Re: Would it be a reasonable to ask IBM's programmers to, write a %sortsfl(sfl: column)

Nathan Andelin wrote:

I agree with you that IBM i installation procedures are a pain. But
I question your assertion about Apache, Java, WebSphere, Tomcat, PHP,
TCP/IP, and stream file operations performing poorly under IBM i
compared to other platforms. My personal benchmarking is a bit
limited, but tends to indicate that stream file & socket I/O is quite
comparable between IBM i and Wintel equivalents. If you have other
information, I'd be interested in hearing about it.

That has not been my experience at all. I haven't done actual benchmarks, but I can take the same Java program or PHP script and run it on i vs. running it on Windows or FreeBSD, and the Windows/FreeBSD copies are visibly faster.

Likewise, I've worked extensively with IFS. When doing heavy disk access, I see visible and noticeable speed differences from IFS vs. NTFS on Windows or UFS on BSD. If I'm careful about how I code my IFS routines, I can get acceptable performance on i, but certainly not the same speed I can get elsewhere.

I don't know what in the OS causes this, and I'm not sure what to blame it on. You guys know I'm a supporter of this platform. I just wish IBM would solve some of these problems. And it drives me absolutely nuts when people talk about OS/400 being great, and RPG being the cause of the problem, when it's the total opposite in my opinion. RPG is really the saving grace of this system. It's the main reason it's still useful.

The remarkable thing about IBM i is not it's support for 5250
interfaces, but rather it's ability to handle complex workloads
efficiently and reliability. That may not do much for Java and PHP,
but it works for ILE based Web applications.

Amen to that! I certainly wasn't the one putting down ILE... Again, I think RPG (and by that I think ILE is implied) is the main attraction of this system. That and DB2, of course.

The reliability and maintainability of the system is another strong point, of course.