Talk:Mian-Chowla sequence

Distinct?

Hi, could you further explain what distinct means in the task description? Thanks. --Paddy3118 (talk) 09:50, 15 March 2019 (UTC)

Scrub that, The OEIS entry made sense to me. --Paddy3118 (talk) 09:54, 15 March 2019 (UTC)

Means a unique sum. E.G. 3 can't be in the sequence since 1 + 3 = 4 and 2 + 2 = 4. --Thundergnat (talk) 12:07, 15 March 2019 (UTC)

Execution Speed?

Hi, it could be nice to have as a third task, the program duration. This gives an indicator of the language efficiency (and the algorithm too). I have tried among several Basic language implementations and the elapsed time, for 100 terms, is spread between 2 seconds and 25 minutes ! --PatGarrett (talk) 11:48, 15 March 2019 (UTC)

In general, Rosettacode discourages emphasizing program execution speed. It is more about comparison of concepts, and focusing on speed tends to lead to heavily optimized code which may become difficult for a new language user to read. It is sometimes useful, or at least entertaining to have an idea of relative execution speeds, but I hesitate to make it, or even imply that it should be, a requirement.

When I am developing tasks, I try to choose goals that most languages should be able to do in about a minute or less of processing time. Selfishly in some part, because all Perl 6 RC entries are run daily against a nightly development build of blead Perl 6 for smoke testing, so tasks that take excessively long really extend the testing time. (Running ~1000 or so tasks takes a while even if they aren't excessively long.) --Thundergnat (talk) 12:07, 15 March 2019 (UTC)

| all Perl 6 RC entries are run daily against a nightly development build of blead Perl 6 for smoke testing

Any thoughts on the RC examples when used for this? Do you create some specific running order for the RC tasks in any way to aid in your verification? --Paddy3118 (talk) 14:08, 15 March 2019 (UTC)

I apologize for if I gave the impression that I am responsible for blead Perl 6 for smoke testing. I help out in small ways but SqrtNegInf is and has been the primary architect and driver of the smoke testing project. I don't believe that tasks are run in any specific order. Some (typically those that require user interaction in some way) are skipped. Tasks that use concurrency are run serially, (to allow them to use the extra processors). Otherwise, they are pretty much run in a big bunch, parallelizing where possible. For more accurate details, it may be better to inquire with SqrtNegInf directly. --Thundergnat (talk) 20:38, 16 March 2019 (UTC)

Thanks Thundergnat --Paddy3118 (talk) 09:29, 17 March 2019 (UTC)

Hello, thanks for the answer. I understand your concern, and also I am impressed by your Perl 6 involvement. Anyway this task is nice because it is also (but not only) an efficiency test for compilers about number crunching and dynamic array storage allocation. Have a good luck. --PatGarrett (talk) 14:14, 15 March 2019 (UTC)

Thanks for the kind words. Really credit for the task should at least partially go to Gerard Schildberger‎; he keeps adding interesting tasks that lead me to do further reading and come up with tasks based off of that. Gerard and I disagree occasionally; but he's smart guy, just takes himself much too seriously IMO. Ah-well. Cheers --Thundergnat (talk) 20:38, 16 March 2019 (UTC)

As well as the algorithm, timings are also dependent on the processor speed and OS and the language compiler/interpreter version - I've (reluctantly) included timing information for the Algol 68 and AWK samples which I ran on a Windows 7 System which appears to be about half the speed of the machine the VB.NET sample was run on - though I expect I have a different versions of the VB.NET compiler, .NET framework and operating system. --Tigerofdarkness (talk) 14:15, 17 March 2019 (UTC)