secp256k1-sys: fix lowmemory feature
#799
Conversation
When upstream switched to multicomb they changed the #define flags needed to reduce the size of the ecmult_gen precomp table. We should have updated our bulid.rs. Before this change, on my system the secp256k1-sys rlib with the lowmemory feature has size 630602. After this change, it has size 610090, a 3.3% reduction. Probably not worth backporting, although it's not a breaking change and we totally could backport it. Fixes rust-bitcoin#795
|
Man do you have any reviewers in this crate ATM? I'm starting to have enough life force back to start expanding my work horizons again. I can't meaningfully review this without digging back into this crate and upstream. Thought I'd ask first to see if its a useful use of my time. Equally I could expand up to miniscript instead, which I've also been ignoring, probably best to only go one way at first. Which is more valuable to you? |
|
@tcharding no, there are no reviewers here. Things just rot unless there is enough popular demand for me to start phoning people. I would greatly appreciate your reviews. In this case the change looks scary but it's really not. You can see some magic |
|
Confirmed the config values were sane by reading the PR description: bitcoin-core/secp256k1#1058 |
When upstream switched to multicomb they changed the #define flags needed to reduce the size of the ecmult_gen precomp table. We should have updated our bulid.rs.
Before this change, on my system the secp256k1-sys rlib with the lowmemory feature has size 630602. After this change, it has size 610090, a 3.3% reduction. Probably not worth backporting, although it's not a breaking change and we totally could backport it.
Fixes #795