FWD » Core Development » Base Language

NativeInvoker has lots of cases like cls.equals(decimal.class) - if not solved by #9060, the entire class needs to be reviewed, and at least the signature needs to be normalized in inboundProcessing; is not just that BaseDataTypeConstant, also anon types of the BDT - see ControlFlowOps.calculateType.

Alexandru Lungu wrote:

I added it as a check next to integer.class (i.e. integer.integerConstant.class) and it worked, but I wonder if:

• #9060 already fixes this

9060b does not fix this particular issue.

• there are other places with such issue

Class checking is highly used in the entire project, a lot of places were completed with checking against the constant types, but surely there are more cases like this that will pop up from time to time.

A simple check against constant types should be fine here.

The library_calls/ tests in Testcases should be executed. These are not yet reworked as ABLUnit tests. See the library_calls/readme.txt.

Artur, that is not what I had in mind. This class uses lots of .equals(Class<BDT>), and the solution needs to be agnostic of the actual type.

See ControlFlowOps.getSignature - this is the source of signInput, from which signature is calculated.

Artur Școlnic wrote:

I committed a new solution to 10004a/15917. Constantin, please review.

Why not do this in ControlFlowOps.getSignature ? A Java array AFAIK we decided that the elements can never be constants.

You mean storing the super class for anonymous classes and constants in the signature?
If getSignature is the only source for the input classes in NativeInvoker, the yes, this is better.

Artur Școlnic wrote:

You mean storing the super class for anonymous classes and constants in the signature?
If getSignature is the only source for the input classes in NativeInvoker, the yes, this is better.

Yes, but look also at Call.invoke:1692, there also calculates the signInput array, but it does not resolve anon or constant types of args. Or you can just prepare signInput in the two NativeInvoker.invoke methods.

• type.equals(BaseDataTypeConstant.class) needs to be BaseDataTypeConstant.class.isAssignableFrom(type)

Otherwise, please do run #10004-5 tests.

I have some issues running library_calls tests, is there a customer test suite that would be appropriate for testing?

Artur Școlnic wrote:

I have some issues running library_calls tests,

We need to run those. What problems do you have?

./junit.sh -c library_calls.TestRunner

Could not find interface for group com.goldencode.p2j.util.osresource.EnvironmentAccessor
Could not find interface for group com.goldencode.p2j.util.osresource.EnvironmentAccessor
UI Theme successfully changed to 'windows10'
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.util.osresource.EnvironmentAccessor
Could not find interface for group com.goldencode.p2j.util.osresource.EnvironmentAccessor
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.ui.ClientExports
Could not find interface for group com.goldencode.p2j.util.LegacyLogManagerClientService
Could not find interface for group com.goldencode.p2j.util.LegacyLogManagerClientService
Could not find interface for group com.goldencode.p2j.util.LegacyLogManagerClientService
Could not find interface for group com.goldencode.p2j.util.LegacyLogManagerClientService
Could not find interface for group com.goldencode.p2j.util.LegacyLogManagerClientService
Could not find interface for group com.goldencode.p2j.util.LegacyLogManagerClientService

Do you have some other p2j.jar somewhere? This should not happen.

Constantin Asofiei wrote:

Do you have some other p2j.jar somewhere?

Just one in the deploy\lib

Artur Școlnic wrote:

I have some issues running library_calls tests, is there a customer test suite that would be appropriate for testing?

No, this test suite needs to pass.

Greg Shah wrote:

The library_calls/ tests in Testcases should be executed. These are not yet reworked as ABLUnit tests. See the library_calls/readme.txt.

I see that there is an extra step of building and installing a shared library using testapi directory (which I did), is there something else that needs to be done in order to run the converted tests with junit?

They are not migrated with JUnit/ABLUnit yet. You have to manually run a ChUI client to execute the test "driver" program. Then you have to compare the output log to the baseline output log to confirm that it is OK.

In the future, we will migrate the test suite to ABLUnit.

This explains my issues running the tests as junit :)
I compared the output binary files from trunk and 10004a, they are the same.

Please compare it with the captured baseline. We might not hold up the branch for any regressions but we should know about them if they exist.

What is the % Done?

There are 153 more files in the baseline and 3 files differ compared to 10004a, which is the same as trunk. I suspect that the missing files are due to a configuration issue on my part. Do we expect the output to be exactly the same as in linux_64bit_p2j_library_calls_output_20140212.zip?

Please see #10193-17 for an related update.

Given that 10193a reached trunk, can we proceed with testing with this?

I tested library_calls and harness+webUI navigation, all passed.

10004a was merged as rev 16105.

Even if I fix this, I get another error. I think it hints that there is something broken in library_calls test-suite.
I try to run EVP_sha256 from libcrypto with an output parameter of type memptr which I expect to be loaded.
However, there is an null/empty pointer returned. This makes the program fail once the memptr that is expected to be loaded is used, resulting in DLL procedure H... using an uninitialised MEMPTR.
Artur, can you rerun the library_calls suite?

I think this is related to #7254

PS: I wonder if #10193 was actually negating the changes in #7254, so we are back with the problem from #7254.

Alexandru Lungu wrote:

Artur, can you rerun the library_calls suite?

I recall having some issues running this suite, but with the latest trunk the issues are much greater, the output files are not generated at all.
The errors in the server.log:

25/12/08 16:25:36.623+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | ** Could not open Dynamic Library: library_calls/testapi/libtestapi.so (8013)
25/12/08 16:25:36.651+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | ** DLL Error : library_calls/testapi/libtestapi.so: cannot open shared object file: No such file or directory (8014)
25/12/08 16:25:36.665+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | ** Could not load DLL procedure library_calls/testapi/libtestapi.so. (3258)
25/12/08 16:25:39.194+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | Put/Get functions not allowed on uninitialized memory. (2914)
25/12/08 16:25:39.211+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | Put/Get functions not allowed on uninitialized memory. (2914)
25/12/08 16:25:40.669+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | Indeterminate extent dimension 0 is out of range 1 to 28000. (11385)
25/12/08 16:25:40.679+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | Indeterminate extent dimension 0 is out of range 1 to 28000. (11385)
25/12/08 16:25:41.133+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | Indeterminate extent dimension 0 is out of range 1 to 28000. (11385)
25/12/08 16:25:41.145+0200 |  SEVERE | com.goldencode.p2j.util.ErrorManager | ThreadName:Conversation [00000001:bogus], Session:00000001, ThreadId:00000005, User:bogus | Indeterminate extent dimension 0 is out of range 1 to 28000. (11385)