Le jeudi 26 novembre 2015 ร 09:04 -0800, JobJob a รฉcrit : > Seems like if s[a:b] is intended to index into s.data, then the > indexing should be consistent with arrays and s[end] should give the > last byte, though since s[1] and s[5] return characters, not a single > byte, there's a little bit of conflict. > Ok I've found a lengthy discussion of this and related issues, and > looks like this behaviour is very likely going to change: > https://github.com/JuliaLang/julia/issues/9297 > > In related news, this looks like an actual bug: > > s = "๐๐" > s12 = s[1:2] # "๐" > s56 = s[5:6] # "?" > sizeof(s12) # 4 > sizeof(s56) # 2 > > The issue seems to be that > nextind(s, 2) #5 > nextind(s, 6) #7 (prob should be 9) > > I think the fix is to change https://github.com/JuliaLang/julia/blob/ > master/base/strings/basic.jl#L141 to return sizeof(s)+1 > > Will file an issue. Good catch! Instead of filing an issue, could you directly apply the fix and make a pull request? Adding a test would also be a good idea to be sure we don't regress on this.
Thanks! > Good talk. :) > Happy thanksgiving to those in the USA
