Lint goto?

[ivarne]

This is a continuation of the lint @goto discussion in JuliaLang/julia#1289 (comment)

I would say the question to ask to find the is, "Is it common for the writer/reader of this code to think it does something different?". goto will often obfuscate and make the logic flow of your program hard to follow, but no one will be surprised about how it works. Lots of users gets surprised when [1:0] results in Int[] not [1, 0], and the only reason it doesn't is that we want type stability and a super efficient UnitRange type with a hardcoded (inlined) step of 1.

I assumed you had lintpragma for all lint messages, so that doesn't really change anything for me, but I'm not the one to dictate the scope of lint.jl. @goto is probably rarely the right answer to the question, so a lintpragma might just be another reason to really rethink your code structure.

[tonyhffong]

I'm fine with taking it out, as it looks like I don't feel as strongly about this as you do.

[StefanKarpinski]

I think that warning about goto is strange – it's not like you can easily use it by accident or anything. If you use it, you meant to and presumably know what you're doing.

[tonyhffong]

Thank you for your feedback. The code and tests have been removed.