List Commands Library for 50g

[pier4r]

(03-18-2018 06:36 PM)DavidM Wrote:  All of this amounts to waaaay more effort (and headache) than was ever intended for a simple command to help shuffle a list. If there's truly no other way to reasonably validate it, then I'll simply remove it from the library. I have no desire to misrepresent the effectiveness of the command by leaving it in place with uncertain results, nor do I intend to become an expert in PRNGs simply to include a command in a library. It's beginning to look like that expertise is required to create a reasonable shuffling routine, though.

I agree that too much effort, unless it is an excuse to learn something, is, well, too much. But removing it? Couldn't you put just a disclaimer "tested under this condition, it perturbates the list. Although for some specific context the perturbation may be not chaotic enough, refer to the discussion XY"? I find LSHUF still pretty useful in a lot of cases.

Because if I understood your argument you say "either I get someone (or myself) that I can trust that says that my command is random enough, or I remove it" it will be just removed, as the chances to find someone that worked for such an application (perturbating lists and related metrics) are slim.

Wouldn't be enough to update LSHUF with your new version, as it seems working better, until the next error report? (if any)

Anyway this is a good chance to expose another reason why I do not update a software (or a product). In the case a feature that was there and it was good enough for me, gets removed. It happened already with listExt although I don't remember the command. Was one that you said "oh, it can be replaced with a couple of RPL commands".

By the way the removal of expected features happens often also in other contexts. See for example a new calculator that doesn't have the previous (loved) programming language (see the 48 with RPL and little RPN. Prime with no RPL). This or that game that gets simplified. This or that software suite that drops features and so on. In all those case, I often consider whether to move on the update or not.