OpenEuphoria

Re: = vs := and = vs ==

• Posted by petelomax May 11, 2014
• 1895 views

Mismatched sequence size is a no op for elements of the longer sequence [even for - and /]. 
{} op object is no op. Returns object unchanged. 

Interesting idea. Not that I have any compelling reason to make such changes, given the following function implements the current behaviour,

function sq_div(object a, object b) 
    if atom(a) then 
        if atom(b) then return a/b end if 
        for i=1 to length(b) do 
            b[i] = sq_div(a,b[i]) 
        end for 
        return b 
    elsif atom(b) then 
        for i=1 to length(a) do 
            a[i] = sq_div(a[i],b) 
        end for 
        return a 
    end if 
    if length(a)!=length(b) then fatal(a,b) end if 
    for i=1 to length(a) do 
        a[i] = sq_div(a[i],b[i]) 
    end for 
    return a 
end function 

then this implements that suggestion

function xq_div(object a, object b) 
    if atom(a) then 
        if atom(b) then return a/b end if 
        for i=1 to length(b) do 
            b[i] = xq_div(a,b[i]) 
        end for 
        return b 
    elsif atom(b) then 
        for i=1 to length(a) do 
            a[i] = xq_div(a[i],b) 
        end for 
        return a 
    end if 
--  if length(a)!=length(b) then fatal(a,b) end if 
    if length(a)<length(b) then 
        for i=1 to length(a) do 
            b[i] = xq_div(a[i],b[i]) 
        end for 
        return b 
    end if 
    for i=1 to length(b) do 
        a[i] = xq_div(a[i],b[i]) 
    end for 
    return a 
end function 

Not exactly difficult, at least in hll, though there are quite a few of them, and like I said I'm not convinced it is necessarily an improvement.

Oh, I have heard that one before

I concluded this is a knee-jerk reaction to the dreaded "true/false condition must be an ATOM", which used to plague me back in the days when I used RDS Eu.

Since moving full-time to Phix and removing implicit sequence ops, this happens, I kid you not, less than once a year.

Besides, I simply just don't buy that expressions such as "{1,2,3} and {4,5,6}" have any meaningful true/false answer, and for my money "programming error" wins hands down.

I have no clear idea what you really mean by that. If you are thinking of some mythical performance-related issue, I have found that it is almost always faster to use a non-sequence-op method, and in fact hll code such as that above can be aggressively pass-by-reference-optimized [although I should admit to recently breaking that handling in Phix] and therefore (at least for quite long/nested sequences) massively outperform the equivalent low-level code.

Pete