Maurizio Lotauro wrote: > Scrive Arno Garrels <arno.garr...@gmx.de>: > >> Maurizio Lotauro wrote: >>> Scrive Arno Garrels <arno.garr...@gmx.de>: > > [...] > >>> Cardinal and Integer should not change since (IIRC) NativeInt and >>> NativeUInt was introduced to handle 32/64 bit cpu. >> >> That's correct Integer will remain 32 bit in x64 however might change >> in the far future and Longint most likely never changes its size even >> in x128. > > I don't think that this will happen,
So I do. > otherwise the introduction of > NativeIn and NativeUInt make no sense. Now this two should follow the > cpu and OS "size". These types will always have the size of a Pointer. > > These are two interested blog from Marco Cantu: > > http://blog.marcocantu.com/blog/Delphi_64_bit_sneak_preview.html > http://blog.marcocantu.com/blog/getting_ready_delphi_64.html > > Notice the changes for Windows messages too. There hasn't much changed, WPARAM and LPARAM will be Pointer-sized integers, Int64 in Win64 and Longint in Win32, so sending Pointers with messages is just a matter of casting them properly, i.e. WParam(MyPointer) etc.. BTW: ICS should be already well prepared for Win64. > > [...] > >>> And why don't invert the test? >>> >>> if FClientNum) < MaxLongint then >>> Inc(FClientNum) >>> else >>> FClientNum := 1; >> >> I don't know why there is/was the ">=" comparison, I guess that Angus >> wanted to build-in some kind of strange fault-tolerance? However >> provided that FClientNum is only ever incremented in >> TCustomWSocketServer.TriggerSessionAvailable >> (which it realy should) you are perfectly right here as well. > > Another think that I noticed. Since the TWSocketClient.CliId is very > imprtant to correct handle the client, why it is writable? Good question, IMO it should be a read-only property. -- Arno Garrels -- To unsubscribe or change your settings for TWSocket mailing list please goto http://lists.elists.org/cgi-bin/mailman/listinfo/twsocket Visit our website at http://www.overbyte.be
