--- Log opened Thu Sep 09 00:00:02 2021
00:58 -!- enick_504 [~afilini-m@2001:bc8:1828:245::2] has joined ##miniscript
10:52 < andytoshi> right, i understand that
10:53 < andytoshi> but why distinguish between initial stack and "during execution" stack
10:53 < andytoshi> if we hit 1000 we hit 1000
10:54 -!- meshcollider [meshcollid@user/meshcollider] has quit [Remote host closed the connection]
11:14 < sanket1729> andytoshi: does this answer your question? https://github.com/rust-bitcoin/rust-miniscript/pull/255#discussion_r704556990
11:14 < sanket1729> I think it is much easier to implement that way
11:18 -!- meshcollider [meshcollid@meshcollider.jujube.ircnow.org] has joined ##miniscript
13:20 < andytoshi> sanket1729: sounds like you actually want to distinguish between the max stack elements when A is dissatisfied vs satisified
13:20 < andytoshi> rather than initial/executing stack elements
13:20 < sanket1729> both
13:21 < sanket1729> I don't how we can easily track only the executing elements without tracking the initial elements separately
13:24 < sanket1729> Oh, and on the recursion related parsing. I think we should just go back to `MAX_RECURSION_DEPTH` way of doing things. I implemented non-recursive parsing for miniscripts from string.
13:24 < sanket1729> But it's useless, because we cannot any methods on it (all of them are recursive)
13:25 < sanket1729> And we would end rewriting all of the library/analysis in non-recursive form
13:29 < andytoshi> why do we need to track "only the executing elements"
13:29 < andytoshi> sanket1729: lol, gross. i mean, we should eventually rewrite the library in non-recursive form..
13:29 < andytoshi> but you're right, we should finish with parsing
13:29 < andytoshi> not start with it :)
13:30 < andytoshi> in any case, no need to do that now
13:30 < sanket1729> Do you think it is worth it
13:30 < andytoshi> yes, but not now
13:31 < sanket1729> I think 1) it makes the code super ugly 2) difficult to review 3) and functions like satisfy would be annoying to implement non-recursive
13:31 < andytoshi> hmm, yeah, fair
13:31 < andytoshi> i think though, in theory we could wind up with super bizarre constructs like 500-level-deep taproots
13:32 < andytoshi> i guess not, that would be really pathological
13:32 < sanket1729> Yeah, it is possible. I have an example in my local branch
13:32 < sanket1729> which we can parse
13:32 < sanket1729> but cannot call any functoins
13:32 < andytoshi> right, it's possible, but is it ever useful?
13:32 < andytoshi> i don't think it is
13:32 < andytoshi> so, let's keep the library recursive for now
13:33 < sanket1729> It's not useful.
13:34 < sanket1729> back to execution stack size
13:34 < andytoshi> lol, well, imagine i have 500 different pubkeys, that are strictly decreasing in the probability that i'd use each one
13:34 < andytoshi> ok
13:34 < andytoshi> sorry
13:34 < sanket1729> We need to track total stack depth
13:35 < sanket1729> But I don't see how we can do that without separately tracking initial stack(witness) and increase in stack depth
13:36 < sanket1729> max_execution_limit_parents is not the same max over all children
13:37 < sanket1729> Slightly updated the comment
13:37 < sanket1729> https://github.com/rust-bitcoin/rust-miniscript/pull/255#discussion_r704556990
13:38 < andytoshi> ah ok, i see what you are saying
13:38 < andytoshi> yep ok, what you're doing is good
13:38 < andytoshi> sorry it took me so long
13:40 < sanket1729> So, should I keep the commit that changes the parsing to non-recursive/or drop it?
13:40 < andytoshi> i think we should drop it
13:40 < andytoshi> it'll just lead to a zillion bug reports about assertation failures
13:40 < sanket1729> ^^ I also got the implementation wrong the first time because of it
13:42 < andytoshi> heh oops, i'm surprised that i missed that
13:43 < sanket1729> So, we cannot dynamically query the stack size right?
13:43 < sanket1729> as in run time
13:45 < andytoshi> correct
13:45 < andytoshi> i don't know why ... LLVM provides this info i'm pretty sure
13:45 < andytoshi> actually maybe not, i recall some ancient blog posts about the elaborate things that rustc does to detect stack smashing
13:45 < andytoshi> which LLVM provides zero help on
13:50 < sanket1729> Yep, looks like people have written custom crates for it. https://crates.io/crates/stacker 
13:52 < sanket1729> dropped the commit and updated the PR
14:10 < andytoshi> awesome, thanks
14:10 < andytoshi> probably can't review til tomorrow, my brain is a bit full today
14:25 < sanket1729> cool, no rusth
16:39 -!- meshcollider [meshcollid@meshcollider.jujube.ircnow.org] has quit [Changing host]
16:39 -!- meshcollider [meshcollid@user/meshcollider] has joined ##miniscript
--- Log closed Fri Sep 10 00:00:03 2021