20

In most the languages I use, you simply cannot nest block comments, because the first occurrence of the "close" comment syntax closes the comment even if it was only an "inner" comment.

For example, in HTML

<!-- outer comment
<p>hello</p><!-- inner comment <p>world</p> -->
<p>this should BE commented</p>
-->

in this case, the outer comment ends on the first --> instead of the corresponding last one, causing the last <p> to print, when it shouldn't.

The same happens for languages that use /* */ for block comments, like in java, php, css, javascript, etc.

But my question is WHY is it that way? Why, by design, it is not allowed? I mention "by design" because I really doubt it is because of parsing problems, I guess the parsers are perfectly capable of keeping track of opening /*s and close the comments with their corresponding closing */s But they simply somehow decided it is not a good idea.

I already know that a workaround for this is to somehow change the inner closing comments, to avoid them to close , and only leave the last closing one. e.g. changing inner -->s and */s for - ->s and * /s . But that is obviously not convenient, and hard to do when you only want to discard blocks of code for debugging purposes. (other techniques are to nest everything in if(false){} blocks, but that is not the point here.

So, what I'd like to know is WHY nested comments are generally not allowed in several modern languages? there must be a good reason other than "others don't do it, we won't either" right?.

And as a plus, are there any other (not so obscure) languages that DO allow nested block comments?

Improve this question
15
  • 1
    HTML is actually a bad example – comments in HTML are more complex. Commented Jun 26, 2013 at 18:58
  • 1
    even if the constraints are old, new versions of the languages should implement it, i guess. (just as they implement any other functionality that would not work on older versions anyway) Commented Jun 26, 2013 at 19:00
  • 1
    Duplicates: programmers.stackexchange.com/questions/81072/… and programmers.stackexchange.com/questions/116330/… Commented Jun 26, 2013 at 19:00
  • 2
    @StinePike Nested comment could be used to add a message, while the nesting comment is for disabling some feature. Commented Jun 26, 2013 at 19:02
  • 4
    @StinePike because sometimes, for debugging purposes, you would like to discard large blocks of code that may already have block comments, and it is very easy to just wrap them with /* */ instead of other workarounds. Commented Jun 26, 2013 at 19:02

3 Answers 3

Reset to default
20

The reason is historical and has to do with the architecture of compilers.

For the sake of efficiency, most compilers traditionally parse the source code in two stages: the lexical analysis and the actual parsing of a token stream (that was produced by said lexical analysis). The lexical analysis is the part that recognises individual tokens, such as keywords, strings, number literals – and comments.

Again for reasons of efficiency, lexical analysis is traditionally implemented via a finite-state machine. These finite-state machines happen to recognise (= handle) regular languages, which fits perfectly for the above-mentioned tokens. However, it is not able to recognise nested constructs – this would require a more powerful machine (augmented by a stack).

Not allowing nested comments was thus simply a decision that traded off convenience for performance, and subsequent languages have by and large adopted the convention.

And as a plus, are there any other (not so obscure) languages that DO allow nested block comments?

There are some. The comments already mentioned Haskell and Pascal. Other languages are D and F#.

Improve this answer
Sign up to request clarification or add additional context in comments.

2 Comments

No, it would not require a stack -- how or why would it? You can trivially count the depth of nested comments with a single integer.
@rw-nandemo In formal language theory that’d equivalent. A regular language (implemented by a finite-state machine) cannot count. The next powerful abstraction is the class of context-free languages, which are canonically implemented via a pushdown automaton (~ FSM + stack). You can implement it differently but the outcome is equivalent.
0

HTML is generally a sloppy, non enforced type of markup. Browsers had to make decisions between what's right and what's realistic, sometimes they choose the second.

If you really want to comment out a piece of HTML, it's generally not a great idea to even use a single comment with HTML tags inside of it, but it's especially bad to start nesting comments.

You can never be absolutely sure how a browser will handle this when it parses the syntax.

Improve this answer

Comments

-3

This behavior is exhibited because anything between comment characters is a comment, including more comment characters. Yes, it would be possible to easily program the parser to treat them as nested comments like you describe, but that isn't quite consistent with a comment. A comment is meant to make everything between the open character and close character non-existent, regardless of what it is. Text, code, and comment charcters all get commented out.

Unfortunately, your suggestion that the reason is "other's don't do it, so we won't either" is also quite correct. People expect block comments to behave in a certain way, and get confused when they don't.

Improve this answer

1 Comment

“isn’t quite consistent with a comment” – nonsense. By the same logic a comment would never end because the end-of-comment mark is part of the comment. And I’ve yet to see a single person who got confused by nested Haskell or Pascal comments. In fact, the very existence of a concise, formal rule describing nested comments tells as that such a rule must be self-consistent. If it weren’t, it couldn’t be formally described.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.