…part the upteeth.

A few days ago I tried syncing my local website copy with the copy on my web host server and the sync failed. Digging into it I discover that the people who made my favorite sync software found a way to deactivate all the older versions that don’t require rent (they like to call it a “subscription”), which I refuse to pay and kept using the older version because it kept working. Hahahahaha…silly me. So I looked around for alternatives and I found what I thought was a good one in FreeFileSync, which is open source and runs on all my platforms.

But I almost instantly ran into a problem syncing my art room Mac files with the NAS copies. (NAS stands for Network Attached Storage. Mine is a box with two drives in it that pretends to be a single drive on my network. The two drives are mirrored so if one fails the other should still have good copies and I can just replace the failed drive.) Nothing I expected would sync because the software was reporting the files on the NAS were timestamped five hours in the future ahead of those on the Mac. I checked the NAS settings and it was applying the correct GMT offset, so I thought it might be a flaw in the software, and not having time to track it down, I manually copied the files that I knew had changed.

Under the hood all modern operating systems sync to Greenwich Mean Time and apply an offset depending on where you actually are. This allows people to move files across time zones and still retain the actual time the file was created or modified. That can be important if you are tossing time critical files across time zones.

So I figured I had a GMT offset problem with FreeFileSync and put it aside until I had time to debug the problem, and maybe file a bug report.

Then the other day I tried backing up the NAS. I have two USB backup drives that I’ve formatted to ext4 because that’s what the NAS uses. I use rsync and I have a preformatted rsync command that does the trick for me. Right away I noticed it was trying to copy everything off the NAS so I stopped it and took a look and finally saw that all the timestamps on the NAS were off by 5 hours ahead. So again I dug into the NAS settings and looked and yes, the NAS was applying the right offset. But then I noticed it’s value for GMT was off. By five hours ahead.

So I checked the setting for the time server and I discover that the time server it had been using was no longer available.  I’ve no idea why, but the main problem was the NAS didn’t throw an error message when it lost its time server, it just simply failed silently and continued. Probably the last time I shut it down when I went on vacation, when I turned it back on and it came back up it looked for its time server, couldn’t communicate with it, and just happily fell back to assuming everything on it was timestamped local time.

It caused me a lot of problems getting timestamps out of sync moving them between the NAS and the art room Mac as I worked.

G*damn silent fail…

So I set the NAS to talk to a different time server that it could still reach and everything is good again. So now at least I know that there isn’t anything wrong with FreeFileSync.

Also, that I will be debugging computer software systems for the rest of my life, retired or not…

A wee mystery solved recently, that had puzzled me off and on ever since I saw that bit of code back in my contractor days.

I was working the absolute worst contract job I ever had, at a local insurance company that had a reputation among contractors for being a brutally hostile workplace. I did not know about this until shortly before I left but it only took two days of working there before I was screaming at my handlers to get me the hell out of there.

I was there to track down and fix a bug in the report application they sent out to all their branch offices, that would randomly cause a blue screen of death. That happens in Windows when the operating system itself crashes and it takes a really serious problem for that to happen, because the operating system will catch most of the errors an application does not and close that misbehaving application down. The problem has to be really severe for it to percolate up the calling stack and take down the operating system.

So of course I needed access to their source code. But in the two week timespan I was there they never managed to get my network account set up or get me a workstation that I could log into in order to do my work. The manager there, when she wasn’t having shouting matches on the floor with the other managers about WE JUST HAD ALL THESE LAYOFFS AND YET NOW THEY’RE BRINGING IN ALL THESE EXPENSIVE CONTRACTORS!!!!!!!, would tell me scornfully I would just have to do my best. So I brought in my own laptop computer. But that didn’t give me access to the code I was supposed to be fixing. Somehow I managed to get one of the other developers to look kindly on me and make a print out of the code I was supposed to be fixing. 

I saw a potential source of the problem almost right away, and I’m proud now to be able to say I tracked down at least one of the blue screen causes (I’m pretty sure it wasn’t the only one). But that’s not what I want to talk about here.

The guy who wrote the code had taken a job elsewhere (couldn’t blame him). Supposedly he was a top level developer complete with a BS in computer science. Fine. Okay. Whatever. That only made the code I saw even more dumbfounding.

Never mind GlobalDummyInteger1, GlobalDummyInteger2, GlobalDummyInteger3, GlobalDummyInteger4, GlobalDummyString1, and so on and so forth. This made my jaw drop:

StringArrayForSomethingElse(1) = StringArrayForSomething(1)
StringArrayForSomethingElse(2) = StringArrayForSomething(2)
StringArrayForSomethingElse(3) = StringArrayForSomething(3)
StringArrayForSomethingElse(4) = StringArrayForSomething(4)
StringArrayForSomethingElse(5) = StringArrayForSomething(5)
StringArrayForSomethingElse(6) = StringArrayForSomething(6)

…and so on for 50 array elements. I must have stared at it for minutes. Then I thought…okay…some of the elements need to be treated differently for…some reason. So I walked down the entire fifty lines of code and they were all doing the same thing, copying the strings in one array to a different array one line of code, one array element at a time.

Understand…the above only takes a few lines of code if you use a For-Next loop…

For count = LBound(StringArrayForSomethingElse) To UBound(StringArrayForSomethingElse)

StringArrayForSomethingElse(count) = StringArrayForSomething(count)

Next count

It’s a loop. The first line initializes a counter and a maximum count based on the lower and upper bounds of the array. Let’s say it’s 1 to 50. It says basically do what follows, starting at the lower bound, and keep doing it for as long as the counter is less than or equal to the upper bound. The second line does the actual work of copying the array elements based on the value of the counter during that pass through the loop. 1…2…3…4…and so on. That last line increments the counter and throws it back to the first line. Every time the loop goes back to that first line the counter is evaluated again for is it less than or equal to the upper bound. At some point the third line is going to increment the counter (per my example) to 51, which is more than the upper bound and when the first line sees that’s its value the loop stops.

Simple. But even simpler if if they’re both dynamic (can have their length changed) arrays because then you can just copy one array to another with a single statement.

StringArrayForSomethingElse = StringArrayForSomething

The point being was pretty dumb to do it the way he did it, and if anything about the arrays changes then you have to go back through all fifty lines of code to fix it. I just could not believe a senior software developer with his bachelor’s degree wrote that code.

As the years passed I would occasionally think back to this and wonder about it. Then Elon Musk bought Twitter and went on a self absorbed rampage through it’s software engineers. And for the past week or so I’ve witnessed a lot of chatter about code reviews and performance metrics and suddenly it hit me: Management at this insurance company was measuring software developer productivity by how many lines of code their developers write per day. 

Or per pay period or whatever. So of course instead of using only a few lines of code to do a task, you use as many lines as possible. That is how you keep your job if that is how your job performance is being measured. And oh golly there are So Many Ways to take a simple line of code and break it out into dozens. If not hundreds. But then you’re just showing off.

It’s a really stupid metric, it almost makes software development look like piece work, but it seems many businesses use it, because management does not understand software development and maybe they need to focus on results and not micromanage the software maturity cycle. And yes, it results in bloated, buggy and hard to maintain code. At Space Telescope I had a project manager who would say negative productivity was a good thing. It was tongue in cheek but it had a serious meaning. Efficiently written code is easier to maintain and less prone to mistakes (bugs). Being able to take many lines of code and reduce them down to only as many lines as necessary is a good thing. Negative productivity.

I’ve no idea what sorts of metrics Musk is using to slash and burn Twitter, but I suspect it’s more of a knee jerk personal reaction to the developer and not their work. What we’re seeing in that foobar is something you often see throughout human history, that the legend is bigger than the man.