dfh@scirtp.UUCP (David F. Hinnant) (12/10/85)
We have an Altos 586 running Xenix 3.0. To us, the C compiler has some rather unusual restrictions. Specificly, the following declarations are NOT allowed: long i; MUST be declared as "long int" short int i; MUST be declared as "short" unsigned long i; doesn't work at all unsigned long int i; ditto However both "unsigned i;" and "unsigned int i;" work OK. The compiler error message from "short i;" is "Type clash" while the error message from "long i;", "unsigned long i", etc. is "misplaced long". Is this normal for Xenix C compilers, or is just the Altos funny? -- David Hinnant SCI Systems, Inc. ...{decvax, akgua}!mcnc!rti-sel!scirtp!dfh
jsdy@hadron.UUCP (Joseph S. D. Yao) (12/18/85)
In article <557@scirtp.UUCP> dfh@scirtp.UUCP (David F. Hinnant) writes: >We have an Altos 586 running Xenix 3.0. ... > long i; MUST be declared as "long int" > short int i; MUST be declared as "short" > unsigned long i; doesn't work at all > unsigned long int i; ditto >However both "unsigned i;" and "unsigned int i;" work OK. The compiler >error message from "short i;" is "Type clash" while the error message from >"long i;", "unsigned long i", etc. is "misplaced long". Several early C compilers didn't allow unsigned longs, because it wasn't easy on their architectures. I auppose that's what's going on there. On the Altos 586 running Xenix 2.[35][abc] we had no problem with {long|short} [int] i; although since (again) early compilers treated long and short as qualifiers to the "real" data type I keep using "long int" and "short int". The fact that they seem to be treated opposite in your compiler must be a bug. Have you tried these all isolated, and with different variable names? -- Joe Yao hadron!jsdy@seismo.{CSS.GOV,ARPA,UUCP}