From: Michael H. <ho...@sa...> - 2003-03-12 17:58:03
|
Grzegorz Jakacki <ja...@he...> writes: > On Tue, 11 Mar 2003, Michael Hohmuth wrote: >> Alexandre Tolmos <kt...@fr...> writes: >> >> > Btw, Grzegorz has got problems compiling with Gcc 3.2.1 (errors in >> > "buffer.cc" pertaining to the "placement new" operator). On my Mac I >> > only have Gcc 3.1. Did you encounter such problems with Gcc 3.2? >> >> No, I did not. Then again, so far I've only tried using Gcc-3.2 >> together with a version 6.0 of the garbage collector. > > Have you run 'make test' ? > > Most likely I did not supply you with enough info on the check-in > procedure (I posted the draft some like two (?) weeks ago, I assumed you > got it). We try to work in branches, tagged sandbox_USERID or > sandbox_USERID_SOMETHING, and test before we merge to the trunk. 'make > test' does testing. Perhaps you were not aware of it, otherwise you would > not be able to successfully run 'make test', as your GC differs from CVS > GC. > > I am very sorry about this confusion, I am planning to put such info on > the website, I am working on this now. It should be ready before the end > of this month. Dear Grzegorz, I am at loss here: I lost track of what we are discussing about. Just to clear up some confusion: Are you commenting on (1) the issue at hand (problems with placement-new operator) or (2) on the Gcc patch I committed? If (1): I have not checked in any patches related to the garbage collector. I just proposed a GC patch for the old garbage collector (which I did not check in) to you in case you would like to experiment with it. If (2): I have checked in my patch to the main trunk of the CVS, as you advised in another post. Yes, I did run "make test". Before I applied my patch, it worked for Gcc 2.95, but OpenC++ did not even compile for Gcc 3.0 and Gcc 3.2 because of the garbage collector's operator-new issue. After my patch, the situation remained the same. I am quite positive that my changes do not break anything that was not broken before I committed them. However, using my experimental GC patch mentioned above, OpenC++ compiles. The testsuite had to be adopted to ANSI C++ because of namespace issues and "void main()", and I have committed the corresponding changes now. It now seems to work with Gcc 3.0, but fails with Gcc 3.2, apparently because `occ' cannot parse a number of STL header files coming with that compiler version. Please note that `occ' currently assumes that you C++ compiler is called `g++' (or `CC' on IRIX), and there does not seem to be an option to change this. Therefore, if your compiler has a different name (for example, Gcc 3.2 is called `gcc-3.2' on Debian Linux), you need to edit the line containing "compilerName" in driver2.cc. As an aside: I subscribed to <ope...@li...>, but it looks like commit messages are not reflected to this list. Is this intended? Kind regards, Michael -- ho...@sa..., ho...@in... http://www.sax.de/~hohmuth/ |