>-----Original Message----- >From: owner-geda-user@xxxxxxxx on behalf of Samuel A. Falvo II >Sent: Tue 26/07/2005 17:30 >To: geda-user@xxxxxxxx >Subject: Re: gEDA-user: PCB GTK version... >> The problem is that GTK and Qt are over-complicated and badly designed. > >The idea is sound, but the execution is horrifyingly poor. GTK 2 in >particular is bad because they are attempting to subsume the rule of >CORBA right into GTK itself, in its attempt to be both >written-language and programming-language independent. The event >distribution model relies heavily on comparing long strings for >equality, instead of hashes or even permanently assigned 32-bit IDs. >Therefore, event notifications (itself made slow by the fact that an >IP-domain socket is used between the X server and the client >application) are made unnecessarily slow. Window refreshes are >triggered based on events. Mouse and keyboard input are based on >events. Etc. Are you sure it does this? There is a specific mechanism in X to deal with this problem: atoms. I can't believe GTK doesn't use them. Anyway, whether or not their particular system can handle GTK, people should look beyond speed. It's possible the temporary speed problem will be solved by improvements in hardware with time. The bigger problem is the design of GTK, if it continues in the vein it is in now then it will be a problem to those using it. (Though, I imagine GTK+ v9 will have something like 100K function calls and a similar number of outstanding bugs, 15GB of libs and a similar amount of docs, and will slow down my 17GHz Pentium 8). What those who maintain pcb have to do is decide whether the potential benefits outweigh the risks. (Which is not really anything to do with me, so I will now shut up).
<<winmail.dat>>