I really don't understand this request based on the input and expected outputs. Why is 'foo -> j -> bar' relevant? 'j' and 'bar' are both direct dependencies of foo, but so is 'a'...

If foo is the only (or one of a few) package to depend on bar, you could do pactree -r bar. This might give you shorter output.

The formatting is messed up, but I may have messed this feature request up even more.

You're right, the shortest path is not 'foo -> j -> bar', but 'foo -> a -> b -> j -> bar'.
If there are multiple paths of the same length, any of them could be printed.

I stumble upon this and liked the idea, but I'm not really familiar with pactree's source code nor do I know how the output would/should look like, so I just did something similar in pacdep; Just pushed to the branch next[1]. pacdep doesn't really do the same as pactree, so you might not necessary be able to get the results you're looking for, but I figured I might as well mention it.

[1] https://github.com/jjk-jacky/pacdep/tree/next

I haven't figured out the optimal combination of the switches, but

$ ./pacdep -PSs gtk2 | grep ^" pixman"
pixman 727.00 KiB <- cairo-infinality-ultimate <- gtk2
$ ./pacdep -PSs gtk2 | grep ^" glibc"
glibc 31.37 MiB <- pixman <- cairo-infinality-ultimate <- gtk2
$ ./pacdep -PSs gtk2 | grep ^" iana-etc"
iana-etc 3.44 MiB <- filesystem <- glibc <- pixman <- cairo-infinality-ultimate <- gtk2

looks fine. Thanks!