".ignore" files should not be stored in a project's source

Many version control systems use ignore files in a folder to indicate resources that should not be tracked , the target ignored files. These ignore files could be in any folder in the project where something must be ignored, natch.

These ignore files are a separate concern, not germane to the purpose of the source code. Should we also have files that indicate how to compile? Compile and config concerns are localized, either in the build scripts or in other types of management system. Likewise, the version control metadata should also be separate from the source.

This is comparable to how in the past Subversion version control system had .svn files throughout a project’s source folders. These were used as part of their vcs “bookkeeping”. Currently the “bookkeeping” was centralized into one .svn file at the project’s root folder, like found in Git. (Not an SVN expert so this is just what I see when I use SVN).

A few that use ignore files all over the place are: Git, Mercurial, CVS, and Subversion. Yes, Subversion too, except that it hides these in this nebulous ‘properties’ objects that are not seen but can cause great pain.

To see how to use these, see for example Git’s: gitignore(5) Manual Page

Any issues?
Other than be a context leak, any issues? None directly. Yet, if you search the web, there are many pleas for making automation tools and apps work around the inclusion of these files. Extra steps must be taken to remove from views in IDE or in searches. And so forth.

What should be done?
Ignore files or properties should be handled just like everything else. They should be localized in the version control system’s metadata folder (db or workspace).

For example, with Git, the .git repository could have a folder called “ignores”:


There are probably many reasons for the current approach: performance, simplicity, etc. WTHDIK

Leave a Reply

Your email address will not be published. Required fields are marked *