Idea: SVN monitor for pending commits?

放肆的年华 提交于 2019-12-06 05:17:30

svn does not require locks on checkouts; however, it supports locks. So, if your teams is more comfortable with this style of working, I suggest that you use the svn locking feature.

Basically, you have more flexibility than before:

  • If a developer just wants to "try something", he can update his working copy and modify the file without locking it.

  • If a developer wants to work on a file exclusively (to avoid merges), he can update his working copy, lock the file, work on it, and, when he's done, commit the change and unlock the file.

Since you were referring to MS SourceSafe I realize you might be in the MS world, but just in case you're not. If you're doing Java, Ruby, PHP, Javascript, Android, Flex work I'd recommend switching to IntelliJ IDE because it has a great view for SVN history built right into your IDE. You can see every check in (Repository tab), and the files changed in that check in (on the right hand side). I included a screen shot just is case. Really easy so you can just get to diffin' and cursing.

I don't think this can really help for these circumstances though. In my opinion, a release is ready when all its features are complete, and tested. When I have a dirty file in my working copy because I was trying to refactor something, or testing something, that's not a reason to keep the release. The information given by the release manager can't help you determine wether the release is done or not.

As for checking for conflicts, the only way to check for that is to monitor the repository. Again, not all dirty files I have locally will be committed.

Michael Kropat

If all the developers do feature development in separate branches, it is easy to see in the SVN repo browser what everyone is working on before code is merged back to trunk. This approach has the added benefit over the proposed approach, that the developers get to signify which "pending" changes they are working on are meaningful (i.e. what is committed to the branch) and which changes are just scratch work (which you won't see in the repo browser).

If you want a truly at glance view of conflicts in pending trunk merges, it is conceivable that you could create a script that tries to merge all the pairs of feature branches back to trunk. You could then graph a table that shows which combinations of branches merge successfully and which ones failed to merge. Either update the graph on a website in the post-commit hook, or schedule it as a batch job once a day to send out in an email to the developers. Should be much quicker to implement than a file-monitoring scheme reporting back to a centralized server.

标签
易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!