This could continue with https://github.com/jtimon/bitcoin/tree/sighash2 and https://github.com/jtimon/bitcoin/tree/sighash3 or simply https://github.com/jtimon/bitcoin/tree/sighash_abstract. With an abstract class the libscript could use a minimal duplicated version of CTransaction if we decide to do that, probably only temporarily, until core is more divided (again, if we decide that including the whole core.o in libscript is too much right now).

May I suggest changing the copyright (in the files header) of edited files directly to be just MIT :)?

I think all nits belong to #4890 but I will happily rebase on top of a corrected version.

Mhmm, @theuni one of the tests builds didn't failed when it had to https://travis-ci.org/bitcoin/bitcoin/builds/36618762

It failed to compile...

@sipa, I know and I fixed that, but the travis builds should had all failed due to the compiling error, but build 5/7 didn't failed. So maybe there's something wrong with it, I don't know.

@jtimon Sorry, I misread your comment. Seems we don't compile the unit tests for OSX. @theuni?

Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/p4989_7ffba2382622ec9529cdd318cc12bc50ebf5b0b3/ 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.

Yes, OSX doesn't build the tests because they can't run there (it's a cross build from Linux). I figured we had enough builders handling the tests, so instead the OSX spends its time testing the dmg packing process.

I'd be fine with switching the tests on so that they build if you guys would prefer that, but I'm not sure it would do much for us.

It's fine, I thought you may not be aware of this or something. Thanks for the explanation.

@jtimon Thanks, you were right to point it out. The tests were created somewhat arbitrarily, I assumed they would be ever-evolving. If you think something needs more coverage, by all means mention it :)

Here's another version of this PR built on top of #5054 : https://github.com/jtimon/bitcoin/tree/sighash_explicit

Changed the class name from SignatureHasher to TxSignatureHasher
Also rebased on top of #4981 since that will probably be merged first.

Went back to being independent from #4981 since the rebase is straight forward anyway. Now the original SignatureHash function is maintained but renamed and moved to the anonymous namespace.

Always instantiate a new CTransaction in TxSignatureHasher's constructor and not only when a CMutableTransaction is provided as suggested by @laanwj.

Closing until at least 0.10