How to track a file (in various versions) used by several projects

[Steve Mullock]

I have used Mercurial (via TortoiseHg mostly) for a number of projects where a scripting 
language (Autoit) controls a piece of custom equipment. I break my code up into functional 
chunks, with the intention of reusing some parts in multiple projects. So for example, the 
code for systemA and systemB might both include "driver.au3". I would have a repository for 
systemA, a repository for systemB etc. There might be 3 - 10 projects that make use of 
driver.au3. They each have their own copy. They are updated independently.  

When I have to update systemA, a change to driver.au3 may be desirable, but I don't want to 
break other systems so only the copy local to the systemA repository is changed. The change is 
recorded the next time I commit systemA as part of the systemA repository. Now I would like to 
prevent the various copies of driver.au3 diverging too far from each other; they will diverge 
temporarily, but I will always strive to merge them again. At the moment I am doing it by 
hand, trying to spot common ancestors and making merges when I start a new system. 

I do have a repository for a directory only containing the "latest" driver.au3 and I copy in 
what I hope are  improvements, but I don't see any structure in that repository.

I am trying to think of a work-flow using Mercurial that would help with this. I suspect that 
the situation is quite common, so someone has already thought of the best way. At the moment I 
am thinking I would need a separate directory/repository containing just driver.au3 for each 
system. Will I have problems if I make a repository in a subdirectory of another repository? 
Can a single driver.au3 file be part of both the systemA repository (in the sub-directory) and 
also be the sole file tracked in the subdirectory repository?

If I am going in the wrong direction, I would be grateful for a warning and perhaps a 
suggestion of a better way.

Steve