be an elaborate practical joke (as the emails of course aren't coming from a teslamotors.com account).

From my sources, I second this.

 

[apacheguy]

If Tesla had a reasonably big firmware team this would be one thing. But I have it on good authority (from an engineer working at Tesla who really would hate me to give more specifics on how they know) that the firmware team is surprisingly small.

Dirk - Perhaps those are just the folks working on UI elements? I'd imagine there are others working on BMS, drivetrain firmware, ECUs, etc.

 

[disillusioned]

If Tesla had a reasonably big firmware team this would be one thing. But I have it on good authority (from an engineer working at Tesla who really would hate me to give more specifics on how they know) that the firmware team is surprisingly small. From my email exchange with that person I understand that the many different versions are a result of the quite un-orthodox firmware release process that Tesla uses.

Anyway, my point is that it appears that the different releases are branches that are being released (after going through QA) without being merged into a linear stream. So if we start at 2.7.56 and go to 2.7.77 and 2.7.85, then .77 could contain fixes that aren't in .85 - but then .106 may have the combination of the changes that were in both earlier releases.

Again, I don't want to get my source in trouble, so I won't get into more detail, but I was quite perplexed by what I was told.
I consider my source reliable - but who knows, this might be an elaborate practical joke (as the emails of course aren't coming from a teslamotors.com account).

This still doesn't sound too unorthodox. You ship the components that are closest to ready and sometimes feature branches aren't ready to be merged back to master and miss the boat. Unless I'm misreading your interpretation of things.

Admittedly, it's a bit... surprising to see the feature branches release to different groups of the actual public; if your point is that there is typically just one production branch that non-beta end users are exposed to, and that Tesla breaks that orthodoxy, well, point taken. But that's also how Facebook and Google and a bunch of others do things, so it's a bit of blending new world with old, 5,000 lbs of steel world. Here's hoping they unit test!

 

[dirkhh]

This still doesn't sound too unorthodox. You ship the components that are closest to ready and sometimes feature branches aren't ready to be merged back to master and miss the boat. Unless I'm misreading your interpretation of things.

Admittedly, it's a bit... surprising to see the feature branches release to different groups of the actual public; if your point is that there is typically just one production branch that non-beta end users are exposed to, and that Tesla breaks that orthodoxy, well, point taken. But that's also how Facebook and Google and a bunch of others do things, so it's a bit of blending new world with old, 5,000 lbs of steel world. Here's hoping they unit test!

Well, there a many different commonly used release methodologies. And indeed companies do A/B testing where different features are released to different users and the results are then evaluated.
What to me is most startling is that AFAIK they have "wide" releases that are supposed to go to the installed base (.56 and likely .77 were that), but then they do releases that go out to smaller and what appear to be fairly random groups, but not as A/B test but as "hey, this feature (or this bug fix) cleared testing, let's get it out there" and so some people get one but not the other, or the other, but not the one, and then later the next "wide" release will include BOTH.
That is not something I have seen documented anywhere or encountered elsewhere.

 

[chickensevil]

So I'm confused, why can't they use the Nav to tell what road they are on? The only time I have seen any issues with it is distinguishing between the HOV center lane and the main highway. Otherwise the roads have always matched up just fine for me even with the various construction and such. Because the nav that determines that is based on Google which is kept pretty current.

So interstate and highway... this should be definition at a minimum include the Interstate system (I-5 / I-95 / I-70 / etc), the US Highway system (US 50 / US 66 / US 1), and possible also state highways since most of those also post at 45MPH and above. What it would stop is city streets and other non-highway streets, and from the way it is worded all it is limiting is that you can't go above 45MPH on those non-highway streets. This really doesn't seem like much of a limitation to me. Or am I missing something? Even if I can't use it up and down the busy parts of route 1 (here in VA), that is fine, because the roads kinda shift and change so much that you shouldn't be really using it there in the first place. And TACC would still be available, and that is still quite useful on those roads. What exactly is the issue again?

 

[donv]

The confusion comes down to the phrase "divided highway"-- at least to me. Around here, there are lots of "highways" (including some US highways and some state highways) which don't have a divider-- and may, in some places, be only 2 lanes. This is the issue.

Hopefully, it turns out to be a non-issue...

I think Tesla is, well, trolling us, to some extent. They put these things out there and see what the reaction is, and then they pull them back.

So I'm confused, why can't they use the Nav to tell what road they are on? The only time I have seen any issues with it is distinguishing between the HOV center lane and the main highway. Otherwise the roads have always matched up just fine for me even with the various construction and such. Because the nav that determines that is based on Google which is kept pretty current.

So interstate and highway... this should be definition at a minimum include the Interstate system (I-5 / I-95 / I-70 / etc), the US Highway system (US 50 / US 66 / US 1), and possible also state highways since most of those also post at 45MPH and above. What it would stop is city streets and other non-highway streets, and from the way it is worded all it is limiting is that you can't go above 45MPH on those non-highway streets. This really doesn't seem like much of a limitation to me. Or am I missing something? Even if I can't use it up and down the busy parts of route 1 (here in VA), that is fine, because the roads kinda shift and change so much that you shouldn't be really using it there in the first place. And TACC would still be available, and that is still quite useful on those roads. What exactly is the issue again?

 

[Andyw2100]

I think Tesla is, well, trolling us, to some extent. They put these things out there and see what the reaction is, and then they pull them back.

I really don't think Tesla included the notes about the restrictions just to see what our reaction would be. I think they could have predicted our reaction pretty well. I'm not sure what's going on, but I just don't think that's it.

 

[dirkhh]

I really don't think Tesla included the notes about the restrictions just to see what our reaction would be. I think they could have predicted our reaction pretty well. I'm not sure what's going on, but I just don't think that's it.

I, too, find that questionable.
But I think this would be the first time that they explicitly add something like this to the release notes which then isn't there.
I'm quite puzzled, actually. Could it be that they need a map data update for this to work? Or there is some other secondary "thing" that needs to fall into place?
Yeah, grasping at straws...

 

[green1]

So I'm confused, why can't they use the Nav to tell what road they are on? The only time I have seen any issues with it is distinguishing between the HOV center lane and the main highway. Otherwise the roads have always matched up just fine for me even with the various construction and such. Because the nav that determines that is based on Google which is kept pretty current.

So interstate and highway... this should be definition at a minimum include the Interstate system (I-5 / I-95 / I-70 / etc), the US Highway system (US 50 / US 66 / US 1), and possible also state highways since most of those also post at 45MPH and above. What it would stop is city streets and other non-highway streets, and from the way it is worded all it is limiting is that you can't go above 45MPH on those non-highway streets. This really doesn't seem like much of a limitation to me. Or am I missing something? Even if I can't use it up and down the busy parts of route 1 (here in VA), that is fine, because the roads kinda shift and change so much that you shouldn't be really using it there in the first place. And TACC would still be available, and that is still quite useful on those roads. What exactly is the issue again?

Because the nav data is HORRIBLE, and frequently I drive on roads that it doesn't know exists, or thinks are still small windy roads when they've since been converted to 4 lane divided freeways. I can't imagine allowing that disaster of a system decide when I get to use AP or not.

- - - Updated - - -

But I think this would be the first time that they explicitly add something like this to the release notes which then isn't there.
I'm quite puzzled, actually. Could it be that they need a map data update for this to work? Or there is some other secondary "thing" that needs to fall into place?

My feeling is that it is actually there, but that so few people have really tested it, and not very thoroughly, that we just haven't seen the outcry yet. I'm reserving judgement.

That said, some of the reports I've see that say it's not restricting, were flat out wrong in their testing. One said that they could set AP on a residential road so there must not be a restriction, when the release notes specifically state that the restriction only activates if you set the AP at above 45mph on a residential road. So they didn't even test the right thing there.

I'd be glad to be proven wrong, and if enough testing proves that the restriction does not in fact exist, I'll update, but until then I've reached my highest firmware version.

 

[Cebe]

I note that all the people saying that Tesla should restrict autopilot are conveniently ignoring my question of how to determine where to do so... seems they don't like that inconvenient problem...

Doesn't the car know it's on a divided highway because of the GPS?

 

[AnOutsider]

Doesn't the car know it's on a divided highway because of the GPS?

The map database isn't exactly known for being current though.

 

[Cyclone]

The map database isn't exactly known for being current though.

Maybe it will only "vote" for it being restricted based on the detailed AP-crowd sourced maps?