在另一个问题中,Mark高度评价了ide,说“有些人仍然不知道”为什么“他们应该使用一个……”。作为一个使用vim进行编程的人,并且在大多数/所有同事都使用vim或emacs进行所有工作的环境中工作的人,ide的优势是什么?我为什么要用它?

我相信这对一些人来说是一个棘手的问题,我对开始一场论战不感兴趣,所以请只回复你认为基于ide的方法更好的原因。我对听到为什么我不应该使用IDE不感兴趣;我已经不用了。可以这么说,我感兴趣的是听取“栅栏的另一边”的意见。

如果您认为ide可能适合某些类型的工作,但不适合其他类型的工作,我也很想知道为什么。


当前回答

Eclipse:

代码高亮显示,在后台编译,在执行过程中指出错误。

与javadoc集成,使用ctrl-Space提示变量名。

当我编译时,就会得到错误。我可以双击一个错误,它会显示相应的行。

很好地与JUnit集成,ctrl-F11运行测试,告诉我测试失败了。如果在输出窗口中出现异常,我可以双击某行,并将我带到失败的行。不仅如此,ctrl-F11还可以确保在运行测试之前编译所有内容(这意味着我永远不会忘记这样做)。

与ant集成。一个命令用于构建和部署应用程序。

集成调试器,包括web服务器的远程调试。

神奇的重构工具,搜索一段代码的引用。帮助我了解改变的影响。

总而言之,这让我更有效率。

其他回答

ide基本上是:

编辑器/代码完成,重构和文档 调试器 文件系统资源管理器 scm的客户 构建工具

都在一个包里。

您可以使用单独的工具或出色的可编程编辑器和额外的工具,如Emacs(也可以使用Vim,但IDEbility IMO稍微少一些)来实现所有这些功能(以及更多功能)。

如果您发现自己经常在一个实用程序和下一个可以集成到环境中的实用程序之间切换,或者如果您缺少这里列出的一些功能(其他文章中有更完整的功能),那么可能是时候转移到IDE了(或者通过添加宏之类的方法来提高环境的ideability)。如果你已经用一个以上的程序构建了一个“IDE”(在我上面提到的意义上),那么就没有必要转移到一个真正的IDE。

拥有IDE有以下优势:

编译通常是“动态的”,这意味着不再切换到命令行进行编译 调试是集成的,在IDE中拥有调试意味着步进调试器实际上使用就地编辑器来直观地显示执行了哪些代码 IDE通常对您正在使用的语言具有更多的语义知识,并且可以在输入时向您显示可能出现的问题。重构比“搜索替换”强大得多。

还有很多,也许你应该试试。

我认为这主要与开发者的认知范围有关。IDE提供了开发人员工作环境的宏观视图。您可以同时看到类层次结构、引用的资源、数据库模式、SDK帮助引用等。而且,由于您的击键会影响和影响如此多的事情,以及体系结构和体系结构交叉点的数量不断扩大,一次只使用一个代码岛工作变得越来越困难。

OTOH,“只有我和vim和手册页”让我对我的工作有了一个更精简的微观——但强烈而精确的——视图。如果我有一个用一种语言和一组静态库构建的设计良好、分区良好、稀疏耦合且高度内聚的代码库,那么这是可以的——这不是典型的情况,特别是当开发团队规模随着时间、距离和个人偏好而增长并重新构建代码结构时。

I'm currently working on projects in Flex and .NET. One of the nicer things about Flex is how few different ways there are to accomplish a standard thing - pull data from a database, open/close/read/write a file, etc. (Yet I'm using the Flex Builder/Eclipse IDE - a typical heavy-weight example like VS, because I'm still learning the basics and I need the training wheels. I expect to evolve back to vim once I'm confident of my patterns.) In this view, I can do what I need to do professionally by knowing a few things really really well.

OTOH, I can't imagine getting to that point with .NET because the view I'm expected to maintain keeps expanding and shifting. There much less conceptual integrity, and over several developers on a project over several months, much less consistency - but the IDE supports that, maybe encourages it. So the developer really needs to (and can more easily) know many more things adequately. Which also has the benefit of helping them answer (or even understand) a lot higher percentage of the questions on StackOverflow. I.e. we can have a deeper knowledge stack. And we can respond to a wider variety of help-wanted ads.

事情可能会朝着两个方向发展。也许对于“仅编辑器”作用域,这就像“如果你只有一把锤子,那么所有东西看起来都像钉子”。使用IDE方法,对于任何您想要固定在一起的东西,您都有广泛的紧固件和相关的工具范围可供选择- nals/锤子,螺钉/螺丝刀,螺栓/扳手,粘合剂/胶枪/夹子,磁铁等等-所有这些都在您的指尖(有向导帮助您开始)。

我从相反的方向来回答这个问题。我从小就在Makefile+Emacs的环境中编程。从我最早的DOS编译器,微软的Quick C,我有一个IDE自动化的事情。我在Visual c++ 6.0上工作了很多年,当我毕业到Enterprise Java时,我使用Borland JBuilder,然后决定使用Eclipse,这对我来说已经变得非常高效。

Throughout my initial self-teaching, college, and now professional career, I have come to learn that any major software development done solely within the IDE becomes counterproductive. I say this because most IDE's wants you to work in their peculiar I-control-how-the-world-works style. You have to slice and dice your projects along their lines. You have manage your project builds using their odd dialog boxes. Most IDE's manage complex build dependencies between projects poorly, and dependencies can be difficult to get working 100%. I have been in situations where IDE's would not produce a working build of my code unless I did a Clean/Rebuild All. Finally, there's rarely a clean way to move your software out of development and into other environments like QA or Production from an IDE. It's usually a clicky fest to get all your deployment units built, or you've got some awkward tool that the IDE vendor gives you to bundle stuff up. But again, that tool usually demands that your project and build structure absolutely conforms to their rules - and sometimes that just won't work for your projects' requirements.

我了解到,要与团队一起进行大规模开发,如果我们使用IDE开发代码,并使用手动编写的命令行脚本进行所有构建,那么我们可以获得最高的效率。(我们喜欢用Apache Ant进行Java开发。)我们发现在IDE中运行我们的脚本对于复杂的构建来说只是一个点击或者自动化的噩梦,用alt+tab到一个shell并在那里运行脚本要容易得多(而且破坏性更小)。

Manual builds requires us to miss out on some of the niceties in the modern IDE like background compilation, but what we gain is much more critical: clean and easy builds that can live in multiple environments. The "one click build" all those agile guys talk about? We have it. Our build scripts can be directly invoked by continuous integration systems as well. Having builds managed through continuous integration allows us to more formally stage and migrate your code deployments to different environments, and lets us know almost immediately when someone checks in bad code that breaks the build or unit tests.

In truth, my taking the role of build away from the IDE hasn't hurt us too badly. The intellisense and refactoring tools in Eclipse are still completely useful and valid - the background compilation simply serves to support those tools. And, Eclipse's peculiar slicing of projects has served as a very nice way to mentally break down our problem sets in a way everyone can understand (still a tad bit verbose for my tastes though). I think one of the most important things about Eclipse is the excellent SCM integrations, that's what makes team development so enjoyable. We use Subversion+Eclipse, and that has been very productive and very easy to train our people to become experts at.

我更喜欢IDE,因为它允许我集成编辑/编译/调试,只需单击一下从错误跳转到生成错误的行。此外,它允许使用操作系统标准接口显示多个信息窗格。简而言之,它为用户提供了一个基于鼠标的输入界面和一个现代的输出界面,而不是依靠上世纪70年代的技术和界面来帮助我。

IDE还有更复杂的用户和用途,我并没有说我使用过他们或了解他们所有人。当我需要的时候,我会学习它们。