Lets say that we use some kind of "library/common"-code in several projects, and that the code will have different versions in these projects. One team (A) develop the lib-code, they're at version 10, an other team (B) use that lib-code (as a part of their project) but version 9.5 of it, and yet another team (C) uses version 8.5 of the lib-code. Now I should do analysis on this code with Coverity, how should I set up the projects and streams? I've read and understood My developers are performing Coverity analysis on their local source files. How should I set up my Streams and Projects?
Is the order important, ie should I add Project-C first and commit version 8.5, then create Project-B and let Project-C's stream contribute to that project etc?
Team B adds stuff outside Team A's domain that the A-guys (nor C-guys) shouldn't bother about, the same goes for Team C. What I'm after is that if someone in Team C does an analysis of a defect in the "library/common"-code, will that info propagate to Team A and their project?
The version of CIM you are using will be pretty important to at least some of the aspects of the answer.
But to answer the version independent part first: No, it does not matter what order you set things up in, except if you ever want to ask "which version did the defect show up in first". If C are really working on the oldest version of this code, I can see wanting to do this in the order you list, however, I imagine that all three teams might be introducing defects of their own, so this order is only important if there is an automatic merge of code from 8.5 into 9.5 into 10.
If that's not the case then it's more akin to three independent codebases which had common origin and share a lot of the code still.
The most important thing in 5.x is to
a) set the scope of triage correctly (you want to make sure that if Team C triages a defect that exists in B and A that their comments are visible to all)
b) that if there is existing triage on one stream in CIM when you create a new stream you "copy" the first stream and rename rather than creating a new stream.
In 6.0 this will become a lot simpler and you will be able to set things up as above from the start, or later on merge triage or split it, keep it the same for all streams or keep separate copies. So the above comments are most relevant to CIM 5.x.
Note that if you didn't do it the way you intended originally, in 5.5 there is a web services method to copy triage from one stream's defects to another, or you can wait till you upgrade to 6.0 and merge the triage information there. (6.0 is due out in April).
I hope this helps.