LCB: Speed difference between LC 8 and LC 9

[[-hh]]

Recently I did a lot of LC Builder coding in preparing a talk and found that

1. LCB is in latest LC 9 up to 7 times faster than in latest LC 8 for lengthy numerical operations
   (simple but huge list building, in average 750 ms in latest LC 9, 5500 ms in latest LC 8).
2. Some new LCB features of latest LC 9 (e.g. reverse) were NOT built into latest LC 8.

So, if one wishes to have a widget that works as it should not only in LC 9 but also in LC 8 one is forced to develop in LC 8.

For me, (1) is (probably) a bug of LC 8, because such a speed difference is (probably) due to a (memory?)-bug that was removed for LC 9.
For new features, (2) is of course OK if these are "experimental".

Has one of you also seen such speed differences (or even crashes of LC 8 after using a widget intensively over several hours)?

[LCMark]

@hh: Okay - well - that's a *significant* speed difference! Can you share the code which shows it?

We did refactor / clean up the LCB VM implementation in 9 ( compared to 8 ), but I wouldn't have expected it to speed up things that much (I'm not going to complain if that is, indeed, the case however).

[[-hh]]

@ Mark.
I understand now (also from a remark by Ali) that LC 8 can't be LCB-upgraded to the state of LC 9.

So this is more a demo of LC 9 speed increase than a demo of a LC 8 bug.

I won't share the code right now _here_ as it is the core handler of my "give away" for LC Global in November. And it's too much work to take out a snippet because there are a lot of parameters settable in the PI.

So I've sent the working demo code (lcb) to your LC-email.