robamu
commented
1 year ago Hmm, I don't think const_new is something that common. This is just a variant which allows creation of structs in const contexts, which might be useful for some cases. The caveat here of course is that I can not perform regular error handling, so I don't really have any other choice than to panic on invalid/unsanitized input. The new APIs should not panic however, do you have concrete examples?
About the subsec_millis being greater than 999: It is assumed that the subsec_millis only contain the millisecond part on top of the full UNIX seconds, so a value greater than 999 would be an additional second.
EthanJamesLew
Hello!
I've been quite interested in a rust implementation of spacepackets and have enjoyed using your project.
I have been grokking the codebase of the Rust spacepacket implementation and noticed the use of panics in the builder methods, specifically in the
const_newandnewfunctions of various structs. While I understand that panics are a part of Rust's error handling mechanism and can be useful for catching exceptional cases, I have concerns regarding their usage in these particular scenarios.First, in the
UnixTimestampstruct, theconst_newfunction panics if thesubsec_millisvalue exceeds 999. Similarly, in thePacketSequenceCtrlstruct, both theconst_newand new functions panic if theseq_countvalue exceeds MAX_SEQ_COUNT. Although panics can be appropriate for handling unexpected or invalid inputs, I'm unsure whether they are the best approach in these cases. I do understand for the latter example that these panic conditions are checked before calling, so they won't panic in those cases.const_newandnew, common for Rust 2021 projects? Would there be preferable ways to callconst_newwithout requiring the caller to check the panic conditions?subsec_millisgreater than 999? Would there be cases where the value may be user-specified?Thank you,
Ethan