Am 18.11.2010 15:27, schrieb Thaddy: > On 18-11-2010 13:21, Žilvinas Ledas wrote: >>> But it might be an advantage for some projects as the discussions >>> over the years implied. >> What about using GC for fpc itself? If it is usable for fpc, then the >> problem of fpc leaking memory when compilation fails with errors can >> be solved using GC. As a result fpc can be integrated to some IDEs >> without a fear or memleaks. >> Just a thought. >> >> Regards >> Zilvinas Ledas >> > It is not a very good idea to use a GC as a means to catch mistakes made > by a programmer. > These kind of mistakes can still cause memory resource related problems > or violations. > > That's not why I (re)wrote my GC interface unit (again) anyway. > > But, there is no reason why it shouldn't work.
It works, but there is a small problem: the compiler itself is a nightmare for any heap manager and it shows *THE* disadvantages of GC: it is sloooooooooooooooooow. Compiling the system unit with the default heap manager: 0.4 sec with the BoehmGC port 98.9 sec (creates a ppu, but crashes on exit) And no, FPC does not a lot huge memory allocations, just a lot of small ones and I mean really a lot :) _______________________________________________ fpc-devel maillist - fpc-devel@lists.freepascal.org http://lists.freepascal.org/mailman/listinfo/fpc-devel