This is incorrect. script.h is very clearly used. Do not assume based on a buggy comment... You must review the header for structures and defines found in script.h before proposing removal.

NAK that bit.

The typedef is a duplicate, correct.

What in keystore.h requires script.h?
What is class CScript; for then?

hmmm. As I understood it, you want to include the header for references... but looking more closely, I see all references are either typedef or virtual. There is never a need for the compiler to explore or size the structure. Apologies.

ut ACK

Thanks, then base58.h doesn't seem to need script.h either.

@jtimon script.h inclusion is only correct if the symbols are defined somewhere. The keystore.h update was correct because it included "class CScript;"

The base58.h update is incorrect, as something is required to define CScriptID and CTxDestination. This gets back to direct/indirect issue that @sipa referenced in #4674.

I understood the criterion in 4674, but in this case I really thought that base58.h was script.h-independent: CScriptID is declared in key.h and I missed CTxDestination. But in fact it could be easily moved with CNoDestination from script.h to key.h, since they only depends on CKeyID and CScriptID (defined in key.h).
Anyway, I'll leave the old version without touching base58 and think better how this fits with other changes I'm working on around script.

Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/p4680_a381ee5d1c1c01c66e681e245f60f2de2b79b264/ for binaries and test log.
This test script verifies pulls every time they are updated. It, however, dies sometimes and fails to test properly. If you are waiting on a test, please check timestamps to verify that the test.log is moving at http://jenkins.bluematt.me/pull-tester/current/
Contact BlueMatt on freenode if something looks broken.