873beca6d [tests] Rename NodeConn and NodeConnCB (John Newbery)
Pull request description:
Final step in #11518
NodeConn -> P2PConnection
NodeConnCB -> P2PInterface
This is basically just a rename. Should be an easy review.
Tree-SHA512: fe1204b2b3d8182c5e324ffa7cb4099a47ef8536380e0bb9d37a5fccf76a24f548d1f1a7988ab8f830986a3058b670696de3fc891af5e5f75dbeb4e3273005d7
ff8a9b0 [tests] Add combinedlogslen argument to test_runner.py (John Newbery)
dba94ea [tests] [travis-ci] Move Travis functional test log post processing to test_runner (John Newbery)
bba1c54 [tests] Improve logging shutdown and add hint for combine_logs (John Newbery)
Pull request description:
Replaces #11779 . Notes from that PR:
> Currently, when a functional test fails, the debug logs are printed sequentially to the travis log. This makes debugging race conditions based on the travis log hard. Instead, all logs events should be combined and sorted by their timestamp, then appended to the travis log.
@MarcoFalke
Tree-SHA512: 56c80067d6a2c92f7e6a35e3ae5160637a0de052d9da593c7be6e02233544a93c66d62456f903f85e2edc09e31ab4bdafd1aed1d9897ae48c634f82631f856f7
e9dfa9bcc [tests] Move version message sending from NodeConn to NodeConnCB (John Newbery)
dad596fc3 [tests] Make NodeConnCB a subclass of NodeConn (John Newbery)
e30d40438 [tests] Move only: move NodeConnCB below NodeConn (John Newbery)
4d5059856 [tests] Tidy up mininode (John Newbery)
f2ae6f32a [tests] Remove mininode periodic (half-hour) ping messages (John Newbery)
ec59523c5 [tests] Remove rpc property from TestNode in p2p-segwit.py. (John Newbery)
Pull request description:
This is the final step in #11518, except for possibly renaming - for motivation, please see that PR.
If this is merged, then migrating the test framework from asyncore to asyncio should be easier (I say should because I haven't dug too deeply into what would be required).
Requesting review from @ryanofsky , since he always has good feedback on these refactor PRs, and I'd appreciate his take on this refactor. Note particularly that I've reverted the change suggested here: https://github.com/bitcoin/bitcoin/pull/11182#discussion_r148859555 . The idea, as always, is to present a simple interface to the test writer.
Tree-SHA512: 94dd467a13ec799b101108cf47d4dccb6f6240b601e375e3d785313333bbb389c26072a50759aca663bbf3d6c8b867b99e36ae8800ab8ea115e0496c151926ce
This commit moves the logic that sends a version message
on connection from NodeConn to NodeConnCB. NodeConn should
not be aware of the semantics or meaning of the P2P payloads.
This makes NodeConnCB a subclass of NodeConn, and
removes the need for the client code to know
anything about the implementation details of NodeConnCB.
NodeConn can now be swapped out for any other implementation
of a low-level connection without changing client code.
c1e5d40 Make debugging test crash easier (MeshCollider)
8263f6a Create walletdir if datadir doesn't exist and fix tests (MeshCollider)
9587a9c Default walletdir is wallets/ if it exists (MeshCollider)
d987889 Add release notes for -walletdir and wallets/ dir (MeshCollider)
80c5cbc Add test for -walletdir (MeshCollider)
0530ba0 Add -walletdir parameter to specify custom wallet dir (MeshCollider)
Pull request description:
Closes#11348
Adds a `-walletdir` parameter which specifies a directory to use for wallets, allowing them to be stored separately from the 'main' data directory. Creates a new `wallets/` directory in datadir if this is the first time running, and defaults to using it if it exists.
Includes tests and release notes. Things which might need to be considered more:
- there is no 'lock' on the wallets directory, which might be needed?
- because this uses a new wallets/ directory by default, downgrading to an earlier version won't see the wallets in that directory (not a big deal though, users can just copy them up to the main dir)
- jnewbery suggested putting each wallet in its own directory, which is a good idea, but out of scope for this PR IMO. EDIT: this is being done in https://github.com/bitcoin/bitcoin/pull/11687
- doc/files.md needs updating (will do soon)
I also considered including a cleanup by removing caching of data directory paths and instead just initialise them once on startup (c.f. #3073), but decided it wasn't super relevant here will just complicate review.
Tree-SHA512: c8ac04bfe9a810c32055f2c8b8fa0d535e56125ceb8d96f12447dd3538bf3e5ee992b60b1cd2173bf5f3fa023a9feab12c9963593bf27ed419df929bb413398d
1135c79 [tests] Tidy up mininode.py module (John Newbery)
f9cd9b1 [tests] Move test_framework Bitcoin primitives into separate module (John Newbery)
Pull request description:
Second part of #11518.
Moves the primitive Bitcoin datastructures and message classes into their own module, and tidies up the mininode.py module.
- First commit is almost entirely move-only
- Second commit is mostly move-only, but also does a little tidying.
Tree-SHA512: 5d74802677f1ab788e43188653106a96fffd9ab1fe3aa6a4eb94e5807de5dd5c8ee212296f45e8d16c7e3d95cfc4891677e812b7944bd3ab604e04b3b88aa06e
fafdad0d4 qa: Remove unused NodeConn members (MarcoFalke)
Pull request description:
* `ver_send` and `ver_recv` were completely unused
* `rpc` was only used once, in p2p-segwit. Imo better only pass it to the constructor in that single test
Tree-SHA512: 7f85554d6d0fd2096516ca3c608811d5370da66cde35d8031bdc921607a7a4efdb26355896012f75f713f8df09e28d46ba46be69fd96a5898fabb1a25cbcb8ad
faaa7db qa: Only allow disconnecting all NodeConns (MarcoFalke)
Pull request description:
Disconnecting the connection with `index=0` makes no sense when there are more than one connections, as the list "rotates around" and populates index 0 after `del`.
Just disconnect all NodeConns in any case.
Tree-SHA512: e5cf540823fccb31634b5a11501f54222be89862e80ccafc28bc06726480f8d2153b8c1b6f859fa6a6d087876251d48a6c6035bccdaaf16831e300bc17ff613d
748157913 [tests] Make comp test framework more debuggable (John Newbery)
Pull request description:
We should remove the comparison test framework entirely (see #10603).
Until we do that, let's make it a bit more debuggable. Currently, if there's an assert in the framework, it's very difficult to track down where we are in the test generator. Make the logging a bit better to help with debugging.
Before this PR:
```
→ ./p2p-fullblocktest.py
2017-10-09 14:05:11.302000 TestFramework (INFO): Initializing test directory /tmp/user/1000/testzdnax_yr
2017-10-09 14:05:11.557000 TestFramework.mininode (INFO): Connecting to Bitcoin Node: 127.0.0.1:11975
2017-10-09 14:05:11.712000 TestFramework.comptool (INFO): Test 1: PASS
2017-10-09 14:05:11.947000 TestFramework.comptool (INFO): Test 2: PASS
2017-10-09 14:05:12.057000 TestFramework.comptool (INFO): Test 3: PASS
2017-10-09 14:05:12.058000 TestFramework (ERROR): Assertion failed
Traceback (most recent call last):
File "/home/ubuntu/bitcoin/test/functional/test_framework/test_framework.py", line 117, in main
self.run_test()
File "./p2p-fullblocktest.py", line 72, in run_test
self.test.run()
File "/home/ubuntu/bitcoin/test/functional/test_framework/comptool.py", line 306, in run
assert test_number != 4
AssertionError
2017-10-09 14:05:12.059000 TestFramework (INFO): Stopping nodes
2017-10-09 14:05:14.203000 TestFramework (WARNING): Not cleaning up dir /tmp/user/1000/testzdnax_yr
2017-10-09 14:05:14.204000 TestFramework (ERROR): Test failed. Test logging available at /tmp/user/1000/testzdnax_yr/test_framework.log
```
With this PR:
```
→ ./p2p-fullblocktest.py
2017-10-09 14:03:54.069000 TestFramework (INFO): Initializing test directory /tmp/user/1000/testuey7t3tf
2017-10-09 14:03:54.329000 TestFramework.mininode (INFO): Connecting to Bitcoin Node: 127.0.0.1:11783
2017-10-09 14:03:54.383000 TestFramework.comptool (INFO): Running test 1: ./p2p-fullblocktest.py line 184
2017-10-09 14:03:54.496000 TestFramework.comptool (INFO): Running test 2: ./p2p-fullblocktest.py line 193
2017-10-09 14:03:54.758000 TestFramework.comptool (INFO): Running test 3: ./p2p-fullblocktest.py line 205
2017-10-09 14:03:54.867000 TestFramework.comptool (INFO): Running test 4: ./p2p-fullblocktest.py line 208
2017-10-09 14:03:54.867000 TestFramework (ERROR): Assertion failed
Traceback (most recent call last):
File "/home/ubuntu/bitcoin/test/functional/test_framework/test_framework.py", line 117, in main
self.run_test()
File "./p2p-fullblocktest.py", line 72, in run_test
self.test.run()
File "/home/ubuntu/bitcoin/test/functional/test_framework/comptool.py", line 309, in run
assert test_number != 4
AssertionError
2017-10-09 14:03:54.868000 TestFramework (INFO): Stopping nodes
2017-10-09 14:03:56.950000 TestFramework (WARNING): Not cleaning up dir /tmp/user/1000/testuey7t3tf
2017-10-09 14:03:56.950000 TestFramework (ERROR): Test failed. Test logging available at /tmp/user/1000/testuey7t3tf/test_framework.log
```
Tree-SHA512: 5525958b0098d661c281bd955c92c72bf81359464376e96d44e6c88f18aea57ba08270ecd564edda4b47d674e3b27d20e5b1060544bf8dd5c6a68103d2bb35b8
The mininode module includes code to support p2p versions below
60001. However, the test_framework does not support versions
of Bitcoin Core before V0.13.0. Remove code supporting
p2p versions before 60001 (which has never been run).
BIP31 support was added to Bitcoin Core in version 0.6.1. Our test
framework is incompatible with Bitcoin Core versions that old, so remove
all special logic for handling pre-BIP31 pings.
p2p connections can now be added to TestNode instances.
This commit also updates the example test to use the new
p2p interface in TestNode to demonstrate usage.
A future commit will update the existing tests to use p2p through the
TestNode.
fafa003 qa: Remove never used return value of sync_with_ping (MarcoFalke)
fa9de37 qa: Make tmpdir option an absolute path (MarcoFalke)
Pull request description:
This should fix issues with the multiwallet test and its symlinks
when the tmpdir is a relative path.
Rather than fixing os.symlink to work with paths relative to a
directory descriptor, which does not work on Windows, normalize
the path instead.
Tree-SHA512: 189690f3d065ea2f0f48e06775c86d513d0916c7c86312432e8e16df160e65539e288c2bd53d49a4180735fa940f6fcd52b506ccd7d9815651a9b1a69850dda6
This should fix issues with the multiwallet test and symlinks
when the tmpdir is a relative path.
Rather than fixing os.symlink to work with paths relative to a
directory descriptor, which does not work on Windows, normalize
the path instead.
4526d21 Add test for multiwallet batch RPC calls (Russell Yanofsky)
74182f2 Add missing batch rpc calls to python coverage logs (Russell Yanofsky)
505530c Add missing multiwallet rpc calls to python coverage logs (Russell Yanofsky)
9f67646 Make AuthServiceProxy._batch method usable (Russell Yanofsky)
e02007a Limit AuthServiceProxyWrapper.__getattr__ wrapping (Russell Yanofsky)
edafc71 Fix uninitialized URI in batch RPC requests (Russell Yanofsky)
Pull request description:
This fixes "Wallet file not specified" errors when making batch wallet RPC calls with more than one wallet loaded. This issue was reported by @NicolasDorier in https://github.com/bitcoin/bitcoin/issues/11257
Request URI is not used for anything except multiwallet request dispatching, so this change has no other effect.
Tree-SHA512: b3907af48a6323f864bb045ee2fa56b604188b835025ef82ba3d81673244c04228d796323cec208a676e7cd578a95ec7c7ba1e84d0158b93844d5dda8f6589b9
eff4bd8 [test] P2P functional test for certain fingerprinting protections (Jim Posen)
a2be3b6 [net] Ignore getheaders requests for very old side blocks (Jim Posen)
Pull request description:
Sending a getheaders message with an empty locator and a stop hash is a request for a single header by hash. The node will respond with headers for blocks not in the main chain as well as those in the main chain. To avoid fingerprinting, the node should, however, ignore requests for headers on side branches that are too old. This replicates the logic that currently exists for `getdata` requests for blocks.
Tree-SHA512: e04ef61e2b73945be6ec5977b3c5680b6dc3667246f8bfb67afae1ecaba900c0b49b18bbbb74869f7a37ef70b6ed99e78ebe0ea0a1569369fad9e447d720ffc4
Split off AuthServiceProxy.get_request method to make it easier to batch RPC
requests without duplicating code and remove leading underscore from _batch
method.
This does not change any existing behavior.
Change AuthServiceProxyWrapper.__getattr__ to only wrap proxied attributes, not
real attributes. This way AuthServiceProxyWrapper can continue logging RPC
calls without complicating other object usages, and special case handling for
the .url property can be dropped.
f97ab35fa qa: Fix bug introduced in p2p-segwit.py (Suhas Daftuar)
a7820422e qa: Treat mininode p2p exceptions as fatal (Suhas Daftuar)
Pull request description:
#11121 inadvertently broke the constructor for the `TestNode()` object in `p2p-segwit.py`, silently breaking at least one of the tests.
Although the python code was raising exceptions due to a `TestNode()` object not existing (or having the right type), mininode was masking these from anyone running the test through the test_runner (like travis), because it catches all exceptions during message delivery and just prints a log message and continues. Such "graceful" handling of errors is almost certainly something we don't want in our test suite, so the first commit here attempts to prevent that type of failure from ever being masked.
The second commit fixes the particular bug in `p2p-segwit.py`.
Tree-SHA512: b6646e3cb1e05c35c28e8899c44104bf2e2d0384643ca87042ab7f6ec0960d89f5bf25a7b95bab6e32d401c20a6018226160500f6ddceb923e81ffb04adb4f2f
8213838 [Qt] tolerate BIP173/bech32 addresses during input validation (Jonas Schnelli)
06eaca6 [RPC] Wallet: test importing of native witness scripts (NicolasDorier)
fd0041a Use BIP173 addresses in segwit.py test (Pieter Wuille)
e278f12 Support BIP173 in addwitnessaddress (Pieter Wuille)
c091b99 Implement BIP173 addresses and tests (Pieter Wuille)
bd355b8 Add regtest testing to base58_tests (Pieter Wuille)
6565c55 Convert base58_tests from type/payload to scriptPubKey comparison (Pieter Wuille)
8fd2267 Import Bech32 C++ reference code & tests (Pieter Wuille)
1e46ebd Implement {Encode,Decode}Destination without CBitcoinAddress (Pieter Wuille)
Pull request description:
Builds on top of #11117.
This adds support for:
* Creating BIP173 addresses for testing (through `addwitnessaddress`, though by default it still produces P2SH versions)
* Sending to BIP173 addresses (including non-v0 ones)
* Analysing BIP173 addresses (through `validateaddress`)
It includes a reformatted version of the [C++ Bech32 reference code](https://github.com/sipa/bech32/tree/master/ref/c%2B%2B) and an independent implementation of the address encoding/decoding logic (integrated with CTxDestination). All BIP173 test vectors are included.
Not included (and intended for other PRs):
* Full wallet support for SegWit (which would include automatically adding witness scripts to the wallet during automatic keypool topup, SegWit change outputs, ...) [see #11403]
* Splitting base58.cpp and tests/base58_tests.cpp up into base58-specific code, and "address encoding"-code [see #11372]
* Error locating in UI for BIP173 addresses.
Tree-SHA512: 238031185fd07f3ac873c586043970cc2db91bf7735c3c168cb33a3db39a7bda81d4891b649685bb17ef90dc63af0328e7705d8cd3e8dafd6c4d3c08fb230341
1817398b3 mininode: add an optimistic write and disable nagle (Cory Fields)
Pull request description:
Disclaimer: I'm not familiar with asyncore, so I'm unclear how safe this is. It works for me (tm).
Because the poll/select loop may pause for 100msec before actually doing a send, and we have no way to force the loop awake, try sending from the calling thread if the queue is empty.
Also, disable nagle as all sends should be either full messages or unfinished sends.
This shaves an average of ~1 minute or so off of my accumulated runtime, and 10-15 seconds off of actual runtime.
Tree-SHA512: 6b61b8058e621dacf0b4dd353c10e3666fbda0691440eb6ebc432491ebada80a781dcd09291bf03e70112a41d3c2a0c91775ed08824b79bf8d0ebed11595c28b
Because the poll/select loop may pause for 100msec before actually doing a
send, and we have no way to force the loop awake, try sending from the calling
thread if the queue is empty.
Also, disable nagle as all sends should be either full messages or unfinished
sends.
This shaves an average of ~1 minute or so off of my accumulated runtime, and
10-15 seconds off of actual runtime.
0063d2c3d [tests] Make p2p-leaktests.py more robust (John Newbery)
Pull request description:
There has been an example of p2p-leaktests.py failing on travis in the new service bits test (introduced in #11001 . It appeared to me that the previous p2p connections had not been fully disconnected before attempting to add new p2p connections.
I've added a sleep and restarted the NetworkThread, but I don't know whether this will fix the problem, since I'm unable to reproduce the failure locally.
@MarcoFalke - not sure what you want to do here? I don't think this change could make things any worse.
Tree-SHA512: f5427c26267185a903c9b75bb3925bf153b8afce70c8e493bf8f585f57d809d20643b4ee69081300b211d22e960242aecc3d719f4ddd230aa08fdc5484b55055
7148b74dc [tests] Functional tests must explicitly set num_nodes (John Newbery)
5448a1471 [tests] don't override __init__() in individual tests (John Newbery)
6cf094a02 [tests] Avoid passing around member variables in test_framework (John Newbery)
36b626867 [tests] TestNode: separate add_node from start_node (John Newbery)
be2a2ab6a [tests] fix - use rpc_timeout as rpc timeout (John Newbery)
Pull request description:
Some additional tidyups after the introduction of TestNode:
- commit 1 makes TestNode use the correct rpc timeout. This should have been included in #11077
- commit 2 separates `add_node()` from `start_node()` as originally discussed here: https://github.com/bitcoin/bitcoin/pull/10556#discussion_r121161453 with @kallewoof . The test writer no longer needs to assign to `self.nodes` when starting/stopping nodes.
- commit 3 adds a `set_test_params()` method, so individual tests don't need to override `__init__()` and call `super().__init__()`
Tree-SHA512: 0adb030623b96675b5c29e2890ce99ccd837ed05f721d0c91b35378c5ac01b6658174aac12f1f77402e1d38b61f39b3c43b4df85c96952565dde1cda05b0db84
Almost all test scripts currently need to override the __init__()
method. When they do that they need to call into super().__init__() as
the base class does some generic initialization.
This commit makes the base class __init__() call into set_test_params()
method. Individual test cases can override set_test_params() to setup
their test parameters.
Separates the act of creating a TestNode object from starting the node.
The test_framework now keeps track of its list of TestNodes, and test
writers can call start_node() and stop_node() without having to update
the self.nodes list.
c6ec4358a [tests] Add bitcoin_cli.py test script (John Newbery)
b23549f6e [tests] add TestNodeCLI class for calling bitcoin-cli for a node (John Newbery)
Pull request description:
We don't test bitcoin-cli at all. That means that we can miss inconsistencies between the bitcoin-cli client and the RPC interface, such as #10698 and #10747. It also means that the various bitcoin-cli options and features are untested and regressions could be silently introduced.
Let's fix that.
This PR adds bitcoin-cli testing in the python functional test_framework:
1. Add a bitcoin_cli.py test script that tests bitcoin-cli. At the moment it only tests that the result of `getinfo` is the same if you run it as an RPC or through bitcoin-cli, but can easily be extended to test additional bitcoin-cli features
**EDIT: `--usecli` option is moved to a separate PR. This PR now only covers the bitcoin_cli.py test.**
2. ~Add a `--usecli` option to the test framework. This changes the test to use bitcoin-cli for all RPC calls instead of using direct HTTP requests. This is somewhat experimental. It works for most tests, but there are some cases where it can't work transparently because:~
- ~the testcase is asserting on a specific error code, and bitcoin-cli returns a different error code from the direct RPC~
- ~we're sending a very large RPC request (eg `submitblock`) and it can't be serialized into a shell bitcoin-cli call.~
~I think that even though `--usecli` doesn't work on all tests, it's still a useful experimental feature. Future potential enhancements:~
- ~enhance the framework to automatically skip tests that are known to fail with bitcoin-cli if the `--usecli` option is used.~
- ~run a subset of tests in Travis with `-usecli`~
This builds on and requires the `TestNode` PR #10711 . As an aside, this is a good demonstration of how tidy it is to add additional features/interfaces now that test node logic/state is encapsulated in a TestNode class.
Addresses #10791
Tree-SHA512: a1e6be12e8e007f6f67b3d3bbcd142d835787300831eb38e6027a1ad25ca9d79c4bc99a41b19e31ee95205cba1b3b2d21a688b5909316aad70bfc2b4eb6d8a52
08ce33f8e qa: Move wait_until to util (MarcoFalke)
Pull request description:
This moves `wait_until` to `util.py` to make it generally available to python tests.
Also, `wait_until` now takes an optional lock that is acquired while testing the predicate.
Previously the lock was always acquired, even when it was not necessary, cf. `disconnect_ban.py`.
Tree-SHA512: 18e452a017a6566fa8ad09bde058e1b841e167039dc63299e70cfa7a6dcbc779581e60ca3e8eb2f1b610767d5208b9376c203eb11015b250fd0542b5eb4215a8
Fixes a couple of bugs from the introduction of TestNode:
- test scripts were no longer able to specify a custom timeout for
starting a node. Therefore tests with nodes that take a long time to
start up (eg pruning.py) would fail.
- the test for whether a node has failed on start up was broken
by changing 'assert x is None' to 'assert not x'. Since
subprocess.poll() can return None (indicating the node is still running)
or 0 (indicating the node exited with return code 0), this was a
regression.
When running the tests locally with a parallelism of 4 on an otherwise
busy system, RPC can take quite a wait to come up.
Change the timeout to 60 seconds just to be safe.
TestNode is a class responsible for all state related to a bitcoind node
under test. It stores local state, is responsible for tracking the
bitcoind process and delegates unrecognised messages to the RPC
connection.
This commit changes start_nodes and stop_nodes to start and stop the
bitcoind nodes in parallel, making test setup and teardown much faster.
cc5d38f4b Add option to attach a python debugger if test fails (John Newbery)
Pull request description:
Adds a simple option to the test_framework to attach pdb if the test fails.
Helpful for catching and debugging intermittent failures: Run the test in a loop with this option. The first failure will cause execution to pause and nodes will be left running for interactive debugging.
@sdaftuar
Tree-SHA512: 01cfae15fa3f04ed6ec6a99fef60a6c6a59723429309e81eacd6767caf12f5758f59b337804291ecab33a38a2958f36e2b513d201bee72a2eeb207a67046f952
In v0.15, we disconnect nodes that send us version messages with
unsupported service bits (1 << 5 and 1 << 7). This commit adds a test
that bitcoind will disconnect those nodes before August 1st 2018, and won't
disconnect those nodes after August 1st 2018.
This commit moves functions start_node, start_nodes, stop_node and
stop_nodes functions into the BitcoinTestFramework class. It also moves
the bitcoind_processes dict and coverage variables into BitcoinTestFramework.
176c021 [qa] Test non-atomic chainstate writes (Suhas Daftuar)
d6af06d Dont create pcoinsTip until after ReplayBlocks. (Matt Corallo)
eaca1b7 Random db flush crash simulator (Pieter Wuille)
0580ee0 Adapt memory usage estimation for flushing (Pieter Wuille)
013a56a Non-atomic flushing using the blockchain as replay journal (Pieter Wuille)
b3a279c [MOVEONLY] Move LastCommonAncestor to chain (Pieter Wuille)
Tree-SHA512: 47ccc62303f9075c44d2a914be75bd6969ff881a857a2ff1227f05ec7def6f4c71c46680c5a28cb150c814999526797dc05cf2701fde1369c06169f46eccddee
Adds new functional test, dbcrash.py, which uses -dbcrashratio to exercise the
logic for recovering from a crash during chainstate flush.
dbcrash.py is added to the extended tests, as it may take ~10 minutes to run
Use _Exit() instead of exit() for crash simulation
This eliminates stderr output such as:
terminate called without an active exception
or
Assertion failed: (!pthread_mutex_destroy(&m)), function ~recursive_mutex, file /usr/local/include/boost/thread/pthread/recursive_mutex.hpp, line 104.
Eliminating the stderr output on crash simulation allows testing with
test_runner.py, which reports a test as failed if stderr is produced.
