Merge bitcoin#25503: test: pass datacarriersize option for tests using large outputs (instead of acceptnonstdtxn)

475aae8 test: pass `datacarriersize` option for tests using large outputs (instead of `acceptnonstdtxn`) (Sebastian Falbesoner)
b1ba3ed test: let `gen_return_txouts` create a single large OP_RETURN output (Sebastian Falbesoner)
f319287 test: assert serialized txouts size of `gen_return_txouts` helper (Sebastian Falbesoner)

Pull request description:

  By specifying the `datacarriersize` option instead of the more generic `acceptnonstdtxn` for functional tests, we can be more specific about what part of the transaction is non-standard and can be sure that all other aspects follow the standard policy. Transactions with more than one OP_RETURN output are [never considered standard](https://github.com/bitcoin/bitcoin/blob/749b80b29e875cc6afa1c2674cccdfd7115cc16a/src/policy/policy.cpp#L149-L153), i.e. we have to change the `gen_return_txouts` helper to create only a single output in order to get rid of the `acceptnonstdxtn` option. Note that on master there is currently no test using the `datacarriersize` parameter, so this PR indirectly also increases the test coverage.

  The change affects the tests `mempool_limit.py`, `mining_prioritisetransaction.py` (call `gen_return_txouts` directly) and `feature_maxuploadtarget.py` (calls `gen_return_txouts` indirectly via the `mine_large_block(...)` helper).

Top commit has no ACKs.

Tree-SHA512: c17f032e00d28f5e5880a4d378773fbc8b1995ea9c377f237598d412628fe117f497a44ebdfa8af8cd8a3b1e3127e0cf7692efbf5c833c713764a71a85301f23

Author: MacroFake

test/functional/feature_maxuploadtarget.py

@@ -46,7 +46,7 @@ def set_test_params(self):
         self.num_nodes = 1
         self.extra_args = [[
             "-maxuploadtarget=800M",
-            "-acceptnonstdtxn=1",
+            "-datacarriersize=100000",
         ]]
         self.supports_cli = False
 

test/functional/mempool_limit.py

@@ -23,7 +23,7 @@ def set_test_params(self):
         self.setup_clean_chain = True
         self.num_nodes = 1
         self.extra_args = [[
-            "-acceptnonstdtxn=1",
+            "-datacarriersize=100000",
             "-maxmempool=5",
             "-spendzeroconfchange=0",
         ]]

test/functional/mining_prioritisetransaction.py

@@ -26,7 +26,7 @@ def set_test_params(self):
         self.num_nodes = 1
         self.extra_args = [[
             "-printpriority=1",
-            "-acceptnonstdtxn=1",
+            "-datacarriersize=100000",
         ]] * self.num_nodes
         self.supports_cli = False
 

test/functional/test_framework/util.py

@@ -499,22 +499,13 @@ def chain_transaction(node, parent_txids, vouts, value, fee, num_outputs):
 
 
 # Create large OP_RETURN txouts that can be appended to a transaction
-# to make it large (helper for constructing large transactions).
+# to make it large (helper for constructing large transactions). The
+# total serialized size of the txouts is about 66k vbytes.
 def gen_return_txouts():
-    # Some pre-processing to create a bunch of OP_RETURN txouts to insert into transactions we create
-    # So we have big transactions (and therefore can't fit very many into each block)
-    # create one script_pubkey
-    script_pubkey = "6a4d0200"  # OP_RETURN OP_PUSH2 512 bytes
-    for _ in range(512):
-        script_pubkey = script_pubkey + "01"
-    # concatenate 128 txouts of above script_pubkey which we'll insert before the txout for change
-    txouts = []
     from .messages import CTxOut
-    txout = CTxOut()
-    txout.nValue = 0
-    txout.scriptPubKey = bytes.fromhex(script_pubkey)
-    for _ in range(128):
-        txouts.append(txout)
+    from .script import CScript, OP_RETURN
+    txouts = [CTxOut(nValue=0, scriptPubKey=CScript([OP_RETURN, b'\x01'*67437]))]
+    assert_equal(sum([len(txout.serialize()) for txout in txouts]), 67456)
     return txouts