That BIP number is just advising them about the use of a version bit, in order to try and ensure that all four full-node Core-based implementations (if and when they all use Xthin) are in sync for this feature.
The 1MB is playing havoc with the mempool, and the amount of orphans are degrading Xthin performance, so
@Peter Tschipper and his helpful collaborators have made several improvements to the first experimental release.
Peter's latest tests are really impressive;
BlockSize XthinSize Bloom Compression
999979 19829 15159 50.43
934561 30866 14712 30.28
998139 23249 11948 42.93
989012 19335 8651 51.15
999951 15367 7280 65.07
988073 20435 8790 48.35
998045 22852 10925 43.67
998156 32325 8283 30.88
998112 17673 7421 56.48
973004 12210 5644 79.69
875175 17048 10101 51.34
999974 21909 7454 45.64
999938 21167 8194 47.24
716767 15091 5769 47.50
999962 34266 10703 29.18
998198 30822 10255 32.39
933360 24900 7150 37.48
998126 21518 12802 46.39
998176 25771 9648 38.73
998049 21104 9833 47.29
998199 23198 8815 43.03
748285 6013 5905 124.44
998192 17312 8314 57.66
934524 34075 12118 27.43
934405 30348 12022 30.79
998042 20041 8518 49.80
933236 21637 8496 43.13
929946 21307 6466 43.65
469112 17118 5784 27.40
998181 26054 14887 38.31
934500 20832 15540 44.86
997994 27173 13355 36.73
989194 24960 10125 39.63
997877 23086 8446 43.22
999915 11636 6787 85.93
843524 10365 5867 81.38
920511 14931 7807 61.65
934569 19239 7796 48.58
813371 3110 4930 261.53
989118 16995 7538 58.20
998139 200964 7047 4.97
605660 30801 6290 19.66
989062 25478 9067 38.82
998149 25152 14452 39.68
998188 29112 14556 34.29
999833 32189 10830 31.06
999922 15286 7792 65.41
998192 25065 12278 39.82
999830 16222 8974 61.63
num blocks 49 total bytes 46344427 xthin bytes 1237436
avg full size:945804 avg xthin size:25253
weighted comp ratio:37.45 unweighted comp ratio:51.12
avg bloom filter size: 9418
The X-relay nodes deployment work within the great firewall is ongoing.
