There's a 3rd party code/headers/libraries tree that my project uses. My
code has dependencies on it. The 3rd party tree even has the build scripts
for building MY code.
Now, what's the correct way to manage the 3rd party code? Every few weeks I
get an update - the entire tree. Since the update can include new files as
well as *removed* files - and can even include structural changes - I use
this process for "upgrading":
1. Schedule a deletion of the old version from my WC. Commit.
2. Unpack the new version to where the old version used to live (on the WC).
Add it and commit.
The problems with this approach, if not obvious:
1. I lose history of the 3rd party tree.
2. The process requires two commits, and is therefore not atomic.
The second problem is minor, but the first may be important, if at some
point in time I will like to know at which version, for example, they've
added that variable to that struct.
What's the correct way of doing it?
Received on Mon May 15 09:06:19 2006