I think the NLP Solver really needs to be ported to native code. It reduces Java code, which seems to be a sort of mid-term goal for LO. It also performs some of the most performance intensive operations in the whole of LO, so it seems we'd wand this to run closer to the metal. Microsoft Excel of course has non-linear algorithms fully integrated. PS - I wasn't sure whether to submit this to Spreadsheet or Extensions. Please move it if it's not in the correct section.
Thank you for reporting this enhancement request! I can confirm that this is a valid enhancement request on: Version: 4.3.0.0.alpha0+ Build ID: 84862db95a5e22b9ef89baa2a8a5deeffefbdef6 Date: Thu Apr 24 21:43:16 2014 +0300 Platform: Ubuntu Linux 13.10 x64 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + As I've been able to confirm the enhancement request I am marking as: New (confirmed) Enhancement Low - not going to benefit most users but would be useful for advanced capabilities + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + LibreOffice is powered by a team of volunteers, every bug is confirmed (triaged) by human beings who mostly give their time for free. We invite you to join our triaging by checking out this link: https://wiki.documentfoundation.org/QA/BugTriage There are also other ways to get involved including with marketing, UX, documentation, and of course developing - http://www.libreoffice.org/get-help/mailing-lists/. Lastly, good bug reports help tremendously in making the process go smoother, please always provide reproducible steps (even if it seems easy) and attach any and all relevant material
Are there any performance measurements about NLPSolver efficiency? I am not sure that rewriting from Java to C++ will provide dramatically speed-up.
(In reply to Todor Balabanov from comment #2) > Are there any performance measurements about NLPSolver efficiency? > > I am not sure that rewriting from Java to C++ will provide dramatically > speed-up. The issue isn't only/mainly concerning efficiency. There is tdf#38948 e.g open since 2011, and requests to get a way to save solver settings to documnents are posted once in a while in an ask/discourse question. See https://ask.libreoffice.org/t/is-there-any-way-to-save-a-solver-model-in-calc-again/77249/8 . I don't remember one of those posts complaining about lack of efficiency. It's about the workflow. It seems the NLP solvers are insufficiently embedded via the current API: set-/get methods for stuctured(object properties don't exist, and the PropertyBag service also doesn't work properly. I would assume this is in consequence of a broken bridge to the implementation in java. If the mentioned problems were fixed, any interested user could easily(?) write a workaround in Basig e.g.