DATE and CDATE in 2017

[rkmcswain]

For me, these system variables return different values in 2017 than they do in earlier versions.

Can I get help verifying this?

Using this bit of code

(progn
  (setvar "dimzin" 0)
  (repeat 100
    (princ
      (strcat
	"\n"
	(rtos (getvar "date") 2 12)
      )
    )
  )
  (princ)
)

In 2017 I generally get the same exact answer for all 100 replies, which means if you subtract the last reply from the first, the net elapsed time = 0.0.

In earlier versions, the 100 answers almost all vary and the difference between the first and last is about 0.000000914 (presumably, the time it takes to process this code)

Why does this matter? Using the function below, note the responses from 2016 and earlier, compared to 2017.

(defun C:foo ()
  (defun gettime ()
    (setq s (getvar "DATE"))
    (setq seconds (* 86400.0 (- s (fix s))))
  )
  (setq begin (gettime))
  ; introduce a delay so we have something to time
  (command "._delay" 500)
  (setvar "dimzin" 0)
  (princ (strcat "\n" (rtos (- (gettime) begin) 2 12)))
  (princ)
)

----

;;; Response in 2016 and earlier = approx 0.5 seconds, which is expected

Command: FOO
0.502994656563
Command: FOO
0.505006313324
Command: FOO
0.502994656563
Command: FOO
0.505006313324
Command: FOO
0.504000484943
Command: FOO
0.505971908569
Command: FOO
0.501988828182

---

Response in 2017 = 0.0 or 1.0 seconds, everytime. Why?

Command: FOO
0.999994575977
Command: FOO
0.000000000000
Command: FOO
0.999994575977
Command: FOO
0.000000000000
Command: FOO
0.000000000000
Command: FOO
0.999994575977
Command: FOO
0.999994575977
Command: FOO
0.000000000000
Command: FOO
0.000000000000

What am I missing here? Is there another sysvar I'm overlooking that would have this effect?

TIA.

[rkmcswain]

@BlackBox_ - do you have 2017? Can you verify this if you have a minute or two? Thanks.

[BlackBox_]

---

@rkmcswain wrote:

@BlackBox_ - do you have 2017? Can you verify this if you have a minute or two? Thanks.

---

Hi @rkmcswain -

I do... Err... I did; reinstalling it now.

Cheers

[Edit] - always forget the '@' malarkey. Haha

[BlackBox_]

@rkmcswain - 

From console; here's what I get in 2016:

_$ (bench '(C:FOO) '() 10)

C:FOO
0.520978868008
0.507017970085
0.507983565331
0.506977736950
0.509995222092
0.507017970085
0.507983565331
0.508989393711
0.508023798466
0.508989393711
Elapsed: 5844
Average: 584.4000

... Here's what I get in 2017:

_$ (bench '(c:FOO) '() 10)

C:FOO
0.999994575977
0.000000000000
0.999994575977
0.000000000000
0.999994575977
0.000000000000
0.999994575977
0.000000000000
0.999994575977
0.000000000000
Elapsed: 5078
Average: 507.8000

... So, it looks as though Autodesk has actually increased performance, and made it completely inconsistent. Haha

[rkmcswain]

Thank you. See how 2017 always returns either 0.0 or 1.0 (rounded)?

Somebody broke something.... 

@BlackBox_

[BlackBox_]

---

@rkmcswain wrote:

Thank you. See how 2017 always returns either 0.0 or 1.0 (rounded)?

Somebody broke something.... 

@BlackBox_

---

You're welcome, @rkmcswain; I'm always happy to not help. Haha 

Cheers

[rkmcswain]

So now I need an alternative for my timers in 2017.... thoughts?

[rkmcswain]

Answer received.

[BlackBox_]

Best I can suggest, is to take the average.

Not to mention, if this is a known issue, where is that documented, so as to not leave users/developers in the lurch?

Or is this only a known issue, now that you've pointed it out?

[rkmcswain]

@BlackBox_ wrote:

Not to mention, if this is a known issue, where is that documented, so as to not leave users/developers in the lurch?

Or is this only a known issue, now that you've pointed it out?

I know those were rhetorical questions, but I feel the need to answer.

1. I don't know.

2. I don't know.

[BlackBox_]

---

@rkmcswain wrote:

@BlackBox_ wrote:

Not to mention, if this is a known issue, where is that documented, so as to not leave users/developers in the lurch?

Or is this only a known issue, now that you've pointed it out?

I know those were rhetorical questions, but I feel the need to answer.

1. I don't know.

2. I don't know.

---

Perhaps you're right.

I know that Autodesk doesn't really like folks asking difficult questions (and I get that), but still; It would be greatly appreciated if Autodesk were to answer.

Cheers

[rkmcswain · Accepted solution]

Thanks to @LeeAmbrosius for pointing out the undocumented (at the time) sysvar MILLISECS. It is documented for 2018 ! We can use this to measure time in milliseconds again. More info here.