Hi, could you wait with that a two days, please? One might be able to patch TStreamerInfo such that we get a workaround for the compiler bug without degrading its performance on any platform (opt on even for .net). I'll check over the weekend. Cheers, Axel. Fons Rademakers wrote: > I'll try to check for the VC++ compiler level and make the rule only > when that compiler is used (we do that already for some other > compilers). > > Cheers, Fons. > > > On Fri, 2003-01-31 at 17:05, Nick van Eijndhoven wrote: > >>Hi Rene, >>Degrading the TStreamerInfo opt level for windows in general >>just because of bugs in the .net compiler is not a good idea >>to my opinion. >>I would suggest that .net users just build their own >>root binaries and use lower opt level as long as the bugs are >>there. Once the .net compiler is in good shape, one could think >>of providing also the binaries for the .net compiler using the >>standard opt level. >> >> Cheers, >> Nick. >> >>Rene Brun wrote: >> >>>Hi Matt, >>> >>>Thanks for sending your changes to the Makefile. I am surprised >>>to see so many bugs in the MS .NET compiler. >>> >>>I imported most of your minor changes to the Makefile for the cleanup process. >>>I also took your changes to the Makefile.win32 and incorporated the same >>>mods in Makefile.win32gdk. >>> >>>I did not take your patch for TTreeFormulaManager.h. I do not want >>>to mess a class because of a MS compiler bug. I changed the destructor >>>to be public instead for all cases. (a pity) >>> >>>I did not take your mods to degrade the optimisation level when compiling >>>TStreamerInfo. Compiling this class with optimisation is very important. >>>We may introduce a case for Windows in general to compile this class >>>with no optmisation. To be discussed. >>> >>>Now in CVS. >>> >>>We do not intend to provide binaries in the short term (next few weeks) >>>for .NET. I encourage Axel in his suggestion to see if binaries compiled >>>under VC++6 work also in the .NET environment. >>> >>>Rene Brun >>> >>>Axel Naumann wrote: >>> >>>>Hi, >>>> >>>>Do we need a dedicated .net binary? Or is it sufficient to have the code >>>>patched such that it builds under .net and only have a vc6 binary >>>>version? I believe we should first test whether the root vc6 binaries >>>>work with .net (e.g. aclic needs to be tested); I don't know of any a >>>>priori reason why that shouldn't work. Can someone do that? (I'd >>>>volunteer to do this over the weekend, incorporating Matt's diff into >>>>current cvs sources, but maybe someone else is faster.) >>>> >>>>Cheers, Axel. >>>> >>>>GENTIT Francois-Xavier DAPNIA wrote: >>>> >>>>> Dear members of the ROOT team, >>>>> I am waiting upon ROOT being distributed for Visual C++.net to install it >>>>>on my PC. I see that it works, seing the message on ROOTTALK by Matthew D. >>>>>Langston. Do you intend to make a download version of ROOT for Visual >>>>>C++.net? >>>>> >>>>>François-Xavier Gentit >>>>>DAPNIA/SPP CEA Saclay >>>>>http://gentit.home.cern.ch/gentit/ >>>>> >>>>> >>>>> >>>>
This archive was generated by hypermail 2b29 : Thu Jan 01 2004 - 17:50:09 MET