pull down to refresh
210 sats \ 1 reply \ @cypher_funk 22 Jun 2023 \ on: Coldkite Coldcard Reproducible Builds bitcoin
Did you see the follow-up thread as well? https://twitter.com/nvk/status/1671856113484955648
Thanks for your reply. Yes I did. I was also able to successfully build the latest MK3 and MK4 builds. I quote myself below.
For M4 5.1.2 I got SUCCESS from make repro The binary on the webpage 893,237 bytes and the build file in ~/firmware/stm32/built/firmware-signed.dfu was also 893,237 bytes.
For M3 4.1.8 I got SUCCESS from make repro The binary on the webpage is 753,981bytes, yet the build file in ~/firmware/stm32/built/firmware-signed.dfu was 722944 bytes
The issue is that older builds of MK3 fail with error 1, or else fail to build entirely.
It seems related to the build date being included in the build file, thus showing up in the diff. If it's just this, it seems minor to me and not anything to be concerned about. But it does cause the error 1 which is confusing to people and there is no documentation on the Coinkite site as to why this is the case.
What is particularity frustrating about NVK's response to this is that he's responding with successful MK4 5.1.2 and MK3 4.1.8 successful builds although that proves the process is working fine and that people who are finding problems must be doing it wrong or else malicious. But as I say, I am able to replicate the success of the videos people have posted with latest builds. The issue is about older MK3 builds. Also I am not sure how to explain the file size difference between even the successful MK3 4.1.8 build between the built file vs the binary on the webpage, although that may just be me misunderstanding something.
reply