Also, at some point one may be able to write df.foo instead of df[:foo].
On Fri, May 2, 2014 at 11:21 AM, Harlan Harris <[email protected]> wrote: > What John said. This said, an old idea for the DF design would be to > include additional metadata for each column, which could include things > like an arbitrary Unicode pretty name that's not restricted to valid > variable name strings. > > > On Fri, May 2, 2014 at 10:08 AM, John Myles White < > [email protected]> wrote: > >> I would discourage using <R> as the name of a column in a DataFrame. >> >> Part of the reason we’re using symbols now is that it will encourage >> people to use column names that are valid Julia variable names. If you >> stick to valid variable names, you’ll always be able to use metaprogramming >> tools like those employed to generate formulas for GLM’s. >> >> — John >> >> On May 2, 2014, at 1:03 AM, Tomas Lycken <[email protected]> wrote: >> >> :"<R>" for symbol("<R>") makes sense to me, so if it's not in the way of >> anything else, I'm all for it. >> >> And yeah, :"<R>" == "<R>" returns true, so I don't see how this could >> really make something impossible to do, which is possible today. I guess if >> there's code out that quotes literal strings like that it'll break, but I >> doubt that there's a lot of it... I have no idea how such a change would be >> implemented, though, so I'm afraid I won't be of much help making it happen. >> >> // T >> >> On Friday, May 2, 2014 9:50:14 AM UTC+2, Stefan Karpinski wrote: >>> >>> We talked at some point about making :"<R>" syntax for symbol("<R>") and >>> requiring something like :("<R>") to express the fairly useless operation >>> of quoting the string "<R>" (this is useless because the result is just the >>> string "<R>"). Barring some problem with this that I'm not thinking of, I'd >>> be in favor of such a change. >>> >>> >>> On Fri, May 2, 2014 at 3:14 AM, Tomas Lycken <[email protected]>wrote: >>> >>>> Sorry to hijack the thread, but since I stumbled over this problem >>>> myself (in the same context) and didn't know about the `symbol("A")` >>>> syntax, this seems like an appropriate place to ask: >>>> >>>> In the dataframe I was working with, I had one column named "R", and >>>> another which I wanted to name "<R>". Using :R was no problem, but it's not >>>> possible to refer to :<R> at all. (Try it in the REPL - it parses it as an >>>> incomplete expression, and if I add something after I get an error "R not >>>> defined"...) >>>> >>>> I think it's cool that it's possible to define symbols from arbitrary >>>> strings using e.g. `symbol("<R>")`, but it's kind of clunky that you can't >>>> refer to them with the colon syntax once they're defined. Is there a way >>>> around this, or do I have to simply "deal with it"? =) >>>> >>>> // T >>>> >>>> >>>> On Thursday, May 1, 2014 7:24:05 AM UTC+2, Ivar Nesje wrote: >>>>> >>>>> Symbols in Julia is a special form of strings that are faster for some >>>>> operations (like comparisons) and much slower for others. Julia uses >>>>> symbols internally to represent variable names. >>>>> >>>>> You create a symbol from a string with the symbol("A") function, or >>>>> the :A syntax. >>>>> >>>> >>> >> >
