On Fri, Oct 24, 1997 2:45 AM, Nick van Eijndhoven <mailto:Nick@fys.ruu.nl> wrote: >Hi Valery and fellow ROOTers, >I think we are now encountering one of the many problems which lie >ahead of us in case we use commercial software products. >This incompatibility of various versions has already been a problem for >years and will never disappear since the companies clearly want to >sell as much as possible of their products. The end of the story is >always that one has to spend quite a lot of money just to keep up >with the sw. and the ROOT team has to provide lots of ROOT versions >even for a specific platform. > Heh. I find my self in a funny position here, being a Mac man by nature. However, I feel I must defend MS here. I think the reason 4.0 and 5.0 are so different, in the case of streams, is that C++ is, truely, a moving target. 4.0 had the old routine based implementation of streams, the one before STL and other stuff was pulled into the C++ definition. The new version has the "standard" (as far as the latest definition of the C++ language) implementation of streams... which is done using templates. As the definition C++ stops moving around so much things will start to settle and we will have less problems like this. I expect, btw, that gcc will break in a similar way when they get updated to the latest standard implementation. My understanding for the reason ROOT has not been upgraded to use MSVC 5.0 is actually not the ROOT developer's fault. CERN has been slow to release a new version of CERNLIB with libraries in a 5.0 compatible format. As soon as this happens, ROOT will change to use 5.0. Correct me if I'm wrong. I do not know when CERNLIB will be released with the MSVC 5.0 format libraries. I, for one, hope that ROOT continues to use MSVC. I like the IDE much better than using the unix tools. Cheers, Gordon.
This archive was generated by hypermail 2b29 : Tue Jan 04 2000 - 00:26:21 MET