Troubleshooting Programs on the IBM 360 Quiz

Answer: 0C7

OC7 was a data exception, indicating you had likely attempted to perform mathematical operation on non-numeric data (or mis-matched data types in some other way). A related error was 0C8 which was thrown if you computed a value too large for the target storage location ("overflow exception"). The 0C5 ("addressing exception") or 0C4 ("protection exception") could result from careless address specification in Assembler programs; inattention to array index management or data linkage specifications in languages like FORTRAN or COBOL.

These exceptions ran the full gamut from 0C1 through 0CF were pronounced "oh-cee-x". The most ominous was the 0C3 ("oh-cee-three") which was defined as "execution exception error" generally indicating a serious flaw in program logic.

Answer: BLL (Base Locator for Linkage) cell and Displacement

The BLL cells each held an address which allowed for relative indexing into memory locations. In your dump you were given which BLL cell to use to reach to the "offending" instruction and the relative "displacement" from the address given in the BLL cell at which to find the instruction.

A little quick math (adding the displacement to the address in the BLL cell) and you now could find the offending location in the dump. BTW, the other two terms (AAI and DIE) are fabrications.

Answer: data file access type

BSAM (Basic Sequential Access Method), QSAM (Queued Sequential Access Method), ISAM (Indexed Sequential Access Method), and HISAM (Hierarchical Indexed Sequential Access Method) were all methods of organizing data on storage medium (often magnetic tape, but also on hard drives). For tape stored data, it was a critical programmer responsibility to understand the data access patterns to ensure that enormous amounts of time were not spent physically repositioning (rewind/FF) to get to the next piece of data to be read. BDAM (Basic Direct Access Method) was used exclusively for data on hard drives.

Answer: True

These were referred to as "overlays" and the programmer not only had to manage code to fit in a given overlay size (as little as 8K), but had to build into the program the logic to unload one overlay and load another keeping track of data/logic states. Code with the architecture to load/unload/re-load overlays while maintaining the integrity of operational logic/data was referred to as "re-entrant" code. By many, this was considered the ultimate skill of the senior programmer.

Answer: 1111 0001

In EBCDIC character codes hex F1 was 1, hex F2 was 2, etc. So, binary 1111 which is 15, would be F in hexadecimal and binary 0001 would be 1 in hexadecimal. So, 1111 0001 would be F1 which was EBCDIC coding for the digit 1.

Answer: 1111 1111

The "exclusive OR" was a frequent command encountered in your dumps of "assembler" code for your high-level program. The result of XOR on two fields was a bit-by-bit comparison with the result being 1 ("true") only if one (but not both) of the bits was 1 ("true").

Answer: Had the tape cleaned

The 9-track tapes used 8 tracks to record the 8 bits of each character stored and the 9th track was a parity bit. Often a spec of dirt could cause the data in a byte to be mis-read and this would mean the parity check would fail.

Answer: key-verified

Key-verifying involved giving the source coding sheets with the already punched deck of cards and using a special type of keypunch machine called a "verifier" and loading the cards into it and then re-punching the source. Anytime the keying did not match, the machine stopped and the keypuncher resolved the difference. Manpower to key everything twice was much cheaper than resources required to run multiple compile cycles until a clean compile was received.

Answer: you could use an incorrect version of the data

GDS (Generation Data Sets) were the early form of file versioning and allowed you to define a Generation Data Group (GDG) and then use relative verstion numbers in the file name (e.g. (0) = current version; (+1) = new next version; (-1) = previous to current version) with the number of versions maintained controlled by the GDG. So, if you had meant to use the current data by had added (-1) to the GDS in the JCL you would be using data one generation old. OOOPS --- and the job might run perfectly fine, maybe even creating a (+1) which at the end of the job would become (0) making your faux pas almost impossible to notice.

Answer: ,TYPRUN=SCAN

For some reason unknown to me, the parameter was TYPRUN rather than TYPERUN. So, adding ",TYPRUN=SCAN" would have the computer only check the JCL for syntax errors.