On Wed, Feb 25, 2009 at 01:46:41PM +0100, Jens Vollinga wrote:
Alexei Sheplyakov schrieb:
+1.5.1 (25 February 2009) +* Fixed compiling problems with GNU g++ 3.4. +* Fixed broken RTTI on woe32. +
Oh, what a nice joke!
what do you mean?
`Fixed broken RTTI on woe32' is a bit sloppy. It might be (mis)interpreted as if we fixed woe32's C++ runtime library itself instead of accommodating GiNaC to its limitations. Hence it sounds a little bit funny.
Should we change NEWS?
Yes. I propose the following wording: * Work around weird C++ RTTI behaviour on woe32 so GiNaC works on MinGW again.
-m4_define([ginac_lt_current], [0]) -m4_define([ginac_lt_age], [0]) +m4_define([ginac_lt_current], [1]) +m4_define([ginac_lt_age], [1]) m4_define([ginac_lt_revision], [0])
This looks incorrect. We did not add or removed any interfaces, we fixed existing ones. So we should simply increment ginac_lt_revision, and that's it.
But crc32(), crctab[], make_hash_seed() are part of the (public) interface, or not?
crc32(), make_hash_seed() are not exported via user visible headers so they are not part of public API/ABI. Also, these functions (and crctab[] variable) are static. They can't be a part of the public interface in first place. Best regards, Alexei