Is it feasible to use random FR4 laminate to build DRO oscillator?

What problems we will see at ~10GHz and higher? In theory it is very tiny area on PCB, less than 10x10mm. Small microstrip distances leads to negligible losses even on worst FR4. So put DR, tune it position until get required peak on spectrum analyzer. How feasible is it? Why I ask this, for example in case of AIA (active integrated antenna), why make two separate boards with interconnections (one for ICs and one for AIA itself), maybe it is good idea to put DRO directly on FR4? No interconnections, no additional assembly technologies, etc. Compact slim design. For some kind wide beam radiating AIA.

λ/4 wavelength (distance which can make a difference in microwave design) represents about 7.5mm at 10 GHz.

So long as you can live with the variation of board parameters; FR4 is not a controlled micrwave substrate so will vary from batch to batch and even over a single sheet, it will work. I have seen DRO oscillators up to 10GHz built using FR4.
Peter

well, about 50 MILLION electric door openers around the world might suggest that you CAN put a 10 GHz DRO on an FR4 substrate!