[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Orekit Developers] Preparation for next release


MAISONOBE Luc <luc.maisonobe@c-s.fr> a écrit :


Hank Grabowski <hank@applieddefense.com> a écrit :


I only have one file listed:

orekit-main/target/test-classes/inpop/inpop10b_TDB_summer_1969_bigendian.dat


This is very strange. The test should work here.

There is a small standalone program called DEFile in the src/tutorials part,
in package fr.cs.examples.bodies. Could you run this program with arguments:
-in /the/absolute/path/to/inpop10b_TDB_summer_1969_bigendian.dat -constant GM_Ven
This should display the value of the constant for Venus as extracted from the file header. 
The value should read: 7.243452440924042E-10.
Then this value is multiplied by au^3/86400^2 to get the desired units of m^3/s^2, which 
should match 3.2485859679756975E14, as expected by the test.
If you get the correct constant but the test fails, it may be because the au value is wrong. Unfortunately, INPOP ephemerides do not put AU in a constant (so we cannot use the program above with -constant AU as we would do in JPL epĥemerides). The AU value is extracted from the INPOP header directly (8 bytes at offset 2680 in the file...). You can get it under debugger at line 395 of JPLEphemeridesLoader.java, and it should be 1.49597870691E11 for this file (the value is really extracted at line 601 of the file, and multiplied by 1000 in the getLoadedAstronomicalUnit method). You can also check it from the command line 
(except for the 1000x factor) under linux using this:

  dd if=/the/absolute/path/to/inpop10b_TCB_summer_1969_littleendian.dat \
     bs=1 skip=2680 | od -t f -N 8 --endian=little


A simpler command is:

  od -j 2680 -t f -N 8 --endian=little \
      /the/absolute/path/to/inpop10b_TCB_summer_1969_littleendian.dat

Luc
By the way, you can use the same method to get GM_Ven value, simply replace the 2680 bytes 
offset above with 7560.
Could you also check the integrity of the binary file? It should be exactly 37520 bytes long and its sha1 sum should be 4c5506e8eac1fd3f3c9b304c0e2b5cbe027dda08. Perhaps the file did not get considered as binary and some EOL conversion occurred on git clone. I doubt about this 
as the results would probably have been completely random.

best regards,
Luc



On Mon, Dec 29, 2014 at 2:23 PM, MAISONOBE Luc <luc.maisonobe@c-s.fr> wrote:


Hi Hank,

Hank Grabowski <hank@applieddefense.com> a écrit :

PS I did a clean checkout of the repository and a clean import (using the

"Existing Maven Projects") setting in Eclipse Luna under Fedora 20 running
OpenJDK 64-bit Server VM (build 24.65-b04, mixed mode) (java -version
producing 1.7.0_71) and I have two tests failing:

DSSTPropagatorTest.testDSSTrestart() fails in certain cases.  It looks
like
they are doing performance tests for 5 consecutive iterations and
sometimes
it fails to complete in a "timely enough" manner.  Subsequent iterations
shouldn't take more than 50% longer than the first iteration, is
apparently
what the test is. If it isn't regularly happening on the test servers then
I would say it's not a big deal.  My VM is somewhat resource constrained
at
this moment so that could be what's causing the issue.



Yes, this is a known (minor) problem. I hesitate to remove the test, which
was added
to check that some long-running pre-computation are not redone every time
a propagation
is started. There is probably a better way to do it, but I did not try to
find one.
We could remove the test if it becomes too cumbersome, as it really checks
an implementation
detail.




JPLEphemeridesLoaderTest.testGMInpop.  It appears to be a just a
difference
on the order of 5e6.  since we are pulling gravitational constants I would
think that we should be able to match to far greater precision than 1e6,
much less 5e6.  Maybe my intuition is off on that though :( :



This one is new. It probably means the test does not pull the right file.
It may
be due to new ephemerides present in the test resources and the one read
first
depends on the local machine (inode time stamps and such low level OS
stuff). This
could be triggered by a new checkout as the directory crawling order may
happen in
a different order.

Could you check what is really read in the test by printing the output of
DataProvidersManager.getInstance().getLoadedDataNames() ? If there is a
JPL ephemerides
in the list in addition to the INPOP ephemerides, then the test should be
tuned
to not load JPL, regardless of the files visit order in DirectoryCrawler.
Please
do this test *without* changing your work space, as any new checkout may
lead to
a different files ordering.

best regards,
Luc


java.lang.AssertionError: expected:<3.24858597E14> but

was:<3.24858591760573E14>
at org.junit.Assert.fail(Assert.java:88)
at org.junit.Assert.failNotEquals(Assert.java:743)
at org.junit.Assert.assertEquals(Assert.java:494)
at org.junit.Assert.assertEquals(Assert.java:592)
at
org.orekit.bodies.JPLEphemeridesLoaderTest.testGMInpop(
JPLEphemeridesLoaderTest.java:114)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at
sun.reflect.NativeMethodAccessorImpl.invoke(
NativeMethodAccessorImpl.java:57)
at
sun.reflect.DelegatingMethodAccessorImpl.invoke(
DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:606)
at
org.junit.runners.model.FrameworkMethod$1.runReflectiveCall(
FrameworkMethod.java:47)
at
org.junit.internal.runners.model.ReflectiveCallable.run(
ReflectiveCallable.java:12)
at
org.junit.runners.model.FrameworkMethod.invokeExplosively(
FrameworkMethod.java:44)
at
org.junit.internal.runners.statements.InvokeMethod.
evaluate(InvokeMethod.java:17)
at
org.junit.internal.runners.statements.RunBefores.
evaluate(RunBefores.java:26)
at org.junit.runners.ParentRunner.runLeaf(ParentRunner.java:271)
at
org.junit.runners.BlockJUnit4ClassRunner.runChild(
BlockJUnit4ClassRunner.java:70)
at
org.junit.runners.BlockJUnit4ClassRunner.runChild(
BlockJUnit4ClassRunner.java:50)
at org.junit.runners.ParentRunner$3.run(ParentRunner.java:238)
at org.junit.runners.ParentRunner$1.schedule(ParentRunner.java:63)
at org.junit.runners.ParentRunner.runChildren(ParentRunner.java:236)
at org.junit.runners.ParentRunner.access$000(ParentRunner.java:53)
at org.junit.runners.ParentRunner$2.evaluate(ParentRunner.java:229)
at org.junit.runners.ParentRunner.run(ParentRunner.java:309)
at
org.eclipse.jdt.internal.junit4.runner.JUnit4TestReference.run(
JUnit4TestReference.java:50)
at
org.eclipse.jdt.internal.junit.runner.TestExecution.
run(TestExecution.java:38)
at
org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.
runTests(RemoteTestRunner.java:459)
at
org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.
runTests(RemoteTestRunner.java:675)
at
org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.
run(RemoteTestRunner.java:382)
at
org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.
main(RemoteTestRunner.java:192)


On Mon, Dec 29, 2014 at 10:36 AM, Hank Grabowski <hank@applieddefense.com



wrote:

I see there are lots of places the BC interpolator was used and is now

creating deprecation warnings.  Should I create an issue in the tracker
and
fix them all ?

On Sun, Dec 28, 2014 at 3:40 PM, MAISONOBE Luc <luc.maisonobe@c-s.fr>
wrote:



MAISONOBE Luc <luc.maisonobe@c-s.fr> a écrit :

Hi all,



Hank Grabowski <hank@applieddefense.com> a écrit :

With the new 2D interpolators in Apache Math we will probably want to





make
the change to use that for the FixedTropospheric classes.  The old
bi-cubic
interpolator could be turning seriously wrong numbers.  I have an item
on
our internal JIRA for fixing that once the new math is incorporated
and
then updating the test.  I can break off time to do that this week or
next.



Fine.

Apache Commons Math 3.4 is finally out. I have updated the dependency
in
Orekit
pom. The deprecated bi-cubic interpolator triggers warnings. I'll let
you make the
required changes.

I have also removed all old deprecated stuff, so only the "new"
deprecations introduced
by 7.0 remains. This removal also implied flattening some temporary
abstract classes
in the events hierarchy, and this in turn changed EventDetector to a
parameterized
interface -this is due to the return types in the fluent API). It's not
a big deal and
7.0 is the good time to do that.



Ooops. No, this was a bad move. I've removed it so EventDetector is not
parameterized. In fact the fluent API belong to the implementation
level,
not to the detector API level. So what have disappeared is the
intermediate
class AbstractReconfigurableDetector, which was intended to be a
temporary
hack. Its methods have been merged in the AbstractDetector class. These
methods do *not* belong to the top level interface.

Sorry for the noise.

best regards,
Luc



best regards,

Luc


On Tue, Dec 23, 2014 at 9:42 AM, Luc Maisonobe <Luc.Maisonobe@c-s.fr>

wrote:

Hi all,



We are planning to release the next version of Orekit soon. This will
be
a major release: 7.0. This version mainly includes the complete DSST
semi-analytical propagator, as the project to include short periodics
terms has been a success, and it also includes the extension to
second
order derivatives for many models (Cartesian coordinates, angular
coordinates, attitude modes, ...).

I think the current status is feature-complete. We need to update to
the
new version of Apache Commons Math (3.4 will hopefully been released
very soon now). This version brings corrections useful for Orekit. We
also need to remove old deprecated methods and classes, as
we can do on a major version release.

If you think anything else should be included for 7.0, please chime
in.
Remember that we can introduce incompatible changes in a major
release
like 7.0, but once it is out we will not be able to introduce such
changes in 7.X and will have to wait for 8.0. So if you need some
breaking new feature, now is the time to introduce it.

best regards,
Luc