This is not the same as: Should we allow, even encourage, "feature comparison style" questions?

Do we want to support questions such as:

  • MySQL vs MariaDB
  • Postgres vs MySQL
  • Oracle vs MSSQL

(I'm not going to list the entire potential cross-set)

Poking around, I've come across this list of questions. Year and link provided so we can track/consider site-age trends. If you find other links, please feel free to edit and add them here. Unvoted means less than 3 votes, downvoted/upvoted is more than 3

Downvoted, not closed

Unvoted, not closed

Unvoted, closed

Upvoted, not closed

Upvoted, closed

Then there's a new question just asked today that is at risk of being closed: https://dba.stackexchange.com/questions/159007/in-what-ways-is-sql-server-2016-less-standards-compliant-than-postgresql-9-6

I want to make sure how we as a site feel about these. I know that SO has a no-shopping-list questions policy, and this feels like a shopping-list.

However, it's also a way for various experts to weigh in for the benefit of future googling, to help everyone, and so we can provide canonical answers that can evolve over time.

To that end, if we are going to adopt them, then let's also set up a preferred format, what we are looking for in a solid question and a solid answer, so we can at least let people get started asking them.

  • 7
    I don't think we need to strive to be a place where anything deemed a future benefit for Googlers should be given a pass. I find these "what are the ways x is inferior to y" questions to be open-ended and lacking of any quality that makes them relate to an actual technical problem the OP is trying to solve. In many cases they appear to be more interested in controversy than an actual answer. In this specific case the answerer would have to have intimate knowledge of the entire SQL standard and enough knowledge of both products in order to compile an exhaustive list. Thesis anyone? – Aaron Bertrand Dec 23 '16 at 5:01
  • Excellent followup point @AaronBertrand – jcolebrand Dec 23 '16 at 16:31
up vote 22 down vote accepted

These types of questions are harmful to the site and should be quickly closed and probably deleted.

I know that SO has a no-shopping-list questions policy, and this feels like a shopping-list.

It's not just SO, this has been established network wide for a long time. Just two meta.SE examples:

It doesn't really matter what we name the problem. We close questions because they're harmful in some way or another, not because they fit some technical close reason description ("shopping list" or whatever). It can be fun to "language lawyer" over these terms, but it is a distraction.

These types of questions are toxic for more reasons than it makes sense to list (irony noted). They encourage just about all the problems that lead to the exclusions mentioned in What types of questions should I avoid asking? and What does it mean if a question is "closed" or "on hold"? (among others).

Even attempts at a 'specific' question, like the latest example (compare standards compliance for Postgres 9.6 and SQL Server 2016) will cause problems:

  • Endless partial answers
  • Many off-topic comments, side-discussions, and arguments
  • Answers (and voting) based on personal opinion, interpretation, or bias
  • Answers that will date rapidly (e.g. with the next service pack)
  • Link-only answers ("I have the perfect answer on my awesome blog!")
  • One Q & A per product/edition/release combination would be needed
  • Likely to be just a rant in disguise - why does product x suck so much?

Even if we could somehow magically avoid all these problems, and obtain the One True Answer - one that comprehensively, impartially, and accurately lists all possible differences - it would be of very limited value.

These types of question might be 'good' and 'interesting' to address somewhere on the Internet (or in a book, video presentation, or wherever); but long network experience has shown they are just more trouble than they will ever be worth in Q & A format.

This answer was written based on this meta question's title, body content, and the most recent example question cited. Having now reviewed the other linked questions in detail, some (the unclosed ones mostly) appear to be of a slightly different type. They are covered in:

The distinction being that "which product is better for this specific practical application" may be on topic in some cases; abstract "compare product x and y" questions are not.


Footnote: The only way I could see this type of question working within our constraints, even in principle, would be to limit such questions to one chat room and one Wiki Answer, with close monitoring. I just don't think it's worth the effort or time investment. Too much risk for too little reward.

  • 2
    Precisely my thoughts and I am glad that another moderator supports my view. I tried to be impartial in my question, but I'm happy with the community consensus as indicated by the upvotes on this answer. Thanks Paul! – jcolebrand Dec 23 '16 at 16:31
  • It is a bummer we can't bounty answers on meta. – Erik Jan 11 '17 at 17:32

I don't think "this vs that" is a good fit for anything that attempts to be canonical. Each DBMS will change over time, and as such, the answers will vary substantially over time.

These types of questions are just troll fodder, in my opinion.

  • 2
    +1, agreed wholeheartedly – Philᵀᴹ Dec 23 '16 at 7:25
  • 2
    +1 but "These types of questions are just troll fodder" — they are often asked and answered in good faith but still don't fit. – Jack Douglas Dec 23 '16 at 10:16
  • What about In what ways is SQL Server 2016 less standards compliant than PostgreSQL 9.6? That's very specific about two versions and spec adherence. – Evan Carroll Dec 27 '16 at 20:12
  • 2
    @EvanCarroll - no offence, but I think Paul covered that pretty well in his answer. – Max Vernon Dec 28 '16 at 18:05
  • Then clearly, your answer doesn't contain your objection. It's not about a DB changing over time. DBs always change over time. If pegging the question to a specific version of a DB doesn't satisfy your concerns, then you should perhaps add your concerns to your "answer." – Evan Carroll Dec 28 '16 at 18:13
  • It rather clearly does answer the question, as evidenced by the up-votes on the answer. Feel free to downvote me, by the way. I don't care either way :-) – Max Vernon Dec 28 '16 at 18:20
  • 2
    ha!! troll folder :) – Shanky Jan 5 '17 at 7:00

You must log in to answer this question.

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