On Mon, Apr 02, 2001 at 11:25:39PM -0500, Craig A. Berry wrote:
> At 8:39 PM -0500 4/2/01, Craig A. Berry wrote:
> > I'm currently building with
> >-Duse64bitall, Compaq C 6.4
>
> All tests successful there.
>
> >, though older compiler/OS versions,
> >different TCP/IP vendors, and VAXen are of greater concern
>
> Building foolperl using DEC C V5.2-003 on OpenVMS Alpha V7.1,
> Multinet V4.2 Rev A-X, I get all tests successful except:
>
> [.lib]socket.t..........FAILED on test 9
> [.op]utf8decode.........FAILED on test 53
>
> socket.t has only 8 tests, which all pass successfully. The reported
> failure is another example of the now infamous mailbox overflow in
> the test suite's pipes. The patch attached here makes the failure go
> away and changes only the printing out of a '#' comment to keep it on
> one line.
Sarathy seems to have applied this patch.
> The utf8decode problem is a genuine problem, perhaps a bug in this
> six-year-old compiler:
>
> ok 51
> # 4.2.1: Malformed UTF-8 character (2 bytes, need 1) in unpack at
> [.op]utf8decode.t line 142.
> ok 52
> # 4.2.2: Malformed UTF-8 character (3 byte(null), need 2) in unpack
> at [.op]utf8decode.t line 142.
This comes from utf8.c:Perl_utf8_to_uv():387..388:
case UTF8_WARN_LONG:
Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)",
expectlen, expectlen == 1 ? "": "s", UNISKIP(uv));
break;
All that was expected was "s", but now we got "(null)"... usually you get
that from stdio if you are trying print NULL?
> 4.2.2: unpack('U*') false positive
> not ok 53
> FAILED on test 53
>
>
> It'd be nice to figure this out, but it's not worth holding up the release.
--
$jhi++; # http://www.iki.fi/jhi/
# There is this special biologist word we use for 'stable'.
# It is 'dead'. -- Jack Cohen
