Large projects get a lot of pull requests. On busy projects, these pull requests can fall behind. The farther they fall behind, the less likely they become to ever get merged.
You've seen this for yourself. Pull request #6 on a project working on pull request #600, still open because it's a good idea – just not quite good enough for anyone to do the tedious manual work of reconciling it with the master branch. Other common failure modes include delaying responding to merges with conflicts, or installing tools like "pager duty" to ensure rapid manual response to conflicts (see more in The life-cycle of merge conflicts ).
"Tedious manual work"? A phrase to make any programmer's ears prick up. There should be some way to automate that.
MergeResolver is an experimental GitHub application for JavaScript projects. When you install MergeResolver on a repository, it waits for a conflict to happen. Conflicts can happen with new pull requests, or when a new commit to master conflicts with one or more open pull requests.
If a conflict happens, MergeResolver clones your repository and sets to work. First, it tries to avoid the conflict by retrying the merge, but on the Abstract Syntax Tree (AST) level.
By treating code as ASTs rather than simply a string of characters, MergeResolver recognizes code's structure and semantics. This ensures all potential conflict resolutions will parse as valid code, and in some cases cleanly reconciles changes that would conflict under Git's built-in, line-oriented merge algorithms (as in the above example).
If AST merging fails, things get more interesting. MergeResolver understands different strategies for reconciling merge conflicts – strategies mined from large datasets of real-world merge conflict resolutions (see On the nature of merge conflicts ). With this knowledge it can generate possible conflict resolutions.
But how does it know which generated conflict resolutions are viable? MergeResolver relies on the only source of truth about a program's requirements: its test suite. MergeResolver finds a viable resolution by looking for a resolution that passes your program's test suite.
Once MergeResolver finds a resolution that passes your test suite, it uses GitHub's Checks API to provide a button (under the Checks tab on the pull request) that can be used to accept the resolution.
If you want to inspect the conflict resolution before accepting it, a link is provided to a diff of the merge resolution commit.
MergeResolver is currently available for direct installation . It is free for open-source projects to try. You can add MergeResolver to your Node-based JavaScript repository here .
Please let us know what works well or not for you, and we'll strive to make it useful enough that it becomes a staple of your team's development workflow.
This material is based upon work supported by the Defense Advanced Research Projects Agency (DARPA) under Contract No. D17PC00096. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Defense Advanced Research Projects Agency (DARPA); or its Contracting Agent, the U.S. Department of the Interior, Interior Business Center, Acquisition Services Directorate, Division III.
以上就是本文的全部内容,希望本文的内容对大家的学习或者工作能带来一定的帮助,也希望大家多多支持 码农网
猜你喜欢:本站部分资源来源于网络,本站转载出于传递更多信息之目的,版权归原作者或者来源机构所有,如转载稿涉及版权问题,请联系我们。
计算机组成与设计硬件/软件接口
[美] David A.Patterson、John L.Hennessy / 郑纬民 / 机械工业出版社 / 2007-4 / 75.00元
《计算机组成与设计硬件:软件接口》(原书第3版)是计算机组成的经典教材。全书着眼于当前计算机设计中最基本的概念,展示了软硬件间的关系,并全面介绍当代计算机系统发展的主流技术和最新成就。同以往版本一样,《计算机组成与设计硬件:软件接口》(原书第3版)采用MIPS处理器作为展示计算机硬件技术基本功能的核心。书中逐条指令地列举了完整的MIPS指令集,并介绍了网络和多处理器结构的基本内容。将CPU性能和程......一起来看看 《计算机组成与设计硬件/软件接口》 这本书的介绍吧!