If the change is made, a good first step is a spread sheet with test cases.
So you have a proposed list of functions that would be changed? Would want too compare expected group actual and old to new. Would want to devise test cases against both common and edge cases. Also curios about time impact, does it take more time or less. Sent from my Samsung Epic™ 4G Pedro Giffuni <p...@apache.org> wrote: Hi Regina; ----- Messaggio originale ----- > Da: Regina Henschel ... > > I have no solution for this problem. > I see.. In theory I am rather busy but I find time to do two or three things at a time ;) There is no need to hurry this type of things for 4.0. >> >> BTW, I am considering doing something drastic there, like replacing all the > probablilty >> distributions with with boost implementations. Would there be any good > reason to >> avoid such approach? > > Such solution had been considered already, but at that time the BOOST > version, > which was used in OOo, was too old. Now it might be possible, but I cannot > judge > it and do not know how to do it actually. It is surely nothing, that can be > done > till AOO4.0. You need someone to implement it, preferably in a branch, and at > least one other person with knowledge in distributions to do QA. It should > not > be decided between holidays, when a lot of persons are on vacation. > Sounds reasonable. I did some very basic stuff and it looks good so far, but TBH, the difference between using the boost versions and the native versions is basically unnoticeable (which is to be expected). Pedro.
