Refactoring is your friend / moving from 6.x to 9.x

[Curry Kenworthy]

Richard:

 > I'd like to follow their progress [...]
 > supporting the QA process benefits my work as well.

Thanks, Richard! Yes, if you're not on the list for that bug, that would 
superb to sign up. I thought it might be considered old or familiar news 
here since tests and observations had been posted and discussed before 
by myself and others, so at first I didn't want to sp*m too much info 
again. Here we go in reverse chronology:

Jan 2019: Current mention, as a reminder and still-the-same 902 status 
update.

Sep 2018: LC Version "Showdown" video with several tests between 6.7 and 
9, to accompany separate Bug Report and Test Stack. (3 Links.)

<http://lists.runrev.com/pipermail/use-livecode/2018-September/250004.html>

2018: Several list discussions that touched on performance. Including 
your JS speed comparison suggestion, which I included. Also Mark's note 
about array work, and some array vs list tests by others.

Jan 2017: LC "Fight Night" video with demo tests between LC 6.7, 5.5, 
8.1.2 and 9 dp 4. At that point 9 was still very much under 
construction, so 8 was the stable data point. Didn't have a chance to 
post followups that year.

<http://lists.runrev.com/pipermail/use-livecode/2017-January/234093.html>

Prior: As I recall, I noticed performance differences from LC 7/8 on and 
mentioned it, but heard back officially that the new LC would go through 
an optimization stage, so I waited accordingly.

Of course all this may NOT cover all performance issues; it's a limited 
set of tests as examples.

 > to rule out all possible impact from type coercions
 > which may involve encoding operations.

True enough; I've mentioned my own rule of refraining from speculating 
much on the causes, but pointing out the areas affected, and countering 
a just-Unicode-as-expected meme that pops up now and then. The "Fight 
Night" was designed to pin down what I had encountered in my own work - 
that several areas were affected, not just text and arrays. Some tests 
were designed to avoid type coercions to the extent possible. (I think 
it was especially "Root Loops" - the original version, not the big calc 
variation.) So I can't say for sure that it's not happening, but in 
those cases it shouldn't be.

Hope you enjoyed the R^3 link antimeme joke! :) Miss our periodic 
catch-up calls; maybe we can do one later this year when we have a chance.

Best wishes,

Curry Kenworthy

Custom Software Development
"Better Methods, Better Results"
LiveCode Training and Consulting
http://livecodeconsulting.com/