6 ms·
That most C++ programmers are so confused about how to properly use their language says quite a lot about C++ itself.
by luriel 14y ago
That most C++ programmers are so confused about how to properly use their language says quite a lot about C++ itself.
- tptacek 14y agoIs your point that programmers in other languages automatically make good choices about data structures and memory usage? Because that seems, respectfully, to be a dumb point. They do not.
- luriel 14y agoNo, but in other languages there seems to be much more consensus about what features one should use, how and when. In C++ it seems everyone has completely different views.
- aidenn0 14y agoI agree its a problem with C++, but IMO it is an inevitable consequence of its strengths: namely that it is regularly used across a very wide range of domains than other languages. I've sen it used on 16-bit microcontrollers, where templates and exceptions were considered harmful, and in large applications, where those features were considered essential.
- rumcajz 14y agoHow would you implement an O(1) list without breaking encapsulation then?
- tptacek 14y agoWhat do you mean, "O(1) list"?
- tg3 14y agoI assume he meant "a list from which an element can be removed in constant time."
- mpyne 14y agoPerhaps I'm being dumb, but doesn't std::list<T> already do exactly this? From SGI's STL docs: "A list is a doubly linked list. That is, it is a Sequence that supports both forward and backward traversal, and (amortized) constant time insertion and removal of elements at the beginning or the end, or in the middle"
- malkia 14y agoAs others have said - yes, as long as you keep the iterator not the object itself. e.g. from T itself you cannot find where it was stored, so with std::list<something>, and then only from something you can't find where it is (you can have to look for it in the list).
- tedunangst 14y agoThe question doesn't make much sense. Your C example totally breaks encapsulation.
- rumcajz 14y agoWell, what kind of problem is that in C? It's C. It has no public/private keywords. Nobody expect struct in C to be well-encapsulated.
- scott_s 14y agoIt's also not a problem in C++ if you don't care. Just because C++ gives you the means to enforce encapsulation doesn't mean you have to. And, as I said below, when your goal is performance and you need to break abstractions, it makes sense not to enforce encapsulation.
- ahupp 14y agohttp://www.boost.org/doc/libs/1_35_0/doc/html/intrusive/usage.html http://www.boost.org/doc/libs/1_35_0/doc/html/intrusive/usag... Not sure if this qualifies as not "breaking encapsulation" but it's as close as you're going to get.
- eklitzke 14y agoThe idiomatic way to access and manipulate STL data structures is through iterators. In this case, an iterator for a std::list is basically a pointer/reference to the person container in the article, so if you're referring to the list objects via their iterators there's no need for O(n) list traversal.
- itsreallynotme 14y ago>in other languages there seems to be much more consensus about what features one should use, how and when Especially in theory. Python so desperately wants to not be Perl in this regard, so much that they codified it into PEP 20: "There should be one-- and preferably only one --obvious way to do it." The theory is great, but if one reads their standard library it is clear that practice does not agree.
- peterwwillis 14y agoHe never mentioned other languages, so no, I don't think that was his point. I think his point was that C++ makes it hard to make good choices about C++.
- tptacek 14y agoI'm saying Java programmers make inappropriate data structure choices about Java programs, and boy do Ruby programmers ever make bad decisions in Ruby.
- peterwwillis 14y agoI got that. Not very nice to say his point was dumb because you wanted to make a different one, though. I think we can all agree that C++ is annoying and leave it at that.
- tptacek 14y agoHuh? I was sticking up for C++ (in this one context).
- hythloday 14y agoTo an extent, they're forced to - when all you have are dynamic arrays and hashtables (I'm thinking of python here), it's hard to go very far wrong. (Isn't it? Am I being naive?)
- JoachimSchipper 14y agoFrom yesterday's bug hunting session: sets (okay, not exactly hash tables) are not ordered data types, so do not use a set for intermediate values when order matters. (My fault - I'm just pointing out that it's not that hard to get Python wrong either.)
- dkersten 14y agoWell, sets are the canonical unordered collection, after all.
- Someone 14y agoNaive? I do not know, but your logic surely is :-) Consider the similar "if you have only one kind of string, it's hard to go wrong". If you need Unicode, but your only type of string is ISO-8859-1, or if you need to cramp many strings in memory, but your only string class uses UTF-32 internally, it's almost unavoidable that you go wrong. The advantage of having only few data structures is that the developers of the system have only few places where they have to spend effort on optimization. So if, like Python, your platform is popular, chances are that those few basic structures have been optimized to death. That does not necessarily make them optimal for every use case, though. The big advantage of C++ is that, if the need arises, you can (see below) make a special-case data structure that beats even the most optimized general-purpose one. Of course, that 'can' is relative. It is really hard to make a robust special-case data structure.
- btilly 14y agoIt is rather hard (though I believe not impossible) to find a use case where dynamic arrays and hashtables are not within a constant factor of the best possible data structure. Admittedly, that constant may be fairly large.