My bos told me that you can't create a loop with glass fibre cables is that true? Why is it different from copper?

What

Copper has flux inhibitors, glass has optic inhibitors. However, the optic inhibitors reach a critical phase mass at a rate that would overload the photo diode receptors. Thus causing a temporal displacement event horizon. This bleeds over from sending to receiver affecting the phase distribution in a negative way. When you mix receiver and sender you are essentially crossing the streams. Crossing the streams is bad.

Is my English so bad or does my question make no sense?

@Gablern At least the second. What is a "loop" if cables for you? A single cable, made to a circle? Well you can do it with fibre, but I have no idea why would do that with either fibre or copper, besides some BUS systems.

@sbiewald I have seen networking system implemented in loops with fiber. You have to filter out packets to keep them from propagating. No clue as to how that is actually done.

Loop - Your Switch has no spanning tree and you put one outgoing cable back in so that the same frames are resended every time until your network break down

@Gablern Ah, we are talking a about Ethernet.
Not sure why accidentally building loops should not work with fibre.

Assuming the protocol is Ethernet in both cases, the same "attack" should work for both types, as switches should handle fibre and copper the same. After encoding/decoding the electrical or light signal and assembling the frame, all the switch sees is an Ethernet packet.
In doubt, ask your boss if he has a spare switch so you can try it out (as modern ones have (R)STP on by default, make sure you disable it on all ports).

@Demolishun Assuming Ethernet on your case as well: (Rapid) Spanning Tree Protocol ((R)STP).
Broad simplification: Your switches send packets on every port regularly. If they receive their own message back, a loop is detected. Additionally, a tree structure is build from the packets, which defines than a unique path in the network, where everyone can reach everyone else without circulating packets. If some connections break, this is detected as well (STP packets are going missing) and the tree is (partially) rebuild.
STP vs. RSTP: The latter one is faster and the network isn't down for 30 seconds while the "root" has to be changed.

Strictly physically speaking, you can make a loop with copper because you can attach some sort of T joint. You can inject electricity into a wire from the side of a wire.

With fibre however, you can only couple in the light at the head front of the fibre, i.e. more or less in a 90 degree angle. Now if you put one head of the fibre to the head of the other end so that you get a loop, there is no head surface where to couple in the light.

Of course you can have any network topology with any transmission medium.

@Demolishun optic inhibitors over the flex horizon is overcome with prism switches at the input terminus in luminous crossover. It is a simple application of Mirozaki’s propagation conservation law. I am confident that phase mass displacement is part of the majority of COTS flex panduit systems.

@irene "Mirozaki’s propagation conservation law", but this law is only application to a quasi temporal intransigent scope. Don't forget the jump to 4th dimensional transvergence in this case. Otherwise I would agree, at least if were talking euclidean space.

@Demolishun Of course we are talking Euclidean space. Non Euclidean conditions can only exist in inorganic transducers and that isn’t practical because of the thermal conditions required to slow the photonic conversion rate. It has only been demonstrated under very specific circumstances and we aren’t sure if it is even real.

If we solved the critical phase mass problem, the research for it would be immediately be snapped up by some giant company and shelved so it can’t cannibalize the optic business model that everyone has been sinking money into for the last decade.