-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Supercharger Live Status
- Thread starter tbleakne
- Start date
Bighorn
Top Supercharger
Bighorn
Top Supercharger
I was at Barstow yesterday and found the numbers to be 100% accurate and upgraded within 2 minutes of a change.
My Nav is reporting Supercharger availability again. Some of the sites seem more accurate, others still questionable. Showing nobody in Ozona, TX, which is better than the almost full it showed before. Junction likely has nobody there, but shows 2 cars and has for a few hours. Three Rivers, TX still shows almost full but highly unlikely anybody is there.
Things seem to be improving slowly.
Things seem to be improving slowly.
dhcp
Member
Refresh my memory - what company comes out with new versions of software that has no bugs? Even relatively mature products, Microsoft, Google, anyone?
Working in that industry, I can tell you that there is only so much QA and beta testing you can do and then you have to release it to the masses. Thousands of miles may test out well but some problems will only surface after thousands of miles
I understand your frustration, but some bugs only come out in scale. If anything, I've thought that (overall) their software is pretty solid compared to a lot of others in the industry.
I agree with everything you said up to the last paragraph.
Tesla re-introduces bugs that have been previously fixed, this has been shown time and time again*, where they could have caught it with a good QA/QC process. Not only that, they add bugs which again, should have been caught with simple QA/QC (song ends 15 seconds before the end, each time you enter the car)
*as in time (equalizer working, equalizer resetting, working again, resetting again, finally working again) and time again (rear thump created, fixed, brought back, not fixed yet). likely more I can't think of.
dhcp
Member
Ok, I would agree with you Regression testing should be part of their QA process. A few legacy bugs could crop back up but regression testing is often part of a software QA process and usually goes pretty quickly since you normally have the "steps to replicate" the bug.
The software engineers should also be aware if they are mucking around areas that had previous issues and alert the QA folks to focus on such.
WannabeOwner
Well-Known Member
Indeed, but in the time I have owned the car I've seen far too many bugs that should have been caught in QA, if not sooner. Cost of fixing a bug post-release is horrific - more so if you include bad PR - I'm staggered that that, alone, isn't enough incentive to cause an improvement. Add to that the effort of fixing released-bugs also wrecks delivery timescale of other intended new work, which then causes the whole "here-we-go-again" mirth in the owner community when yet-another-buggy-version is released and promised deadlines are missed.
I have not, fingers-crossed, found anything of concern on the safety-side, so why is it frequent on the Infotainment side? If a team with great QA skills exists in the company then leverage that knowledge to improve the weaker team(s).
My expectation would be that regression testing was automated and run frequently - preferably on a daily-build in DEV so that a bug introduced today, which can be caught by automated testing, is fixed tomorrow - before other dependencies exist which are then a nightmare to fix, and mean that deadlines become very unpredictable - fix one thing, break something else. I've only been in that situation once, thankfully decades ago, where we discovered we had no idea when the product would ship, and I never, ever, want to be there again!.
I'm happiest when our QA testing finds nothing
While all software has bugs, Tesla's releases often contain obvious bugs and design flaws that other companies would never allow into their released products.
Tesla tries to overcome the flaws in their QA process by using "start-stop-resume" software distribution, essentially using the first batch of cars receiving new updates as late "beta testers", and when major problems are encountered, Tesla suspends distribution, fixes the bugs, and the starts the process over again.
While Model S and X owners have surprisingly accepted the less than desired software quality and the seemingly random software release distribution, it's not clear Model 3 owners will be as forgiving.
Tesla can and must do much better, especially as the amount of software involved with EAP/FSD increases - when even small problems can become huge when the cars are operating under minimal or no human control.
While Tesla should benefit from improvements in their specifications-design-implementation process, they should also benefit from changes to their external testing program.
First, Tesla should open up their beta testing program, and allow owners to "opt in" or "opt out" of the testing. Among Tesla's customers, there are likely quite a few owners who have considerable experience with software systems and testing, and could be high quality beta testers, helping to find more of the obvious problems before software is released, when it's easier to get the problems fixed.
The other change Tesla should implement is the ability to "roll back" the software to a previous official release. That way, if the released software (whether a beta or official release) has a major problem, the latest release could be backed out - and then reinstalled when the problem is corrected. Rolling back to arbitrary releases can be difficult to test, which is why I'd recommend roll back to a specific release - which is something Tesla should be able to do as part of their standard testing before distributing any beta or official release.
And, owners should be given more control over the software being installed in their cars. Release notes should be provided either via e-mail, My Tesla or in the car, prior to installing the software (today you can only learn what's in the release, after the release has been installed). And, owners should be able to select when new releases are installed - with the ability to request a release installation, and not waiting for Tesla's seemingly random distribution sequence. Owners could also indicate their tolerance for changes, indicating if they want new releases to be installed quickly after availability or if they would prefer to wait a couple of weeks and give Tesla a chance to fix problems before the software is installed.
Yes, all software has bugs - but for cars that average $100K, Tesla should be able to do better...
Tesla tries to overcome the flaws in their QA process by using "start-stop-resume" software distribution, essentially using the first batch of cars receiving new updates as late "beta testers", and when major problems are encountered, Tesla suspends distribution, fixes the bugs, and the starts the process over again.
While Model S and X owners have surprisingly accepted the less than desired software quality and the seemingly random software release distribution, it's not clear Model 3 owners will be as forgiving.
Tesla can and must do much better, especially as the amount of software involved with EAP/FSD increases - when even small problems can become huge when the cars are operating under minimal or no human control.
While Tesla should benefit from improvements in their specifications-design-implementation process, they should also benefit from changes to their external testing program.
First, Tesla should open up their beta testing program, and allow owners to "opt in" or "opt out" of the testing. Among Tesla's customers, there are likely quite a few owners who have considerable experience with software systems and testing, and could be high quality beta testers, helping to find more of the obvious problems before software is released, when it's easier to get the problems fixed.
The other change Tesla should implement is the ability to "roll back" the software to a previous official release. That way, if the released software (whether a beta or official release) has a major problem, the latest release could be backed out - and then reinstalled when the problem is corrected. Rolling back to arbitrary releases can be difficult to test, which is why I'd recommend roll back to a specific release - which is something Tesla should be able to do as part of their standard testing before distributing any beta or official release.
And, owners should be given more control over the software being installed in their cars. Release notes should be provided either via e-mail, My Tesla or in the car, prior to installing the software (today you can only learn what's in the release, after the release has been installed). And, owners should be able to select when new releases are installed - with the ability to request a release installation, and not waiting for Tesla's seemingly random distribution sequence. Owners could also indicate their tolerance for changes, indicating if they want new releases to be installed quickly after availability or if they would prefer to wait a couple of weeks and give Tesla a chance to fix problems before the software is installed.
Yes, all software has bugs - but for cars that average $100K, Tesla should be able to do better...
SR-75
Member
I really miss your API for Supercharger Status. Thanks to you, at least you show everyone what is possible. I hope that our existing Supercharger App will expand their features to include the info that you have shown possible.
cpa
Active Member
First, I am mildly dismayed that bugs in software are accepted and tolerated throughout the industry. If I provided financial statements that did not foot, had omissions, or had mis-matched references, I would not be in business very long and probably have my license in jeopardy. But then, I know zero, zip, and nada about the software industry and its professional and ethical standards.
That said, I believe strongly that a large percentage of testers be computer idiots like me. Using experts, whether professional or hobbyists, to test things does not represent a valid sample of every potential user. And I believe that it is very hard for computer wizards to remove themselves from their bubbles. They do not understand how the rest of us think, react or solve problems.
That said, I believe strongly that a large percentage of testers be computer idiots like me. Using experts, whether professional or hobbyists, to test things does not represent a valid sample of every potential user. And I believe that it is very hard for computer wizards to remove themselves from their bubbles. They do not understand how the rest of us think, react or solve problems.
Columbus, TX is not reporting properly. When I arrived this evening the nav said 4 of 6 stalls available, but there was no one there. After I plugged in it changed to 3 of 6 stalls available, so it recognized I was there, but apparently it thinks two of the stalls are occupied when they are not. Also I didn't get a charging almost complete notification on my iPhone-- first time that's happened.
WannabeOwner
Well-Known Member
I haven't found that that works well. The strike-rate of problems found is way too low. We had a whole tele-sales call centre (70-ish people, training completed but basically "ordinary people", pending the "busy time" in the run up to Christmas), test a new eCommerce website. They, collectively, found half a dozen trivial issues whereas the company boss and I each had lists several pages long. He was quite disappointed that the call centre hadn't found more as I recall it!!
My expectation (I work in the software industry) is that the folk in the Testing department are not the same [breed] as the DEV folk. Its a particular skill - devising a test, and running it / tweaking it, etc. and having a nose for what may go wrong, as well as being trained in the psychology of how users behave and what they are likely to [wrongly] do. The Software industry (either at the design phase, using some sort of mockups, or later on in the QA phase) also makes use of Focus Panels - watching users actually using the software; where their eyes go, what they click on, interviewing them "Why" they tried pressing that (unexpected!!) button, and so on. Those techniques are good at finding issues IME.
I have found that clients' responsibility for web-based projects we have been involved in has moved from IT (back in 2000) to Marketing. IME Marketing haven't got a clue what makes a good, usable, website, their primary focus is on pixel-perfect graphics, slick appearance, and differentiating themselves from their competitors. So gone, from the old days, are objectives like "Can you put an item in your shopping basket and checkout" and "Is it easy and intuitive to use" ("differentiating from competitors" usually means "It doesn't work like anything else anyone is already familiar with"). I'm an old dinosaur, but i'd like to bring back the eCommerce objective which Clients had in 2000 which, back then, was "build it looking just like Amazon so everyone will be familiar with it"
As a sysadmin turned computer scientist turned software engineer, I enjoy discussions of Tesla software quality, bugs, features, etc. as much as anyone else, but they're really off-topic for a thread on the Supercharger live status feature on the Tesla CID.
Thanks,
Bruce.
(NOT a moderator but I could maybe play one on TV)
Thanks,
Bruce.
(NOT a moderator but I could maybe play one on TV)
+1
Probably mentioned, but if would be nice if it was not required to freeze the screen to monitor the status of a SC. Not to to mention the distraction while driving of trying to "click" the icon to see the status. It would be great if it just showed the status bar the way it now just shows the SC icon.
rolson1011
2016 P90DL, Model 3 on reservation
Similar threads
