Hello,
I've found this wonderful website and maybe I am asking too many question in too short period of time. I apologize for that.
Well, here it is my third question in two days:

I have a Router R1 with 2 Ethernet cards E0 and E1 connected to a second router R2 through S0. Second Router R2 has also two Ethernet ports E0, E1 and a serial interface (connected to R1).
R1
!
int e 0
ip address 172.12.10.0 255.255.255.0
!
int e 1
ip address 172.12.20.0 255.255.255.0
!
int s 0
ip address 172.12.30.0 255.255.255.252
!
router rip
network 172.12.0.0
R2
!
int e 0
ip address 172.12.40.0 255.255.255.0
!
int e 1
ip address 172.12.50.0 255.255.255.0
!
int s 0
ip address 172.12.30.0 255.255.255.252
!
router rip
network 172.12.0.0
!
The book says that the routing table of r1 will show knowledge of networks 172.12.40.0 and 172.12.50.0 as learned through RIP and the routing table of r2 will show knowledge of networks 172.12.10.0 and 172.12.20.0 also learned through RIP.
My question gets here:
How the routers learned about subnets 10, 20, 40 and 50 ?
It is not supposed that 10,20,40, and 50 get lost when used the classful address of class B (172.12.x.x) ?
Or the VLSM means that RIP can not adverstized different subnet marks ( and in this case S0 it is using a 30-bit subnet versus 24-bit subnets of the rest of the interfaces ?
So, If this it's the case, I can interpret Variable as Different ?
I am not sure If I ask a clear question, but I will be very happy to explain my concern in a different way, if asked to.
Thanks a lot for reading as far,
Regards,
Andres