arts icon indicating copy to clipboard operation
arts copied to clipboard

Release ARTS3

Open erikssonpatrick opened this issue 6 years ago • 3 comments

erikssonpatrick avatar Sep 06 '19 09:09 erikssonpatrick

Things to fix and consider for ARTS-3 (for now draft for meeting later today):

Mandatory:

  • Remove obsolete code and features (#41, #205, #206, #209)
  • Naming and description of WSVs and WSMs (#104, #105)
  • Built-in doc of groups (#37)
  • Finish started and considered changes (#83, #157, #136, #60)
  • Remove or "activate" cfiles in /controlfiles that are not used in any test (#210, #211)
  • ARTSCAT-6 (#17)
  • Comment out parts in AUG that no longer are valid (did we decide?)
  • (Revision of WSVs, WSMs and functions for scattering data (#98, #99 and much more, but not needed if #208 gets done))
  • New version of DOIT (#207)

Would be very nice:

  • Allow spherical harmonics for scattering data (#208)
  • Target C++17 (#212)

erikssonpatrick avatar Oct 25 '19 12:10 erikssonpatrick

Patrick,

Please can you clarify Remove all obsolete code and features? What defines this? Because, to me, it seems we should first of all remove all control files that are not run or included by the test-setup. We cannot guarantee that they are working anyways. Then we would have to remove all methods that are not used in any of the remaining control files? They could be working and they could be used by others so we need an opinion on each of them. An opinion here means that someone with the opinion creates a feature test-file to allow the method to remain in ARTS. But what is next? Do we remove all inner functions of ARTS that are not used? Or do we even rearrange so that the lot of the helper-functions that don't really need a header because they are only called once has their header removed?

About Complete switch to new "radiative transfer system" #41 , I sent in some code a while back that mostly solves this. There is some problems but the difference between new and old method seems smaller than between either and the perturbed case. So there is probably something fundamental wrong in both.

riclarsson avatar Oct 28 '19 08:10 riclarsson

All: Of course, you can not edit my post. So let's go for this. Add your points in a comment and I update my initial comment. And when you think your stuff is correctly reflected in the list, remove your comment. All to not get a very long thread here.

Richard: Added ARTSCAT-6. Yes, #41 is basically closed. I just wait for a last test by Simon before closing that issue. When it comes to the control files we provide, I did not add much. The main reason is that I don't think we manage to make a proper revision of those files before March. But I agree that cfiles not used should be removed, or be included in some test. Added this as mandatory. I don't think we shall make Python mandatory, but that can be discussed. And started a new category. I mainly meant that if there are internal functions that are not used, they shall be removed. There are likely still also WSMs that are replaced by newer ones, and they shall be removed. I can not see that we can define a test for exactly every option in ARTS ...

erikssonpatrick avatar Oct 28 '19 08:10 erikssonpatrick