59

If a comment has [short title](http://lo-o-ong.example.com/url) link, then http://lo-o-ong.example.com/url is counted towards the total comment length limit.

The answer to the same issue (to ignore the link URLs) is: the current limit is enough, or to span a comment over several physical comments, or to use URL shorteners.

The current limit is especially limiting for non-English URLs. Compare the markdown for:

[link](https://en.wikipedia.org/wiki/Computational_complexity_theory)

vs. the same article in Russian:

[ссылка](https://ru.wikipedia.org/wiki/%D0%92%D1%8B%D1%87%D0%B8%D1%81%D0%BB%D0%B8%D1%82%D0%B5%D0%BB%D1%8C%D0%BD%D0%B0%D1%8F_%D1%81%D0%BB%D0%BE%D0%B6%D0%BD%D0%BE%D1%81%D1%82%D1%8C)

Splitting manually into multiple comments due to a long URL is busywork -- if a computer can do it; it should do it for you i.e., if the database limit for comments can't be increased, then let's allow the JavaScript to submit multiple comments. It could be restricted only to the case when the reason that the comment exceeds the size is a long href (visible comment text is under the limit), to avoid encouraging flooding (a very simple implementation will do -- it is enough if it works in the common case).

Shortened URLs are bad. Writers have to obfuscate their URLs. Readers don't see where they go. Links often serve as a reference to additional material so we shouldn't discourage comments that back up their statements with references.

How different is this Russian Wikipedia link from an obfuscated one?

The rendered version looks like this: ссылка. Hover over to see the decoded url: http://ru.wikipedia.org/wiki/Вычислительная_сложность (it is clear if you know Russian — the assumption is that the links are used on Stack Overflow на русском where people do understand Russian) that it is less obfuscated than a shorten url e.g., ссылка (https://goo​.gl/DPvnVF).

  • 1
    This is a good suggestion and makes perfect sense. – paulmz Oct 3 '15 at 15:28
  • 4
    How different is this Russian Wikipedia link from an obfuscated one? Not everybody can map arbitrary ASCII/Unicode numbers to the characters in a foreign language just by looking at the escape codes. – Deer Hunter Oct 3 '15 at 20:23
  • 5
    The only thing I could see going wrong is that this would be that I could then post 1000 links in a comment and... well you get the idea. – Diminutive Colossus Oct 3 '15 at 23:05
  • 8
    @DeerHunter: obviously, browser decodes percent-encoding for you – jfs Oct 4 '15 at 0:02
  • 2
    @TinyGiant: empty titles are trivial to check: [](http://example.com) [](example.com) vs. [non-empty example](http://example.com) non-empty example – jfs Oct 4 '15 at 0:05
  • This makes perfect sense, until you remember that the column size for comments in the DB is around 600 characters (the current limit), and it would need to be expanded for this to have any impact. – Kevin Brown Oct 8 '15 at 16:18
  • 2
    @KevinBrown: you've missed the part about splitting comments on the client. – jfs Oct 8 '15 at 16:20
  • 3
    Instead of not counting the URLs, I would suggest making them all of the same length. E.g. in Twitter all URLs have the same character count (23). This can be done by e.g. storing URLs separately and only a hash/index in the comment itself. – Igor Skochinsky Apr 7 '17 at 9:02
  • @IgorSkochinsky it is one possible way to implement it (essentially, it embeds the URL shortener into SE itself but it can be transparent for the user). – jfs Apr 7 '17 at 13:53

You must log in to answer this question.

Browse other questions tagged .