All geek questions in one place

Git or no Git - A large binary version of the file, but no merging is required

There are many previous conversations about version control and SCM for binary files in the code base in StackOverflow. But I did not find any information about this particular project and the case used, which I am studying:

I have one parent binary - very large at several gigs in size. From this file, I have hundreds of “children” that are the same size, but each child is a little unique compared to the parent with very slight differences.

I will never need to unite children with parents, so I need serious tips and advice on maintaining the differences between 1 parent and n children:

  • Save only binary differences for each child.
  • When this child is needed (for download, implementation, etc.), recompile it using the parent file + differences.
  • NO MERGE is needed - I'm just curious to keep the differences in order to reduce the file size for each child.

I worked a lot with GIT, and I saw great StackOverflow posts about GIT here about the amazing ability to process binaries for version control, like this one here.

But my needs are simpler - I want to create a huge C or C ++ framework for preserving the differences in binary files and recompiling the source files using these differences. PLUS is the parent file. It. Is there a quick solution like GIT, but without additional features?

Thanks a lot - I try to avoid reinventing the wheel here.

+4
source share
1 answer

It looks like you want data deduplication, not version control. If so, try ddar . You can use it to store related binaries and it will make sure that the storage is efficient.

+4
source

Source: https://habr.com/ru/post/1437567/

More articles:


All Articles