On 26/10/2009, at 11:31 PM, Dennis Lovelady wrote:
I disagree about the implication. From the function's perspective,
arguments are received; return values are passed.
From the caller's perspective parameters are passed and return values
received :) so your point is ...?
If that's true then because RPG's default behaviour is to presume no
widening then I should be able to specify no widening in the C
Are you sure we can call this "RPG's default behavior," Simon? If
it were,
wouldn't the change from default require coding something like
*CWIDEN,
rather than the opposite?
RPG doesn't do any widening. You have to tell it to widen by use of
*CWIDEN. I think the only reason *CNOWIDEN exists is to tell RPG to
deal with C's peculiar return mechanism. It's not necessary if #pragma
argument (fn, nowiden) is specified in the C code. You can see that no
widening occurs by a simple test (or by reading my original append and
seeing how RPG gets the leading X'00' from the C return value).
This supports my presumption that C always widens the return value so
I wonder what happens if I remove #pragma nowiden from the C function
and tell RPG that *CNOWIDEN is actually in effect. I make those
changes and now when I run the program I see compileFlags contains
X'75'. Curiouser and curiouser.
My assumption has always been that under the covers the binding
process is
aware of the language types of the caller and called functions.
Certain
"conversions" may be done in order to make the languages with
different
attributes play well together.
Based on previous discussions I'm fairly sure that after compilation
language differences disappear and are not visible or known to the
later steps (e.g., bind or run). I think there is special support for
CALL from command line to a C program (i.e., null-termination of
literals) but that's all. For inter-language calls the expectation is
that the caller passes the right stuff.
One who is brought up in the C world rather than the RPG world might
argue
that RPG is completely ridiculous in its approaches that differ from
C.
Except that it is only C and it's derivatives (including interpretive
languages based on C) that adopt these peculiarities. RPG, COBOL, PL/1
and a bunch of other unrelated languages all behave more like each
other than not. C is singularly the odd one out. Granted that if you
started with C and have little exposure to other languages then many
"normal" things (like fixed length character variables, decimal data
types, bidirectional parameter passing, a proper inter-program call
mechanism, a compiler that actually does things for you instead of the
bare minimum thus requiring various lint checkers to get "good" code,
etc. etc.) might seem ridiculous but that is really edging toward the
endless language wars which I'd rather not continue.
For
example, the length attribute rather than a null-terminated string.
That's
so IBM!
Including Pascal and it's derivatives so perhaps not "so IBM".
Besides, surely it makes much more sense to consume 2 bytes to hold
the current length than constantly having to scan for a magic end-of-
string marker? I've been told that C only does it that way because it
was developed on PDP hardware which had an efficient machine
instruction for doing exactly that. Thus 40 years later we are still
hamstrung by implementation decisions made with little foresight--
although I do have the advantage of hindsight when I say that.
But it is what it is, and we have to take the things we don't like
along with those that we consider cool.
Ah, but nothing about C is cool--it's an appalling language for almost
any task.
On the other hand, if they all worked alike, there'd be only one,
right?
That's true! There can be only one ...
However, you have not addressed any of the questions in my original
append. Do you have any information regarding those?
Midnight here so off to bed for me.
Regards,
Simon Coulter.
--------------------------------------------------------------------
FlyByNight Software OS/400, i5/OS Technical Specialists
http://www.flybynight.com.au/
Phone: +61 2 6657 8251 Mobile: +61 0411 091 400 /"\
Fax: +61 2 6657 8251 \ /
X
ASCII Ribbon campaign against HTML E-Mail / \
--------------------------------------------------------------------
midrange.com
