[Gnu-arch-users] Multiple id's for a file

[Catalin Marinas]

Hi,

Going back to the problem of tracking changes to a non-arch
repository. I track the Linux kernel releases in a local arch tree
from which I branch and do my development. It happens quite often to
add new files (mine or others) to my branch which will make their way
into the mainline kernel. In my mainline kernel arch repository the
files would appear with a different id. The situation is simpler if I
add a tag to the files written by me but it doesn't solve those taken
from others. A solution is to copy the id's from my branch to the
newly added files in the mainline tree but this cannot be achieved if
I would use a public arch repository of the mainline kernel.

For this situation, it would make sense to have multiple ids for a
file. When my file is added to the mainline tree, I could simply add
its id to the existing id and the merging would preserve the history
for that file (it would report a conflict at the first merge but it's
still better than removing the file in my branch and re-adding it with
a different id).

Catalin