27/02/2004 10:30:22, Leopold Toetsch <[EMAIL PROTECTED]> wrote: >[EMAIL PROTECTED] wrote: > >> Compiler version: >> >> Microsoft (R) 32-bit C/C++ Standard Compiler Version 13.00.9466 for 80x86 > >Thanks update. > > >> t\op\00ff-dos.t 255 65280 2 4 200.00% 1-2 > >Strange >
I've sent a patch that makes this problem "go away", though I do not understand
the cause. The 'fix' was to c&p the entire script into a new file, save it using
"codeset=ANSI" & "filetype=unix", overwriting the original.
Then the test completes without error.
Running "cvs diff -u -b" is the (attached)patch I've sent. I'll leave it to
others to interprete the diff.
>
>> c:\Perl\bin\perl.exe t\harness --gc-debug --running-make-test -C ubtests
>> failed, 10.96% okay.
>
>But -g is running. Even more Strange.
>
If I understand correctly (and it's quite likely I do not yet), the -C tests are
running with the "computed goto" core (core 4?). As far as I can determine from
some breif research on the web, msvc does not support this. Certainly, if I try
to compile config/auto/cgoto/test.c with default options I get
T:\config\auto\cgoto>cl test.c
Microsoft (R) 32-bit C/C++ Standard Compiler Version 13.00.9466 for 80x86
Copyright (C) Microsoft Corporation 1984-2001. All rights reserved.
test.c
test.c(9) : error C2059: syntax error : '&&'
test.c(12) : error C2059: syntax error : '*'
which I thought meant that this test would not be run, but I haven't figured out
what would stop it from being run when a "nmake fulltest" is invoked?
Looking through the output from "nmake testC", the only tests that are passing
are:
t\op\debuginfo..........ok
t\op\hacks..............ok
imcc\t\imcpasm\cfg......ok
imcc\t\imcpasm\opt0.....ok
imcc\t\imcpasm\opt1.....ok
imcc\t\imcpasm\opt2.....ok
imcc\t\imcpasm\optc.....ok
imcc\t\imcpasm\pcc......ok
imcc\t\imcpasm\sub......ok
These are what constitute the 10.96% passed figure, which tends to indicate that
these tests are not generally indicative of success in using the computed-goto
core (and should be placed somewhere else?).
The upshot is probably that the PLATFORMS line for this should probably now
read:
win32-ms-cl_13.00.9466 - Y/2 - - - Y Y/2
and note *1 can disappear.
I'd post a patch to this effect but I have no real confidence in my
interpretation.
>
>Thanks,
>
>leo
>
>
-- regards, Nigel.
00ff-dos.t.diff
Description: Binary data
