Difference between revisions of "Talk:Efficient Hex"

From Second Life Wiki
Jump to navigation Jump to search
m (thanks to Strife for happy optimism on dividing the one article in two)
(briefly restate some open questions found in archived Talk:Hex)
 
Line 1: Line 1:
= No Reproducible Measures Of Run Time Quoted =
I guess this [[Efficient Hex]] article is ok without quoting any milliseconds.
I'd also be ok with keeping only the extreme examples (i.e., the leading "Different, Clever and Fast" and "Different and Small" and the trailing "Brief, Clear, and Conventional"). We might make progress faster if we strove to analyse only those, rather than also working on the middling examples of "Clever and Fast" and "Clever and Small".
On the other hand, when thinking like an engineer I can imagine a use for any feature. So I'd be ok keeping those middling examples too.
-- [[User:Ppaatt Lynagh|Ppaatt Lynagh]] 14:37, 29 October 2007 (PDT)
= Persuasive Ways To Measure Fast =
The little I know about measuring script run time on closed source servers is:
1. I see the [[Efficiency Tester]] and [[Code Racer]] harnesses exist for calling [[llGetTime]] and [[llGetTimestamp]] to measure the run time cost of a fragment of code.
2. I see I can reproduce the results that those harnesses give me if I try just one sim on one day.
3. I don't yet understand how to reproduce consistent relative or absolute measures of run time from one sim to the next, from one day to the next.
So I have withdrawn my claims re the relative run time costs of the different exemplars. My claims were intuitively persuasive but actually unscientific, in the sense of not demonstrably reproducible. Likely I'll now stand by here, waiting indefinitely for the contribution of reproducible claims by someone more expert than I. Over in [[Talk:Efficiency Tester]] and [[Talk:Code Racer]] I may still contribute talk on occasion, trying to make progress there, since I am now incidentally academically interested in this undeveloped or unpublished science of measuring script run time on closed source servers.
-- [[User:Ppaatt Lynagh|Ppaatt Lynagh]] 14:37, 29 October 2007 (PDT)
= Not Much Talk Here Yet =
= Not Much Talk Here Yet =


Line 10: Line 34:


::Great to hear, yes thank you. -- [[User:Ppaatt Lynagh|Ppaatt Lynagh]] 13:21, 29 October 2007 (PDT)
::Great to hear, yes thank you. -- [[User:Ppaatt Lynagh|Ppaatt Lynagh]] 13:21, 29 October 2007 (PDT)
::As you can see, I've now archived the [[Talk:Hex]] and then briefly restated its open questions here. I hope that work helps too, -- [[User:Ppaatt Lynagh|Ppaatt Lynagh]] 14:37, 29 October 2007 (PDT)

Latest revision as of 14:37, 29 October 2007

No Reproducible Measures Of Run Time Quoted

I guess this Efficient Hex article is ok without quoting any milliseconds.

I'd also be ok with keeping only the extreme examples (i.e., the leading "Different, Clever and Fast" and "Different and Small" and the trailing "Brief, Clear, and Conventional"). We might make progress faster if we strove to analyse only those, rather than also working on the middling examples of "Clever and Fast" and "Clever and Small".

On the other hand, when thinking like an engineer I can imagine a use for any feature. So I'd be ok keeping those middling examples too.

-- Ppaatt Lynagh 14:37, 29 October 2007 (PDT)

Persuasive Ways To Measure Fast

The little I know about measuring script run time on closed source servers is:

1. I see the Efficiency Tester and Code Racer harnesses exist for calling llGetTime and llGetTimestamp to measure the run time cost of a fragment of code.

2. I see I can reproduce the results that those harnesses give me if I try just one sim on one day.

3. I don't yet understand how to reproduce consistent relative or absolute measures of run time from one sim to the next, from one day to the next.

So I have withdrawn my claims re the relative run time costs of the different exemplars. My claims were intuitively persuasive but actually unscientific, in the sense of not demonstrably reproducible. Likely I'll now stand by here, waiting indefinitely for the contribution of reproducible claims by someone more expert than I. Over in Talk:Efficiency Tester and Talk:Code Racer I may still contribute talk on occasion, trying to make progress there, since I am now incidentally academically interested in this undeveloped or unpublished science of measuring script run time on closed source servers.

-- Ppaatt Lynagh 14:37, 29 October 2007 (PDT)

Not Much Talk Here Yet

We have been slow to conclude that we should divide the brief & clear & conventional exemplars at the hex article from the clever & small & fast exemplars of the Efficient Hex article.

We have now proposed dividing the one article in two. If we next reach a consensus that that division is a step forward, then of course we should move all the talk of different unpersuasive ways to measure fast to here, away from the Talk:Hex.

-- Ppaatt Lynagh 07:32, 29 October 2007 (PDT)

I'm happy and optimistic with this solution. I'll take a closer review later but I don't foresee any show stoppers. -- Strife Onizuka 12:40, 29 October 2007 (PDT)
Great to hear, yes thank you. -- Ppaatt Lynagh 13:21, 29 October 2007 (PDT)
As you can see, I've now archived the Talk:Hex and then briefly restated its open questions here. I hope that work helps too, -- Ppaatt Lynagh 14:37, 29 October 2007 (PDT)