erratic movement of controls with repeat

[jacque]

i ran Craig's test script on OSX High Sierra, LC 9.0.2. I ran 10 full runs and saw no difference in performance between them. I don't think it's LC or the OS.

I'd guess graphics card or CPU.

iMac (Retina 5K, 27-inch, Late 2015)
3.2 GHz Intel Core i5
32 GB RAM
AMD Radeon R9 M390 2048 MB

[dunbarx]

We do get off into the woods on occasion...

I will try this handler at home on a macBook Air.

The machine I was referring to is an iMac, 3.2GHz,, late 2012., OS 10.13.4

Graphics card: NVIDIA GeForce GTX 675MX 1024 MB

Since Jacque sees none of this, then if it is just something peculiar to my machine, I will actually feel just fine, in that LC per se is not the issue.

But it does show the pitfalls of trying to accommodate all these OS's, processors, etc. What would one do if most customers were happy with an app, but one guy was complaining all day long, and nobody took him seriously?

Craig

[richmond62]

it does show the pitfalls of trying to accommodate all these OS's, processors, etc.

It does indeed. However, you don't want to fall down the slippery slope that
Political Correctness did, where, in trying to keep everybody happy all of the
time you end up with something that is so tame it is virtually worthless.

There is no way, on earth, that a developer can predict and take into account all
the possible combinations of "OS's, processors, etc." So one has to work out some
sort of "average end-user", which, full of pitfalls that that is, at least is better
than either saying "to hell with everyone" or "I want to keep everyone happy all of the time."

[FourthWorld]

I will try this handler at home on a macBook Air.

Please report back what you find.

But it does show the pitfalls of trying to accommodate all these OS's, processors, etc. What would one do if most customers were happy with an app, but one guy was complaining all day long, and nobody took him seriously?

Easy to remedy by taking him seriously, and running through good diagnostic measures to pin down the root cause.

This case is mysterious for many reasons, but the central premise of the question is sound: execution speed of a routine should not change if nothing in the system changes.

In addition to pinning down the component with the steps above, it may also help to pin down the problem itself:

Is the issue intermittent or consistent?

Is it always the case that the second and subsequent runs of that script are noticeably slower?

If not consistently the second, is there any consistency to how many iterations need to be run to see it slow down?

Once it slows down, does it stay at that speed, get ever slower, or resume previous quick speed?

[dunbarx]

Is it possible to post a video? I just ran this again, and the differences between runs are startling.

Craig

[dunbarx]

Is the issue intermittent or consistent?

--- Consistent in that it always happens in some form

Is it always the case that the second and subsequent runs of that script are noticeably slower?

-- Not always, oftentimes it takes one or more subsequent runs to see the effect.

If not consistently the second, is there any consistency to how many iterations need to be run to see it slow down?

-- Nope, sometimes one or two, sometimes twelve, for the first noticeable slowdown.

Once it slows down, does it stay at that speed, get ever slower, or resume previous quick speed?

-- Once the slowdown starts, the next handful become slower and slower until the final slow speed is a fifth of the "correct" speed. And then, reliably, the "correct" speed will reassert itself, and then another form of the process repeats. It never rights itself. The number of runs from one "correct" run to the next can be anywhere from 5 to 20.

Odd, eh?

Craig

[FourthWorld]

Is it possible to post a video? I just ran this again, and the differences between runs are startling.

You can post a link to a video.

[dunbarx]

The same handler on my MacBook Air at home shows no sign of the odd behavior.

Is there any value to pursuing this? I do not believe I have anything unusual in my work iMac. I will restart tomorrow, and with only LC open, see what transpires.

Craig

[FourthWorld]

Is there any value to pursuing this?

I don't know. I can't see the problem here. Maybe just confirming that it appears to be limited to one machine is enough.