asciilifeform: which is to see what happens to p2,q2 when you make 4096+ bit moduli.

asciilifeform: at any rate, it is now possible to do the thing i ACTUALLY SET OUT to do

asciilifeform: i hesitate to use word 'correct' in relation to something gpg does...

asciilifeform: as crappable into, e.g., python, to arithmetize on.

asciilifeform: pgpdump -i displays the actual int.

asciilifeform: which, fortunately, we do not need to care about

asciilifeform: and so it allocates chunks on the heap

asciilifeform: mircea_popescu: mpi was written by braindamaged folk who believed that bigint ought to be able to 'grow infinitely'

asciilifeform: which transforms to an actual sequence of meaningful bits of the bignum via a gnarly process.

asciilifeform: mircea_popescu: not stack. the raw mpi turd.

asciilifeform: i'ma never live this down, will i.

asciilifeform: ^ the corrected patch.

asciilifeform: mpidump shows the thing you end up seeing in pgpdump -i.

asciilifeform: and padding at the end of each chunk.

asciilifeform: that does not correspond to an integer, as such, it has pieces of it, chunks, each having a header

asciilifeform: mircea_popescu this is still a raw mpi hexdump

asciilifeform: ( aaaaaanyway my ORIGINAL aim when i wrote the tester, was to look at what happens at 4096 and above. which i haven't even done yet ..! )

asciilifeform: 2 ain't 32 tho.

asciilifeform: but we knew this 20yrs ago.

asciilifeform: and probably ought to go to bed

asciilifeform: ok turns out that nobody can count x2.

asciilifeform: mircea_popescu: and what did the p and q in the privkey look like ?

asciilifeform: the output of 'YieldPrime', per my model, ought to end up in http://btcbase.org/log/2016-08-18#1524021 ☝︎

asciilifeform: so far i am at a loss as to how one becomes the other.

asciilifeform: how the fuck these turn into the sane-looking p's and q's.

asciilifeform: now for the $64k question:

asciilifeform: which is why i post the actual in and out, instead of contents of my head.

asciilifeform: anyway the patch ^ shown above is the ticket.

asciilifeform: did i push that button??

asciilifeform: and what did mircea_popescu use for a compiler, that ate this.

asciilifeform: dafuq would an unintegral number of bits refer to ??

asciilifeform: i'ma let one of you fellers tell me when i wake up why 11 00 00 00 10 00 00 00 00 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00 also didn't happen

asciilifeform: see folks, this is why i didn't qntra this yet, dun wanna pons&fleischmann it.

asciilifeform: ought to be log_hexdump("\nYieldPrime: \n", ptest, nbits/8.);

asciilifeform: not one of us, it turns out, can count

asciilifeform: let's try a small variation on the theme.

asciilifeform: dun forget to glue a saf to an ety first.

asciilifeform: which brings us to here and now.

asciilifeform: AAAAAND this is where we dump the buffer.

asciilifeform: (apologies for l0g bloat, but this has to be done..)

asciilifeform: the requested keysize we set the 2 high order bits */

asciilifeform: doing that for RSA, to make sure that the modulus does have

asciilifeform: /* Set high order bit to 1, set low order bit to 0.

asciilifeform: which then gets, for good shitgnomatic measure,

asciilifeform: i'll walk through here:

asciilifeform: thing is, nextprime cannot walk all that far

asciilifeform: ... it looks as if there is an extra layer of shuffle not accounted for by the dump?

asciilifeform: mircea_popescu: this is useless

asciilifeform: so this thing parallelizes 'embarrasingly'.

asciilifeform: note that this is an upper bound, it applies regardless of what kind of magic is used.

asciilifeform: is this obvious or do i need to draw picture.

asciilifeform: if we know B - k shared topmost bits, then the work required to break in comparison with work W, supposing we knew B bits, is at most W*(2^k).

asciilifeform: btw, here is a handy elementary proof of a certain thing,

asciilifeform: not necessarily top, or bottom of prime.

asciilifeform: the other thing,

asciilifeform: and all of p1,q1,p2,q2 share at least top 20 bit.

asciilifeform: but chances are that p1 and q1 share top 32 bit

asciilifeform: nothing odd, of course, to the naked aye.

asciilifeform: http://wotpaste.cascadianhacker.com/pastes/f6355da8-4a82-40d2-a6a0-d1217232dbac/?raw=true << this, for instance, is mircea_popescu's pub.

asciilifeform: does mircea_popescu fully apprehend where this train is going ?

asciilifeform: this is promising to be a very awkward occasion.

asciilifeform: the diddled rng being the first.

asciilifeform: soooo the subkey idiocy is apparently the 2nd half of a bipartite poison.