Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Compiling Parrot with Visual C++ 2005

2 views
Skip to first unread message

Ron Blaschke

unread,
Dec 10, 2005, 10:50:31 AM12/10/05
to perl6-i...@perl.org
The following test results are taken from Revision 10428 on a Windows
XP, Visual C++ 2005 Express Edition box.

Doesn't look too bad. I only had to change one thing: Remove the
declaration C<extern char **environ;> from C<classes/env.pmc>. For
that see also ticket perl #37665 [1].

Also, there are a lot of warnings for IO and string fuctions, e.g.
fopen and strcpy, going like this:

This function or variable may be unsafe. Consider using xxx_s instead.
To disable deprecation, use _CRT_SECURE_NO_DEPRECATE. See online help
for details.

Anyway, here are the test results.

Failed Test Stat Wstat Total Fail Failed List of Failed
-------------------------------------------------------------------------------
t/examples/pasm.t 1 256 7 1 14.29% 5
t/examples/pir.t 1 256 17 1 5.88% 10
t/pmc/env.t 2 512 12 2 16.67% 9-10
t/pmc/pmc.t 1 256 24 1 4.17% 17
6 tests and 363 subtests skipped.
Failed 4/180 test scripts, 97.78% okay. 5/4061 subtests failed, 99.88% okay.

[1] https://rt.perl.org/rt3/index.html?q=37665

Ron

Leopold Toetsch

unread,
Dec 10, 2005, 4:16:01 PM12/10/05
to Ron Blaschke, perl6-i...@perl.org

On Dec 10, 2005, at 16:50, Ron Blaschke wrote:

> The following test results are taken from Revision 10428 on a Windows
> XP, Visual C++ 2005 Express Edition box.
>
> Doesn't look too bad.

Indeed.

> Failed Test Stat Wstat Total Fail Failed List of Failed
> -----------------------------------------------------------------------
> --------
> t/examples/pasm.t 1 256 7 1 14.29% 5

Looks like PIO_win32_open doesn't process/convert '/'

> t/examples/pir.t 1 256 17 1 5.88% 10

Another IO test

> t/pmc/env.t 2 512 12 2 16.67% 9-10

Ok

Can you compare expected with result and have a look please.

> t/pmc/pmc.t 1 256 24 1 4.17% 17

And that one is fixed^Wskipped now (bogus test)

> Ron

leo

Ron Blaschke

unread,
Dec 12, 2005, 11:56:36 AM12/12/05
to Leopold Toetsch, perl6-i...@perl.org
Leopold Toetsch wrote:
> On Dec 10, 2005, at 16:50, Ron Blaschke wrote:

Here's a side by side comparison, Revision 10460, of
Visual C++ 7.1 and 8.0.

arithmetics.t suffers from -0.0 vs 0.0 problems. We've had something
similar some time ago. They seem to have changed that in 8.0, as the
test passes there.

Let me know at which things I should have a look, if any. Don't know
if anyone else is working on them. Also, if I should add a few lines
at README.win32.

Visual C++ 7.1
==============


Failed Test Stat Wstat Total Fail Failed List of Failed
-------------------------------------------------------------------------------
t/examples/pasm.t 1 256 7 1 14.29% 5

t/examples/pir.t 1 256 17 1 5.88% 10

t/op/arithmetics.t 1 256 26 1 3.85% 7


t/pmc/env.t 2 512 12 2 16.67% 9-10

t/pmc/eval.t 1 256 18 1 5.56% 13
6 tests and 364 subtests skipped.
Failed 5/180 test scripts, 97.22% okay. 6/4064 subtests failed, 99.85% okay.

Visual C++ 8.0
==============


Failed Test Stat Wstat Total Fail Failed List of Failed
-------------------------------------------------------------------------------
t/examples/pasm.t 1 256 7 1 14.29% 5

t/examples/pir.t 1 256 17 1 5.88% 10

t/pmc/env.t 2 512 12 2 16.67% 9-10

t/pmc/eval.t 1 256 18 1 5.56% 13
6 tests and 364 subtests skipped.
Failed 4/180 test scripts, 97.78% okay. 5/4064 subtests failed, 99.88% okay.

Ron

Jerry Gay

unread,
Dec 12, 2005, 12:37:12 PM12/12/05
to Ron Blaschke, Leopold Toetsch, perl6-i...@perl.org
On 12/12/05, Ron Blaschke <mailin...@rblasch.org> wrote:
> Here's a side by side comparison, Revision 10460, of
> Visual C++ 7.1 and 8.0.
>
great!

> arithmetics.t suffers from -0.0 vs 0.0 problems. We've had something
> similar some time ago. They seem to have changed that in 8.0, as the
> test passes there.
>

it seems msvc 6.0 matches the behavior of 8.0 (meaning 7.1 is probably
the broken one.)

> Let me know at which things I should have a look, if any. Don't know
> if anyone else is working on them. Also, if I should add a few lines
> at README.win32.
>

i don't think anyone is looking at these. the few active windows
developers have either been too busy to work on parrot, or (like me,)
involved in other parrot projects. if you could take a ny of these,
that would be appreciated.

the env test failures are known (#37301), but not addressed. making
the tests work would be preferable to skipping them, for what it's
worth :)

the eval tests aren't failing for me:

>prove -Ilib t/pmc/eval.t
t/pmc/eval....ok
1/18 skipped: too much old calling conventions
All tests successful, 1 subtest skipped.
Files=1, Tests=18, 3 wallclock secs ( 0.00 cusr + 0.00 csys = 0.00 CPU)

so that's something i can't help with unless it's reproducible.

the examples tests have been failing for some time now, but i haven't
taken the time to look at them closely. it would be nice to get them
passing (or skip them on windows) before the next release, but i think
env and eval tests are more important.

hope that helps.
~jerry

Ron Blaschke

unread,
Dec 13, 2005, 1:58:08 PM12/13/05
to jerry gay, perl6-i...@perl.org
jerry gay wrote:
> On 12/12/05, Ron Blaschke <mailin...@rblasch.org> wrote:
>> arithmetics.t suffers from -0.0 vs 0.0 problems. We've had something
>> similar some time ago. They seem to have changed that in 8.0, as the
>> test passes there.
>>
> it seems msvc 6.0 matches the behavior of 8.0 (meaning 7.1 is probably
> the broken one.)

Interesting. Didn't know that it works with 6.0.

>> Let me know at which things I should have a look, if any. Don't know
>> if anyone else is working on them. Also, if I should add a few lines
>> at README.win32.
>>
> i don't think anyone is looking at these. the few active windows
> developers have either been too busy to work on parrot, or (like me,)

> involved in other parrot projects. if you could take any of these,
> that would be appreciated.

Good, I'll have a look at things.

[lot of useful stuff snipped]

Thanks a lot for your response.

Ron


0 new messages