calinvass said:
For example if you send a light beam towards a massive object the propagation speed will reduce if μ and ε are increasing, but the spacetime curvature will compensate for that reduction.
If you measure the speed of light or the permittivity of permeability of the vacuum locally, you'll always get the same value, c. So basically the answer is no.
So if you make up a bunch of identical atomic clocks, and a bunch of identical meter sticks, and use the atomic clocks to locally measure the time it takes for the light to transverse the meter sticks, you'll always measure the same value. Similar things would happen if you measure the premittivity or permeability of free space, though I'm not quite sure of what physical instruments you'd use to callbirate them. It's easier to talk about clocks and meter-sticks, they are more fundamental.
You would also observe that these identical atomic clocks you create don't stay synchronized with each other when you compare their readings using light signals. Sometimes people get caught up in the idea that there is (or should be) a fundamentally Euclidean structure for space-time, when they do this they impute the disagreements in the behavior of clocks with gravity to the clocks being affected by gravity, and they imagine that there is some non-observable and non-physical "master clock" that measures some non-observable, non-physical, "absolute time", and that if you measure time acording to this non-physical time, you can keep the Euclidean structure.
Actually, this doesn't actually work exactly right, but it can be made to work in a way that's almost right, as an approxmiation that can be useful. So people sometimes use this point of view. It's not really a problem as long as one realize that it's not-quite-right approximation. Unfortunately, sometimes people take it too seriously, as something that's not just an approxomation. To really learn GR in a way that's not just an approxmiation , one ultimately needs to abandon Euclidean geometry.
There are several issues that I've observed over time that serve as stumbling blocks. One issue is an emotional attachment to Euclidean geometry that can inhibit learning about different geometries, a conviction that it's the way things "should work". One can first learn about non-Euclidean geometries in the simpler concept of special relativity - which doesn't include gravity. Related issues are a lack of desire or difficultin in learning special relativity, and to jump right into general relativity. The mathematical issues in learning special relativity are much lower than they are for learning GR, though there is still the emotional obstacle of "letting go" of Euclidean geometry. (The sort of geometry that special relativity uses is called Lorentian geometry.) There are also treatments of SR that don't stress the geometry, which can be useful too. Though it turns out that there are some similar emotional difficulties in "unlearning" some familiar notions to learn these other approaches to SR as well, the stickiest issues are learning a concept of time that's compatible with special relativity, which basically rules out any sort of "absolute time", which just isn't compatible with either SR or GR at a fundamental level.
Whereas the geometrical approach to SR stresses the Lorentzian geometry, a typical non-geometrical approach might talk at length about the Lorentz transform, without casting the Lorentz transform into a geometrical mold.)
So my general recommendation is for people to study special relativity before getting too involved in trying to figure out GR. I'd say I'd have a limited amount of success - it's easier to tell someone to learn SR than it is to do it. Still, it's a task that's much easier and more likely to be benificial, to learn a bit about special relativity, than it is to jump right into the mathematical abstractions of differential, Riemannian and/or pseudo-Riemannian geometries.