Akim Demaille writes:
> 
> | Attached are backports of patches I found on the mailing
> | lists (c, cp and java). Checked today's CVS with 1.35, CVS with the
> | patch attached with 1.35 and with 1.50. cp and java don't show
> | regressions. for c:
> | 
> | - CVS and CVS+patch, both bison-1.35: no regressions.
> | - bison-1.35 and bison-1.50, both CVS+patch:
> | 
> | --- test-summary-1.35	2002-10-14 00:16:08.000000000 +0200
> | +++ test-summary-1.50	2002-10-13 13:55:39.000000000 +0200
> | @@ -39,11 +39,43 @@
> |  FAIL: gcc.c-torture/compile/20020927-1.c,  -O3 -g  
> |  FAIL: gcc.c-torture/execute/loop-2e.c execution,  -Os 
> |  FAIL: gcc.c-torture/execute/loop-3c.c execution,  -Os 
> | +FAIL: gcc.dg/noncompile/920923-1.c  (test for errors, line 69)
> | +FAIL: gcc.dg/noncompile/920923-1.c  (test for errors, line 72)
> | +FAIL: gcc.dg/noncompile/920923-1.c  (test for errors, line 74)
> | +FAIL: gcc.dg/noncompile/920923-1.c  (test for errors, line 77)
> [...]
> 
> Could someone given some details on these errors?  I have frequently
> used pre-1.50 to bootstrap GCC, with success.  Unfortunately, due to
> some stupid policy here, I have stopped downloading GCC and checking
> it with the current Bisons.  Nevertheless, I can't imagine what
> incompatibility can have been introduced in the meanwhile.

attched is the log of the first testcase run and the testcase.

> Also, I'm not sure how
> 
> | - CVS and CVS+patch, both bison-1.35: no regressions.
> | - bison-1.35 and bison-1.50, both CVS+patch:
> 
> should be read: it seems to say that there are no regressions with the
> patch and 1.35 (line 1),

comparing a build "current CVS" and "current CVS + backport patch",
both built with 1.35: no regressions.

> and there are regressions with the patch and 1.35 (line 2).

yes, regressions building "current CVS + backport patch", built with
bison-1.35 and bison-1.50.