6
\$\begingroup\$

The question is whether it should be allowed for the OP to edit to include timings of answers to a code related to benchmarking. Currently, there is a moderator who doesn't like this for this question. I think it stems from these rules.

However, it seems to me that the rules are inappropriately applied here, or at least nonsensical. How are we supposed to know what the best benchmark is if the OP isn't allowed to add those results to his post? I mean, I can't run my code on his computer, so my timings don't mean as much unless I run all of the different codes (and languages) on my machine (and then it's not necessarily the same as OP's timings).

I am not asking for the OP to include the answer codes, just to be allowed to have a edit at the bottom where the latest benchmark numbers from the answers run on his/her machine can be posted so that way there is a reasonable way to measure between answers.

\$\endgroup\$
7
\$\begingroup\$

If the code in the question already contains a benchmark, or it is obvious how the performance will be evaluated, then a simple comment on each answer would suffice. For example:

Thanks to your suggestions, the running time dropped from ~ 25 seconds to 13 seconds.

The question should not be cluttered with information about answers, nor would it be appropriate to kick the question to the front page with an edit to add that information.

If the code in the question did not include a benchmark, and a performance comparison of multiple answers is desired, then a Community Wiki answer would be appropriate. The wiki answer could contain the benchmark code and a discussion about the pros and cons of each solution. Other users can write comments about the validity of the benchmarking methodology there.

\$\endgroup\$
  • 1
    \$\begingroup\$ I forgot it would kick the question to the front page. I can see that mattering. \$\endgroup\$ – Chris Rackauckas Jul 20 '16 at 1:42
  • \$\begingroup\$ kick the question to the front page: it gets considered "active" by a new/edited answer, anyway - incorporating (/updating) information about answers in the question shifts focus /?lastactivity from the answer to the question. \$\endgroup\$ – greybeard Jul 29 '16 at 5:06
5
\$\begingroup\$

No.

Questions aren't about polling the results of the answers, questions should contain absolutely no information about results of answers.

If you want the OP's timings of the answer, get the OP to comment them onto the answer or edit them into the answer himself.

\$\endgroup\$
3
\$\begingroup\$

The chief problem is that the code was changed at the same time as the benchmarks. See this revision, which is the one that was rolled back.

If the asker wants to post code revisions, the places to do so are either

  1. An answer. I'd argue that "Edited code in compliance with @answerer's suggestions, which changed the benchmarks as follows:" would be enough to make it not just a code dump answer.

  2. Another question. Then the revised code can get new answers based on the new code and benchmarks.

If posting benchmarks requires changing the code, then changing the code still has to be done in the normal manner. No invalidating answers.

\$\endgroup\$
  • \$\begingroup\$ 2. is definitely favoured on Code Review. Follow-ups are the way to go. \$\endgroup\$ – Mast Jul 26 '16 at 18:38

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .