Hi Roberto, On Mon, 5 Aug 2002, Roberto Bagnara wrote:
I am reiterating the following request because of the obvious reasons of functionality and efficiency. Moreover, the issue of C++ ABI compatibility (which suggested not to apply the given patch) seems now to be completely pointless:
Thank you for reminding us of this patch. I am completely in favor of your patch. Yet, the problems of ABI-compatiblilty between GCC versions are completely orthogonal to GiNaC's compatibility as defined by DSO libraries. So, it will go in for GiNaC-1.1.
GCC 2.95 is ABI-incompatible with GCC 3.0, which is ABI-incompatible with GCC 3.1, which will almost certainly be ABI-incompatible with GCC 3.3.
^^^^^^^^^^^^^^^^ Not almost certainly, GCC 3.2 is already ABI-incompatible with GCC 3.1 in the strict sense. It is, however, hoped that GCC 3.3 will be compatible with the (soon to be released) GCC 3.2. However, since the incompatibilities are AFAIK only in corner cases of the name mangling rules, this may not even affact us at all. In fact, I think that we were not at all hit by any of these incompatibilities since GCC 3.0.
Moreover, hasn't the time of GiNaC 1.1 come yet?
Hmm, it contains some changes that are fine for GCC-3.x but measurably affect performance for GCC-2.95.x. I don't know if the time has come yet. Can't you live from CVS for a little while longer? Best wishes -richy. -- Richard B. Kreckel <Richard.Kreckel@Uni-Mainz.DE> <http://wwwthep.physik.uni-mainz.de/~kreckel/>