Re: Informix/4GL REALLY SUCKS

In article <67p3lr$lob@cssun.mathcs.emory.edu>, Jonathan Leffler <johnl@informix.com> writes >On Mon, 22 Dec 1997, Daniel Wright wrote: >} Gee....my copy of K&R (vol. I) > >I mentioned K&R; I don't think David Williams did. > >} doesn't really allude to floats or doubles being anything other than >} standard data types. > >Nor does mine. And I didn't say that it did. > >I did observe that on some machines, Informix software was compiled to >use the dec_t structure instead of float and double, because those >machines did not have hardware support for float and double (which >meant that the C compiler provided a software package to do float and >double calculations). > That's part of what I meant. Also note that dbexporting a float on one machine and dbimporting it onto another may not always work. There was an Informix bug about this. I'm not sure that the range of values for a float is defined - is it platform specific? Certainly on most platforms char = 8/16 int = 16/32/64, most are now 32 going on 64. long = 32/64 most are now 32 going on 64. float = undefined, no real standard. >I also observed that at the time K&R was originally written, it was >not possible to pass structures by value. Some C compilers allowed That was the other thing I was thinking - passing structures. > >Yours, >Jonathan Leffler (johnl@informix.com) #include <witticism.h> > Thanks for clearing it up Jonathan. I should really dig out my K&R and check what is says and not quote from vague memorys!! -- David Williams