thing that irk me....

hank · 2019-02-12T03:22:49.000Z

I spent a few years doing QC checks on software after the programmers reported bugs fixed.
They were never happy when I said “well, you are the only one who made any change and the bug disappeared, but this list of new problems popped up …”

I got “but the bug is gone, right?” pushback.

And I replied “Your bugfix change breaks a bunch of new things; it’s unsatisfactory.”

And once they shipped the product to the client, it wasn’t in good hands. The client had a manager who wanted to be given the correct results for every acceptance test that would be run, so he could write them down and show them to his boss as having been found acceptable.

The Volkswagen diesel testing scandal was no surprise.

And I’m living on a planet among people who build fission power plants.

Scary.

SIGShooter · 2019-02-12T04:42:01.000Z

I spent 35 years writing financial software until I retired and I can assure you that most programmers are terrible at vetting their programs for bugs. Typically they just check for what they’re coding for and maybe what the users usually do rather than all possible scenarios (which actually is hard to do in some cases).

Naturally I was the exception with perfect and bug-free code

Of course the Volkswagen scandal wasn’t bad coding or debugging, rather it was intentional fraud.

Lightbringer · 2019-02-12T05:02:28.000Z

SIGShooter:

I spent 35 years writing financial software until I retired and I can assure you that most programmers are terrible at vetting their programs for bugs. Typically they just check for what they’re coding for and maybe what the users usually do rather than all possible scenarios (which actually is hard to do in some cases).

Wellp, if everything didn’t have to be done yesterday (even if only assigned today), there’d be more time to debug code and make sure it would be quite stable.

I did exactly that, regardless of when it was “needed”, and my boss snarkily called it “the G factor”. Ie, if it could take him 1 day to do something, it’d take me 3 days instead, at least according to him. Then again, he had no clue what built-in functions like memset() were, so he had his own function to clear string space (even if not even needed). :person_facepalming:

What I do now is ask, preemptively, “Do you want it done fast or do you want it done right?”. Answering “both” is unacceptable. If minutes count, and you need it now now now, I’ll do it fast, with emphasis on getting it done as fast as possible, with zero error-checking or sanity-checking. If you want it done right, it’ll take more time.

I’ll actually try to break it, to imagine the impossible happening, to check edge-conditions, etc. And even then, sometimes you just miss something.

SIGShooter:

Of course the Volkswagen scandal wasn’t bad coding or debugging, rather it was intentional fraud.

Gotta admit, though, it was quite clever!

First, they designed to the test, based on the fixed and published driving cycle of the test. Then, check to see if the car’s likely on a treadmill or actually on the road.

James Kirk would be proud!

teacher · 2019-02-12T05:07:13.000Z

Lightbringer:

…… ……

So he does a 30sec fix to unjam it and get it spinning again. Happy Day, right? Nope. Got chewed out for not sticking to His Job, and pretty much blackballed from that crew.

:person_facepalming:

And that…. in many cases, is one of the biggest problems with unions. Common sense has left the building……

greenlight · 2019-02-12T07:39:52.000Z

My workplace UI was updated and now it takes me more time to get my work done.

Zulumoose · 2019-02-12T08:27:41.000Z

My workplace UI was updated and now it takes me more time to get my work done.

Just had a first training on our new workplace software, and it looks like everything is going to take several times as long to get done.
The focus, as always, seems to be on putting processes in place for their own sake, rather than for making things simpler and easier to achieve.

Henk4U2 · 2019-02-12T08:49:26.000Z

A thing about firmware and software. This was way back, so I had to dig in the depth of my little grey cells.
A long, long time ago people invented the PC, later followed by the PC-XT and the PC-AT.

These had quite straight forward mother boards, without any integrated IO, audio of video.
The standard video-cards were not very impressive. If you were into playing games, you needed a good one.
So several companies were founded, focussing on making good, better, or best, video-cards.
And pretty soon some benchmark software was introduced to test these video-cards.

Then a certain brand emerged, with a reasonable price and a staggering performance on “the” benchmark.
In real life the (games, spreadsheet, text) the performance of that video-card was rather modest.
Later they found out that video-card was primarily designed to get the best scores at “the” benchmark!
This was in the eighties. The first IT-millions were made using benchmarks and PC-magazines! Sans internet.

kiriba-ru · 2019-02-12T09:19:38.000Z

SIGShooter:

Of course the Volkswagen scandal wasn’t bad coding or debugging, rather it was intentional fraud.

I would choose the word “cheating”.
This is just one from thousands examples, which shows how useless are all this fixed tests and rules.
Some EU cars had asymmetrical frame between bamper and longitudinal beams. Two spacers connected them in triangle with front angle placed in the point that devided car width in 3/5 and 2/5. It worked well for years (EU safety tests require one front crash-test with 40% cross-section between car width and crash block, this triangle helped to distribute force between two longitudinal beams perfectly), but suddenly US safety departament involved new tests with 25% overlap. And something went wrong:
https://www.iihs.org/iihs/sr/statusreport/article/47/6/1
This problem probably claimed many lives. Much more than VW diesel cheat. But it is harder to explain it with politics, so nobody is interested in such discission on tv prime time etc.

teacher · 2019-02-12T10:00:11.000Z

/\. Well said kiriba-ru… .

RobertB · 2019-02-12T11:26:10.000Z

Yourrid:

Flying older, worn out jets for a living has gotten me used to seeing the phrase “could not duplicate” on things I write-up as being inoperable, yet are very difficult to test on the ground.

I was flying into Detroit a few years ago in one hell of a snow storm. The cabin is pressurized by the same hot air (which comes off the engines) that also heats our wings and keeps us from turning into a giant flying ice cube. The engines weren’t producing enough air to do both at the same time. So we were either going to depressurize the cabin and potentially suffocate our passengers, or we were going to turn into a flying brick. Coming in on approach, the computer to keep our tail straight malfunctions (so its swinging all over the place), the autopilot turns itself off, my attitude indicator locks itself in 30 degree right bank (so I don’t know if we are turning or not), and I’m trying to keep the wings from icing over while praying we see the runway with enough time to pull up!

Wrote everything up for maintenance to fix. 2 days later it’s back in service with a couple “could not duplicate” pencil whips :rage:

None of this stuff was visible on the flight data recorder?

wle · 2019-02-12T12:42:39.000Z

yeah!

xevious · 2019-02-12T13:07:11.000Z

Yourrid:

Flying older, worn out jets for a living has gotten me used to seeing the phrase “could not duplicate” on things I write-up as being inoperable, yet are very difficult to test on the ground.

I was flying into Detroit a few years ago in one hell of a snow storm. The cabin is pressurized by the same hot air (which comes off the engines) that also heats our wings and keeps us from turning into a giant flying ice cube. The engines weren’t producing enough air to do both at the same time. So we were either going to depressurize the cabin and potentially suffocate our passengers, or we were going to turn into a flying brick. Coming in on approach, the computer to keep our tail straight malfunctions (so its swinging all over the place), the autopilot turns itself off, my attitude indicator locks itself in 30 degree right bank (so I don’t know if we are turning or not), and I’m trying to keep the wings from icing over while praying we see the runway with enough time to pull up!

Wrote everything up for maintenance to fix. 2 days later it’s back in service with a couple “could not duplicate” pencil whips :rage:

Doesn’t the FAA have some kind of “silent complaint” mechanism in place so people can report on issues that are failing to be addressed that are critical to aircraft safety? What you say scares the hell out of me… just to think of the passengers that might be on such a flight, let alone the risk to your own life. Far better to ruffle some irresponsible feathers than to attempt survival of a crash.

xevious · 2019-02-12T13:17:41.000Z

kiriba-ru:

SIGShooter:

Of course the Volkswagen scandal wasn’t bad coding or debugging, rather it was intentional fraud.

I would choose the word “cheating”.
This is just one from thousands examples, which shows how useless are all this fixed tests and rules.
Some EU cars had asymmetrical frame between bamper and longitudinal beams. Two spacers connected them in triangle with front angle placed in the point that devided car width in 3/5 and 2/5. It worked well for years (EU safety tests require one front crash-test with 40% cross-section between car width and crash block, this triangle helped to distribute force between two longitudinal beams perfectly), but suddenly US safety departament involved new tests with 25% overlap. And something went wrong:
https://www.iihs.org/iihs/sr/statusreport/article/47/6/1
This problem probably claimed many lives. Much more than VW diesel cheat. But it is harder to explain it with politics, so nobody is interested in such discission on tv prime time etc.

VW became the poster-child for a kind of practice that has been going on… “soft” corruption that escapes prosecution. I’m continually amazed at the relatively small fines that are levied against businesses when caught violating regulations. In the end, they still end up better off than had they fully complied with the law… so, they will attempt to violate regulations again, just be more clever about it.

We live in an illusion of civility. At the core of humanity is still an ugly festering cancer of hostility and treachery. People who will throw others to the wolves because they refuse to respect their fellow human being, even if they’re of a lesser status, let alone destitute/homeless. Discrimination is rampant… not just on race and gender, but on economic status and other petty social factors. It’s hard to be optimistic in this age of raging arrogant degenerative populism.

xevious · 2019-02-12T13:26:11.000Z

hank:

I spent a few years doing QC checks on software after the programmers reported bugs fixed.
They were never happy when I said “well, you are the only one who made any change and the bug disappeared, but this list of new problems popped up …”

I got “but the bug is gone, right?” pushback.

And I replied “Your bugfix change breaks a bunch of new things; it’s unsatisfactory.”

And once they shipped the product to the client, it wasn’t in good hands. The client had a manager who wanted to be given the correct results for every acceptance test that would be run, so he could write them down and show them to his boss as having been found acceptable.

The Volkswagen diesel testing scandal was no surprise.

And I’m living on a planet among people who build fission power plants.

Scary.

I used to be an SQA engineer and then an SQA manager. I had a knack for breaking software. I know exactly what you’re talking about, as I’ve experienced it first hand. There were some developers who were pretty good at QC’ing their own code, but others were terrible. As long as it compiled that passed a quick simplistic test, they’d throw it “over the wall” for the QA folks to test it. At times it got so bad, I had QA engineers sit with the developer and test it right then and there. That worked, but it was a drain on resources. Our schedules slipped, but at least we had better code going out.

The only way out of this mess will be when AI development studios are created that make simplistic code writing obsolete (that’ll all be generated based on models and business rules). Developers will \be required only for the more complex kinds of logic requirements that AI can’t easily handle and held to a higher standard.

slmjim · 2019-02-12T13:55:28.000Z

Lightbringer:

… What I do now is ask, preemptively, “Do you want it done fast or do you want it done right?”. Answering “both” is unacceptable. …

A variation of the three-legged stool of realistic acquisition:

1) It can be done fast.

2) It can be done good.

3) It can be done cheap.

One may only choose two. The third will automatically be unavailable, I.e. It can be done good and cheap, but it won't be fast.

slmjim

junkyardbiker · 2019-02-12T13:56:25.000Z

I am a Machinist, In a small shop of a larger complex run by non-machinist folks who love to come in and tell me how to make parts, or that I should be using a different tooling setup, or that Im not allowed to do the processes on the job sheet THEY wrote. Complaining about how when they ask me to remove a few thousanths of material from a surface it also removes the cadium coating on top of it and they have to get it re-coated. I had someone report us because we “make to much swarf” when we machine parts. Called it wastefull usage of material. Well, when you you want a 3” bore cut into a block and the entire stock that you ordered is oversized, we have to remove all the extra stock for the part to come out to spec? Complaints of hazmat usage (machine coolant and tapping fluid) and our shop rags being contaminated with it (you want clean parts out at the end right?) Then the building managers come in and take away our beautiful, perfect, meticulously maintained Monarch lathe and give us a used, beat up, Tiawanese lathe instead…… Oh, and then complained about the money spending because none of our toolposts and tooling worked with the “new” lathe and we asked for tooling. And the scrap bin is off limits but gets thrown in the dumpster.

Lightbringer · 2019-02-12T18:14:49.000Z

slmjim:

A variation of the three-legged stool of realistic acquisition:

“Cheap, on-time, works. Pick any two.”

hank · 2019-02-12T19:49:15.000Z

Nearly got hit this morning by a guy talking (to himself? wireless cell phone?) as he blew through a stop sign from a cross street onto the main street we were driving on, making a left turn across our path. He cussed us out for honking the horn as we screeched to a halt and let him have both sides of the street. Ungrateful!

SIGShooter · 2019-02-12T20:14:19.000Z

hank:

Nearly got hit this morning by a guy talking (to himself? wireless cell phone?) as he blew through a stop sign from a cross street onto the main street we were driving on, making a left turn across our path. He cussed us out for honking the horn as we screeched to a halt and let him have both sides of the street. Ungrateful!

Lucky you weren’t hit. Many years ago I was t-boned by a woman who ran a stop sign like you and was gabbing away on a cell phone. Had my 3 kids in the car and fortunately I saw her out of the corner of my eye and skidded my car sideways so that she only hit us with a glancing blow (although it totaled my car). To top it off she had no license and no insurance. Police wouldn’t even come to the site because it was only a traffic accident with no injuries (that’s San Francisco for you).

In the end none of us were injured so that was the important thing.

SIGShooter · 2019-02-12T20:22:17.000Z

xevious:

I used to be an SQA engineer and then an SQA manager. I had a knack for breaking software. I know exactly what you’re talking about, as I’ve experienced it first hand. There were some developers who were pretty good at QC’ing their own code, but others were terrible. As long as it compiled that passed a quick simplistic test, they’d throw it “over the wall” for the QA folks to test it. At times it got so bad, I had QA engineers sit with the developer and test it right then and there. That worked, but it was a drain on resources. Our schedules slipped, but at least we had better code going out.

That’s why lots of companies are and have been going to off-the-shelf software solutions when possible. Might not be as full-featured as in-house software but it’s often better quality, cheaper, and offers cleaner upgrade paths. I wrote securities valuation and pricing programs so it was not something that could be purchases, fortunately for me