Hi Norbert, all, On Thu, 4 Aug 2011 20:47:43 -0500 Norbert Thiebaud <nthieb...@gmail.com> wrote:
> The va_end manpage, on linux says: > "Each invocation of va_start() must be matched by a corresponding > invocation of va_end() in the _same function_" > (emphasis is mine) > so it's not just ugly, but also 'wrong'... FWIW, I tried a clean new from the group up reimplementation of SfxItemset once in cws new_itemsets: http://hg.services.openoffice.org/cws/new_itemsets/file/924a55dab4dc/svl/inc/svl/itemset.hxx http://hg.services.openoffice.org/cws/new_itemsets/file/924a55dab4dc/svl/source/items/itemset.cxx the intention was to be able to optimize on the code (with seemed to be impossible with the bit-rotten old one) as SfxItemsets pose a major part of the load/save-instructions. I never got to integrate that cws, as its performance currently is just on par with the old implementation (IIRC, ~8% faster on Linux, but a bit slower on Windows). I still think it would be a much better base for optimization and finetuning than the current codebase, let alone be more readable (for example it is doing the same thing in half the LOC). Maybe somebody is interested in integrating the replacement Itemset? I added an EasyHack for it: https://bugs.freedesktop.org/show_bug.cgi?id=39849 Although that is a truely challenging task, it would also be most rewarding. Best, Bjoern -- https://launchpad.net/~bjoern-michaelsen _______________________________________________ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
