Commit | Line | Data |
---|---|---|
bc9c0cf3 C |
1 | Date: Fri, 26 Apr 1991 18:01:04 -0300 |
2 | From: mjlx@eagle.cnsf.cornell.edu (Mike Lijewski) | |
3 | To: arnold@audiofax.com | |
4 | Cc: david@cs.dal.ca | |
5 | Subject: testing 2.12 on a machine with unsigned chars | |
6 | ||
7 | I chose to use the alloca which you supply. The RS/6000 has a builtin | |
8 | alloca which is accessible using a `#pragma alloca', but I chose not | |
9 | to use it. Initially, I tried to use it by conditionally compiling it, | |
10 | similar to the way alloca.h is included on sparcs. But this has | |
11 | some problems. Firstly, the RS/6000 compiler complains about the | |
12 | placement of the #pragma, something to the intent that the pragma must | |
13 | precede all C code. This would be easy enough to fix by conditionally | |
14 | including the #pragma elsewhere in the relevant files. A more | |
15 | difficult problem is that the awk.tab.c generated by bison uses | |
16 | alloca. To fix this the right way, bison would have to be modified to | |
17 | output the appropriate conditionally compilable code as it does now | |
18 | for sparcs. If you think it is worth while to use the builtin alloca, | |
19 | I would be happy to get it working, except for the bison problem, and | |
20 | send you the diffs. The FSF might also be interested in "fixing" | |
21 | bison to use the builtin alloca on the RS/6000. |