I posted this message a long ago. Please review the benchmark results and tell me whether this overhead is worth or is too much.
Kornél ----- Original Message -----
From: Miguel de IcazaFrom: Ben MaurerI want to know the real impact (we have the tools, it should take a few minutes at most to find out the numbers).The best way for Kornel to do that (just as an FYI since he has probably never used the tools) is to run: mono --stats foo.exe mono --profile foo.exeYou are right I have never used these tools.:) I tested "internal static string GetText (string message)" without INSIDE_CORLIB. It is a very common code. getTextLock and SafeGetText are required anyway. GetResource is not changed significantly. I tested GetTextResourceManager without resource files (just used the transparent GetTextResourceSet) because the current ResourceManager and ResourceSet implementations are not currently effective enough and there are no localized resources yet. And of course localizing resources from files will result in worse performance anyway but it depends on ResourceManager and not Locale. I attached the source files and the results: mcs /out:OldLocaleBenchmark.exe LocaleBenchmark.cs OldLocale.cs mcs /out:NewLocaleBenchmark.exe LocaleBenchmark.cs NewLocale.cs mono --stats OldLocaleBenchmark.exe >OldLocaleBenchmark.stats mono --profile OldLocaleBenchmark.exe >OldLocaleBenchmark.profile mono --stats NewLocaleBenchmark.exe >NewLocaleBenchmark.stats mono --profile NewLocaleBenchmark.exe >NewLocaleBenchmark.profile I have runned this on Windows. Kornél
LocaleBenchmark.tar.gz
Description: GNU Zip compressed data
_______________________________________________ Mono-devel-list mailing list Mono-devel-list@lists.ximian.com http://lists.ximian.com/mailman/listinfo/mono-devel-list
