WAIT

nhighton

I've heard several different ideas of how long WAIT 1 is. Netlinx Studio help says a millisecond - 1/1000th of a second, but other times says 'measured in 1/100th of a second'. I've also seen people on this forum stating that WAIT 11 is 1.1 seconds, so measured in 1/10th of a second!

Discuss & clarify!

Thanks,
Nick

Joe Hebert

WAIT 10 = 1 second

ericmedley

You can get into the milliseconds by using a timeline. However, it should be noted that this is not a dead-on millisecond measurement. I'd not use a timeline for measuring stuff. For that I'd go to processor ticks. The guys here on the forum have routines they regularly use to accurately time stuff. Me, I'm too right-brained to care about accurate timing.

svTech

You can also get finer wait times. WAIT 0.1 is a wait of 0.01 secs.

ericmedley

svTech wrote: »

You can also get finer wait times. WAIT 0.1 is a wait of 0.01 secs.

This is a relatively new thing.

Joe Hebert

ericmedley wrote: »

This is a relatively new thing.

Relativity new is relative but as far as I can remember WAIT .1 (0.01 seconds) has always been legal in Netlinx. At least I think it has.

jjames

Joe Hebert wrote: »

Relativity new is relative but as far as I can remember WAIT .1 (0.01 seconds) has always been legal in Netlinx. At least I think it has.

Agreed - I've used this as far back as 2005 (I know - nothing compared to some old-schoolers here.)

DHawthorne

I'm about as old-school as they come, and I had no idea you could put a fractional value on a WAIT.

I bet it came in with Netlinx, I can't imagine Axcent supported it.

Joe Hebert

DHawthorne wrote: »

I'm about as old-school as they come, and I had no idea you could put a fractional value on a WAIT.

I bet it came in with Netlinx, I can't imagine Axcent supported it.

That sounds right Dave. I don’t think we had anything less than a WAIT 1 (0.1 seconds) in Axcess.

Joe Hebert

As a follow up regarding WAITs in Netlinx:

Netlinx Help File wrote:

NOTE: If greater precision is required, the time parameter can be expressed as a decimal fraction, for example 0.1 to specify a wait time of 1/100th of a second. The range is 0.1 - 0.9.

jjames

DHawthorne wrote: »

I'm about as old-school as they come, and I had no idea you could put a fractional value on a WAIT.

I bet it came in with Netlinx, I can't imagine Axcent supported it.

I should have qualified my post that I've only ever dealt with NetLinx, and have briefly looked at an Axcent program or two.

mpullin

DEFINE_PROGRAM
WAIT 10 tp_feedback();

calls tp_feedback() every second.

jjames

mpullin wrote: »

DEFINE_PROGRAM
WAIT 10 tp_feedback();

calls tp_feedback() every second.

*GASP*
Are you trying to start a debate why not to run feedback in define_program or a timeline?

mpullin

jjames wrote: »

*GASP*
Are you trying to start a debate why not to run feedback in define_program or a timeline?

I'm just saying.

ericmedley

DHawthorne wrote: »

I'm about as old-school as they come, and I had no idea you could put a fractional value on a WAIT.

I bet it came in with Netlinx, I can't imagine Axcent supported it.

I'm so old skool they tore the school down Dave.

In all these years of doing Netlix I didn't know you could go below a 10th of a second. I too am from the old Accent days. I don't use waits that much anyway.

ProgressiveCode

ericmedley wrote: »

I'm so old skool they tore the school down Dave. In all these years of doing Netlix I didn't know you could go below a 10th of a second. I too am from the old Accent days. I don't use waits that much anyway.

I started programming AXcess in 1997 when I worked at AMX as the AXcess programmer for the Synergy department. It did not support fractional wait values. Wait 15 = approx 1.5 seconds. I believe NetLinx has always supported fractional wait times. Since timeline events are a part of NetLinx and they have an even sharper theoretical resolution, it makes sense that waits can be similarly defined. Wait 15.5 = approx 1.55 seconds. A timeline event with a length of 1550 would also = approx 1.55 seconds.

vining

So if the the overall wait flag is set, meaning at least one wait is in the queue how often then does the wait queue get checked and the wait time get decremented? I always assumed every 100ms but does this now mean that the queue is checked every 10ms or worse every 1ms instead? Is it adaptive and change depending on the resolution needed? It doesn't seem efficient to check a queue every 1ms or even 10ms (when wait pending) if 99.99999% of defined waits will only need a 100ms resolution especially given the amount of waits that could be active at any given time. That could be a good size for loop running every 10ms, checking and decrementing. I've never used a wait less than 1 although I think I have heard of this decimal stuff before, just never needed a lesser wait time so I never gave this concept any mind.

Joe Hebert

Please WAIT...

vining wrote: »

So if the the overall wait flag is set, meaning at least one wait is in the queue how often then does the wait queue get checked and the wait time get decremented? I always assumed every 100ms but does this now mean that the queue is checked every 10ms or worse every 1ms instead? Is it adaptive and change depending on the resolution needed? It doesn't seem efficient to check a queue every 1ms or even 10ms (when wait pending) if 99.99999% of defined waits will only need a 100ms resolution especially given the amount of waits that could be active at any given time. That could be a good size for loop running every 10ms, checking and decrementing. I've never used a wait less than 1 although I think I have heard of this decimal stuff before, just never needed a lesser wait time so I never gave this concept any mind.

WAITs are serviced after each event that is handled and after every pass thru mainline. Here’s a picture from the AMX help file (and TN264) which illustrates that.

One may or may not need a WAIT of less than 1 but perhaps a WAIT 3.5 might fit better than a WAIT 3 or a WAIT 4 in some situations. It's good to know we have the option.

vining

I think I'd rather have it checked by an internal wait timeline that starts when a wait is placed in queue and is killed when the last queued wait expires. Then have the timeline run at 100ms unless a decimal value is passed in and then it would then run at 10ms. Either timing rep rate would decrement the wait times by the timeline rep time. Any wait going into queue would turn on the TL if not running and the timeline would turn itself off when all waits are completed. It would automatically shift the rep rate depending on whether a decimal value is present in the queue or not.

Then the mainline would only have to check the wait TL every pass to see if it active or not and if active process it when its rep time is reached.

DHawthorne

Since WAITs are checked in mainline, anything that bogs mainline down can delay the way (not likely more than a few milliseconds on a NetLinx master, unless there is really a lot going on). It's a minimum time, not a precision timer.

nhighton

Thanks guys, as usual more information than required - I'm sure you wouldn't have it any other way

lightingcontrol

Is the timeline helpfile wrong??

In the time help file
From the Timed Waits helpfile:
Time is expressed in 1/10th second units

WAIT 10 (* wait one second*)

From the TIMELINE_CREATE helpfile:
"Time is measured in 1/100 of a second."


My testing of waits shows that my processor (NI2000) is actually 1/1000 as mentioned in the thread here!

ie. a time line interval of 1000 is one second

jjames

It's a typo - it's always been measured in thousandths.

jjames

Wait - I'm confused now . . . what do you want? WAIT or TIMELINE?

wait 10 = one second
timeline time array of {1000} is one second

Nothing's changed in the past 7 years of me doing this - can't speak for anything prior to that.

papadouk

i've been programming AMX since 2000, and as jjames stated:

wait 10 = one second
timeline time array of {1000} is one second

about time precision in WAITs, i never used it... When i need to deal with 1/100s or 1/1000s i use Timelines