最近Stack Overflow上有一群讨厌perl的人,所以我想我应该把我的“关于你最喜欢的语言你讨厌的五件事”的问题带到Stack Overflow上。拿你最喜欢的语言来说,告诉我你讨厌它的五件事。这些可能只是让你烦恼的事情,承认的设计缺陷,公认的性能问题,或任何其他类别。你只需要讨厌它,它必须是你最喜欢的语言。

不要拿它和其他语言比较,也不要谈论你已经讨厌的语言。不要用你最喜欢的语言谈论你喜欢的事情。我只是想听到你讨厌但能容忍的东西,这样你就可以使用所有其他的东西,我想听到你希望别人使用的语言。

每当有人试图把他们最喜欢的语言强加给我时,我就会问这个问题,有时是面试问题。如果有人找不出他最喜欢的工具的5个缺点,那他对它还不够了解,不能提倡它,也不能利用它赚大钱。他还没有在足够多的不同情况下使用它来充分探索它。他把它作为一种文化或宗教来倡导,这意味着如果我不选择他最喜欢的技术,我就错了。

我不在乎你用什么语言。不想使用特定的语言?那就不要。你通过尽职调查做出了明智的选择,但仍然没有使用它?好吧。有时正确的答案是“你有一个强大的编程团队,有良好的实践和丰富的Bar经验。改成Foo是愚蠢的。”


This is a good question for code reviews too. People who really know a codebase will have all sorts of suggestions for it, and those who don't know it so well have non-specific complaints. I ask things like "If you could start over on this project, what would you do differently?" In this fantasy land, users and programmers get to complain about anything and everything they don't like. "I want a better interface", "I want to separate the model from the view", "I'd use this module instead of this other one", "I'd rename this set of methods", or whatever they really don't like about the current situation. That's how I get a handle on how much a particular developer knows about the codebase. It's also a clue about how much of the programmer's ego is tied up in what he's telling me.

憎恨并不是衡量人们了解多少的唯一尺度,但我发现它是一个相当不错的尺度。他们讨厌的事情也让我知道他们对这个话题的思考有多好。


当前回答

C#

它是一种很棒的语言,特别是在LINQ中,但是与c++相比泛型支持较差。它有如此多的潜力,但目前的实现只对强类型集合和类似的琐碎事情有用。下面举几个例子:

A generic argument cannot be restricted to enums (only classes or structs). A generic argument cannot be a static class. Why? This seems like a completely artifical restriction. You cannot specify that a generic type must have a constructor with a certain signature because you cannot have constructors on interfaces. Why not? It's just another method with the special name ".ctor". Similarly, you cannot specify that a generic type must have a static method, because those also cannot be declared on interface. Something like static T Parse(string s) would often come in useful. The compiler is too eager in prohibiting some casts which the programmer knows would actually work, so they require uglyness like (TheRealType)(object)value No covariance, eg. IList<string> cannot be converted to IList<object>, even though string[] can be converted to object[]. (Microsoft might be fixing this in C# 4.0, though.)

其他回答

Clojure

Lack of built-in syntax for optional and keyword parameters in function definitions. Sure, you can add it easily enough, but that means library writers don't use it. Pervasive destructuring hasn't proven to be a good substitute yet Lack of method combination (before/after/around methods of the sort found in Common Lisp) Too much reliance on Java interop, e.g. there's no built-in file IO Sometimes I want static typing. This one isn't pure hate; I usually prefer dynamic, and attempts to mix the two have been largely unsatisfactory There's no built-in fast binary serialization format for the built-in data structures, though I hear people are working on it

下面是关于Perl 5的更多内容,来自创建了大量Perl模块,特别是在Moose上工作过的人的观点。

The horrible brokenness that is overloading and tied variables. Both of these features are a failed attempt to allow transparent extension to the built-in types. They both fail in various ways, and require module authors like myself to either implement horrible hacks to support them, or to say "never pass an overloaded object to the foo() method". Neither alternative is really acceptable. Lack of proper hooks into the compilation process and the meta-model. Moose in general, and role usage in particular, could be made much safer if the Perl core allowed us to affect the compilation process via a sane API that allowed us to hook into the meta-model (packages, classes, etc.) Lack of named parameters built into the language. Instead, everyone reinvents this. It's annoying. Similarly, lack of optional types. I don't want a static language, but the ability to specify types and constraints, particularly on function/method parameters, would be great. Perl 6 gets this right. Types are optional, but very rich, and there's no fundamental difference between built-in and user-defined types. The backwards compatibility police. This is more of a cultural issue. A number of the above issues can never really be fixed, since Perl 5 has a very strong commitment to backwards compatibility. So even if something were to be added that effectively replaced the current ball of shit that is tie and overloading, those features will never be removed. Of course, backwards compatibility is also one of Perl 5's greatest strengths. Bonus hate: Perl's built-in exception mechanism is a joke. The fact that exceptions may be a string or object makes for an eternity of fiddly exception-catching code, and the lack of a catch in the language syntax is the wart on the wart.

c#中缺少预处理器。

我知道他们把它放在一边是因为有些人会滥用它,但我认为他们把孩子和洗澡水一起倒掉了。代码生成被认为是一件好事,在c++中,预处理程序是我的第一个代码生成器。

Haskell

Sometimes the type system feels backwards. What if I don't want the compiler to infer types for my variables? What if I want the opposite, where it does constraint checking on said variables? For example, instead of inferring the type of the elements of a list, it instead makes sure that they all belong to a particular typeclass. This is a subtle but huge difference that makes it difficult for me to program UIs. It can be done, but it takes more effort than it does in some other languages. Haskell rocks for the non-UI parts, but the UI I leave to an untyped language. Allowing the construction of infinite values leads to some really frustrating errors sometimes. NoMonomorphismRestriction. Bytestring handling bites me in the ass sometimes and you don't know it until your program crashes because you mixed them up improperly. Something is wrong here, when we are losing type information that should have prevented this. Typeclasses should be automatically derived for trivial cases, like witness types, but there's a strong potential for abuse there.

C++

The inconsistencies in the libraries related to char* and std::string. All C++ libs should take std::strings. Characters are not bytes with respect to iostream. I do a lot of byte-oriented work. Having a "byte" type and a "character" type would significantly make it simpler. That, too, would permit scaling to Unicode somewhat easier. Bit operations should be easy on a value. I should be able to access and set the n'th bit of a value without playing AND/OR dancing. The lack of a standardized interface for GUIs. This is where Microsoft has really been able to position themselves well with C#. A standard interface binding that OS makers provide would go really far for my work.