bryan@cs.utexas.edu (Bryan Bayerdorffer @ Wit's End) (03/02/90)
Here's another one: I ran into a minor problem compiling the Amiga workbench code using Lattice 5.04. Part of amiwbench.c looks like this: #include "hack.h" . . . #ifdef LATTICE #include <strings.h> ... #endif It turns out that hack.h causes <stdlib.h> to be included. the external defs for the memory functions (memccpy, etc.) have argument types that differ from those in strings.h, where these functions are also declared. Buzzzzzz... Error 72: External item attribute mismatch. Since these functions aren't used, I got around this by making a copy of strings.h, deleting the external defs for the memory functions, and including this new file in place of strings.h Anybody else have this problem? Did I not R some part of TFM? I know this can happen to *other* people, but... :-)
keni@dtix.dt.navy.mil (Lorber) (03/02/90)
In article <448@mohawk.cs.utexas.edu> bryan@cs.utexas.edu writes: > > It turns out that hack.h causes <stdlib.h> to be included. the >external defs for the memory functions (memccpy, etc.) have argument types that >differ from those in strings.h, where these functions are also declared. >Buzzzzzz... Error 72: External item attribute mismatch. > Since these functions aren't used, I got around this by making a copy >of strings.h, deleting the external defs for the memory functions, and >including this new file in place of strings.h > >Anybody else have this problem? Did I not R some part of TFM? I know this can >happen to *other* people, but... :-) No, this is NOT caused by not R'ing TFM. This is a bug that crops up here and there due to an error made (as far as I can tell) by Lattice on one of their patches: are you really running 5.04a or 5.04b, and not the straight 5.04? The workaround should be OK, as far as I can tell. (My apologies to Lattice if this wasn't their fault - they've been quite helpful and pleasant when I talked to them.) Ken Lorber keni@dtix.dt.navy.mil news filler takes my space and your time!
bryan@cs.utexas.edu (Bryan Bayerdorffer @ Wit's End) (03/02/90)
In article <1138@nems.dt.navy.mil> keni@dtix.dt.navy.mil (Kenneth Lorber - Code: 3531) writes:
=-No, this is NOT caused by not R'ing TFM. This is a bug that crops up
=-here and there due to an error made (as far as I can tell) by Lattice
=-on one of their patches: are you really running 5.04a or 5.04b, and not
=-the straight 5.04? The workaround should be OK, as far as I can tell.
=-
As I said, this happened using 5.04 (no later patches applied).
However, someone else claims not to have hit this problem using the same
version of the compiler. Possibly some 5.04s are more equal than others. Mine
was patched up from 5.02.