Re: is there a different between write(1: %addr(data): len) and 'QtmhWrStout'

Scott

I don't know all the environmental setup for this, but when using sockets in our RPG2SQL product, if the program getting the socket is run interactively, the first socket returned is 3. If run in batch, it is 0. So it seems that interactive jobs also assume those 3 descriptors - of course, our code might have already set up something - haven't looked at that part much yet.

Vern

Scott Klement wrote:

hi Lim,

Not in all cases.

QtmhWrStout always writes to stdout. write(1: ....) always writes to descriptor 1. But they aren't always the same thing.

If I understand correctly, the original implementation of C on OS/400 didn't have descriptors. So stdin, stdout, stderr were not associated with a particular file descriptor.

Later, when descriptor support was added, IBM had to deal with the fact that some people expected 0,1,2 to be mapped to stdin,stdout,stderr (because that's how Unix works). So they invented an environment variable, QIBM_USE_DESCRIPTOR_STDIO that controls this.

Furthermore, 0,1,2 always match stdin, stdout, stderr when you are in QShell or PASE. I believe they also do when you are communicating with the HTTP server (i.e. CGI programming)

But there are situations where they don't match, such as in the native ILE environment (interactive or batch job) or when spawning a child job manually via the spawn() API and mapping pipes to descriptors 0,1,2, or when starting Java directly from RPG. In those cases, you have to set the QIBM_USE_DESCRIPTOR_STDIO variable...



hockchai Lim wrote:

is it correct for me to assume that write() (using handle ID of 1) and QtmhWrStout essentially does the same thing of writing the data to stdout?

thanks