fix enum redeclaration bug: the sequence
enum {a=1}; f() { enum{a=1};
would leave the inner `a' value 2, not 1. from Chris Torek

generate a semantic message instead of dying (4.3BSD-tahoe/libexec/2)

utah rcsid 1.20 87/09/21 02:23:15: Code for the relook()/schain hack from
allegra!jpl which is quite a bit prettier than the old code and works the
same.

utah rcsid 1.19 87/07/10 22:45:12: Arthur Olsen's hack to force lint to
examine initializations.

utah rcsid 1.18 87/05/18 22:30:36: Fix goof in previous revision which
screws up when a symbol returned by relook() is the first element in an
schain.

utah rcsid 1.17 87/04/30 13:53:40: Prevent 'schain botch' -- if relook()
succeeds, fix up schain so that it passes through the copy instead of the
(mishashed) original.

utah rcsid 1.16 87/02/25 00:16:05: Prevent confusion when introducing a new
definition of an identifier in an inner scope. Previously an incomplete
definition of an external array could get 'adjuste

utah rcsid 1.16 87/02/25 00:16:05: Prevent confusion when introducing a new
definition of an identifier in an inner scope. Previously an incomplete
definition of an external array could get 'adjusted' by the definition of
an auto array with the same name.

show more ...

utah rcsid 1.15 87/02/08 20:23:37: Add typedefs to the set of storage
classes which may have null storage.

utah rcsid 1.14 87/01/15 06:57:19: Changed the compiler error in tsize()
into a normal user error -- you can get this by trying to dereference a
function pointer in ordinary code.

utah rcsid 1.13 87/01/15 05:50:53: Added fixinit() routine for recovering
the initial state after an error.

utah rcsid 1.12 87/01/15 02:00:24: Made test for missing declared arguments
in defid() more stringent. Made some minor lint fixes and touched up a
couple of comments.

utah rcsid 1.11 86/11/12 23:17:23: A variation on a suggestion by Ray
Butterworth -- when an array is not initialized and not external, its
definition must allocate space. I also cleaned up the corr

utah rcsid 1.11 86/11/12 23:17:23: A variation on a suggestion by Ray
Butterworth -- when an array is not initialized and not external, its
definition must allocate space. I also cleaned up the corresponding 'null
dimension' error for empty dimensions in places other than the last.
Following Butterworth's suggestion, a lint error was added for an explicit
dimension which is converted to a pointer in the normal formal argument
conversion code (contingent on lint's -h flag).

show more ...

utah rcsid 1.10 86/08/16 00:01:25: Don't allow addresses (or unloadable
things) to be placed in bit fields.

utah rcsid 1.9 86/08/03 01:06:07: Don't generate code after errors.

bug report 4.3BSD/lib/60

bug report 4.3BSD/usr.bin/44

additional check for empty stack

(1) Prevent compiler errors for 'int i = { 1, 2 };' and similar disasters.
(2) From Carl Sutton (sutton@tektronix) -- Save symbol table 'next' pointer
in step 2 in clearst() in case relookup() causes

(1) Prevent compiler errors for 'int i = { 1, 2 };' and similar disasters.
(2) From Carl Sutton (sutton@tektronix) -- Save symbol table 'next' pointer
in step 2 in clearst() in case relookup() causes 'r' to equal 'p' (resulting
in '*p' getting clobbered).

show more ...

Fix a bug that was confusing lint -- to free a symbol table element,
set its type to TNULL, not NULL.

Better error reporting for undefined structures. You used to get
'unknown size' for uninitialized undefined structures and a compiler
error if you tried to initialize one...

update from donn@utah-cs

use common header file for C, Fortran, Pascal.

don't core dump on "struct { a y; a z; };".

changes for single precision floats; lint cleanups

fix illegal storage class declarations for parameters.