IBM Mainframe Forum

What is the "C and unix" way to you? As compared to some other environment?

Ive been trying to learn C and C++ for 10 years now, and the reason I cant get over the hump is that everything about C/C++ seems broken, I dont see how I can take such a language seriously with its dangling pointers, array buffer overflows, unsafe I/O functions that you have to back up with runtime checks, very weak typing. I have yet to get a straight answer from anyone as to why someone would want to program in a language that requires the programmer to micromanage all these language details instead of using a language thats safe in all these areas. Of course this is why Java and C# are so popular, they do just that, take away all the pointers, have garbage collection, leave the programmer to worry about his own program rather than spend all his time protecting himself from a unsafe language. So when looking for a safe language that compiles to an executable I only see 2 options...Ada and PL/I. I like Ada a lot so far and I honestly cant understand how anyone would want to flagulate themselves by using C/C++ in a large project when they can save so much time using a language like Ada. I dont know enough about PL/I yet to understand what its about, however I did find this article that compares C to PL/I which I believe addresses a lot of the major problems about C that people just seem to put up with:
http://www.uni-muenster.de/ZIV.Eberhard ... 1andC.html
Also PL/I was used by Gary Kildall to code his CP/M OS, so it must be a fairly capable language.

As for the unix OS, I found this book which addresses most of its major problems:
http://simson.net/ref/ugh.pdf
Also worth mentioning is the Plan 9 OS, that unix inventors worked on. I think there is a lot of room for improvement on unix that no one has gotten around to doing since the OS was never a unified standard.

This is not to contradict but rather to suggest that when one is starting to learn multiple new concepts from the very basics, it is best not to form heavy conclusions until considerable foundation has been established.

Ive discovered this to be true, there are no shortcuts to systems programming, which is why Im very curious about your opinion since youve worked on both z/OS and unix. Would you rate z/OS as a more stable and robust OS than unix?

i was asked to be the technical project lead for an effort to completely migrate the entire application inventory of an MVS environment

Im curious if you translated PL/I programs to C, and what your opinion on PL/I is. Do you think PL/I is as capable a systems programming language as C?

Hello,

Im curious if you translated PL/I programs to C

We had neither c nor pl/i. . . Keep in mind that these are not so common on the mainframe for application development. We ported all of the database definitions, data, cics screens, batch processes, and application code (mainly COBOL and CA-IDEAL) from the mainframe to HP-UX. The idea was to NOT re-write, but keep the existing application code. If all of the code had to be re-written, we could not have done this - at least not in any acceptable timeframe.

Would you rate z/OS as a more stable and robust OS than unix?

Personally, i wouldn't try to compare them. Properly administered, both provide a stable environment (i once allowed one of the Production Unix servers to run continuously for 6 months - just to prove it could). I have also worked on mainframes that have not had an unscheduled IPL for multiple years. I would rate them both much better than Windows for a "data center" environment.

CP/M was written in PL/M. not PL/I. As far as I know, the only major operating system written in PL/I was Multics. In any event, both CP/M and Multics have been dead and buried for years.

CP/M was written in PL/M. not PL/I. As far as I know, the only major operating system written in PL/I was Multics

Steve- PL/M is a dialect of PL/I designed by Gary Kildall
http://en.wikipedia.org/wiki/PL/M
You mention that its I/O system is built into the language, Im not sure thats really a bad thing as I/O is a major source of problems in C and C++.

I can see the advantage of a language like C in doing things like writing device drivers. Its weak typing and unrestrained pointers allow you do a lot of assembly-like tricks to interpret bits in different ways. But for doing large projects C and C++ is a complete disaster. Using header files and classes to create modularity is totally inadequet. Languages like Ada and PL/I were designed for making large projects and have constructs for building independent modules built into the language.

Even Ken Thompson (co-inventer of unix) and Bob Pike have invented a langugae called Go to try and make C a more secure systems language. Here is a article about what Bob Pike said about C and C++:
http://infoworld.com/d/developer-world/ ... lexity-375

In any event, both CP/M and Multics have been dead and buried for years.

As far as Multics and CP/M being "dead and buried", I first want to mention that MS DOS was a hacked version of CP/M and that Gary Kildall was actually preparing to launch a lawsuit against Microsoft for patent/copyright violations against CP/M. Multics was an over ambitious project that tried to do too many things that were beyond the hardware capabilities in its day. But there is nothing wrong or dated about the Multics OS itself which is why I think its worth peoples time to investigate older operating systems rather than try and re-invent the wheel by fixing/improving *nix. I invite you again to take a look at the "The Unix-Haters Handbook" book I linked to previously:
http://simson.net/ref/ugh.pdf
This book is not a book of jokes and rants by disgruntled unix users, its writen by experts who point out what is wrong with all the features of the OS. Its a worthwhile read.

Hello,

I realize this is an old topic, but we have had several requests like this. The following link was sent to me and might be of interest for someone looking to try out the mainframe. . .
http://www.ibm.com/developerworks/downl ... index.html
Other than the following quote (which is on the linked page), i don't know anything about the "sandbox" - if it is like other similar offerings, it has a time limit.

Try out IBM mainframe and enterprise modernization products running in a cloud environment. Hands-on exercises step you through real-world scenarios. No installation required! Here for the first time? Set up your system and access the self-help information to get the most out of your experience.