🚀 go-pugleaf

Jorgen Grahn <grahn...@snipabacken.se> wrote:
> On Thu, 2018-03-01, John Levine wrote:
>> In article <slrnp9e76p.eu5.grahn+nntp@frailea.sa.invalid>,
>> Jorgen Grahn  <grahn...@snipabacken.se> wrote:
>>> That too, but I'm personally more interested in the time frame when
>>> tools I use today were created: the Unix shells, C compilers,
>>> vi/Emacs, troff ...
>> 
>> I was writing C programs on Unix by late 1975 and I don't ever
>> remember using a program to do variable crossrers.
> 
> (Note that I didn't mention such tools.)
> 
>> Whatever it needed to do, grep was good enough, or the tag schemes
>> later written for emacs and vi.
> 
> That's what I still do, 40 years later :-) Tags don't work too well
> with C++ (since you can have a dozen different functions called foo)
> but they /kind of/ work, and no better alternative has shown up as far
> as I'm concerned.
> 
> I don't think I've seen anyone use cxref, ever.  I once saw a guy use
> cscope, but that was just once, and I was too lazy to try it out.
> 
>> Before that I did use the PDP-10 MACRO-10 cref program from time to
>> time, even though we had text editors that had some sort of string
>> search.
> 
> /Jorgen
> 

The thing is, the compiler already has all that information, all nicely
organized. It's trivial to generate the output. The only way to get a
proper xref is by using a program that recognizes variable scope and
storage class - although it's *bad* form, I think in C you can have an
extern or static variable foo and also use different local foo's in a bunch
of functions. If you have a program that does enough syntax analysis to
determine this you have the front half of a compiler.

-- 
Pete