Hi Evan, Evan Ward <evan.ward@nrl.navy.mil> a écrit :
MAISONOBE Luc wrote:Hi Evan, Evan Ward <evan.ward@nrl.navy.mil> a écrit :I have the refactoring finished, but two test cases that use empty EOP histories are failing. What should the behavior be when the user requests a frame that uses EOP data, but none is available? Currently a null correction is used when there are less than 4 EOP entries or when there is no data for the date. Is this the intended behavior? IMHO an exception should be thrown in both cases because a frame with the requested accuracy can not be created. (the fail fast strategy)I don't agree. My rationale is that Orekit can be used in mission analysis cases, where time ranges from dates for which EOP corrections are available to dates where EOP corrections are not available yet. For such cases, the best guess is to use 0 correction. Of course, in this case the user could ask for not applying the correction, but I'm not sure forcing this him/her to do that is good.Ok. That is a little different UTCScale, which throws an exception when there is no data. I'll add a comment to FramesFactory explaining that EOP corrections are only used if they are available.
I have reviewed the changed you committed to the time-stamped caches. This is great!The first quick tests I did show a real improvement even for single-threaded applications. It is particularly interesting for EOP as it was the most important bottleneck remaining after Earth gravity field force model has been improved a few weeks ago (when Holmes-Featherstone has been introduced).
Thanks a lot. Luc
EvanLucRegards, Evan---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.
---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.