Re: Win32 modules & cygwin

On Mon, Feb 19, 2007 at 09:46:30PM -0800, Yitzchak
Scott-Thoennes wrote:
> Just a heads up: I've had some time to work on this. 
I've got a
> patch I'm testing for parts 2 and 3 below (Jan already
did part 1)
> plus moving Win32API::File.
>
> As soon as everything passes on cygwin and
mingw-under-cygwin, I'll
> post it for others to test on VC with both dmake and
nmake.

Patch attached assumes win32/ext/Win32API has been moved to
ext/Win32API.

> Steve and I
> worked on static module builds for the various
compilers a couple of
> years ago (in preparation for this), and I don't
remember if we got
> everything or not.  I guess we'll find out.

Looks ok.

> I just installed borland again and will try to test
that too.

Starting it building now.

> I'm not sure what needs to be done to keep wince from
trying to
> compile the static Win32CORE module.

Looks like wince only picks up a subset of the extensions,
so maybe
nothing needs doing.

> On Thu, Dec 07, 2006 at 11:21:15AM -0800, Jan Dubois
wrote:
> > On Thu, 07 Dec 2006 09:14:43 +0000, Steve Hay
<steve.hayuk.radan.com>
> wrote:
> >
> > >Jan Dubois wrote:
> > >Is it worth doing likewise with
win32/ext/Win32API/File ? I've no idea
> > >whether that works under Cygwin or not.
> >
> > Yes, it works under Cygwin, and yes, I think it
should be moved too.
> >
> > I just started with Win32 to make sure everything
still works after the
> > move.  It is not done yet; I plan to do the
following additional steps:
> >
> > 1) Move the remaining Win32: functions
from win32/win32.c into
> >    ext/Win32/Win32.xs and just add forwarding
stubs to win32/win32.c
> >    that will "require Win32" and then
call the function in Win32.xs.
> >
> >    This makes the functions available to Cygwin
too.  Currently "they"
> >    maintain their own copy of theses function in a
statically linked
> >    extension called win32core or something like
that, but that isn't
> >    part of bleadperl.
> >
> > 2) Create an ext/win32core extension containing
the stub functions
> >    from win32/win32.c.  Make sure Cygwin will
statically link this
> >    extension.
> >
> >    With this change Cygwin should behave
identically with regards
> >    to all the Win32: builtin
functions right out of the box.
> >
> > 3) Statically link the ext/win32core module for
MSWin32 builds too,
> >    allowing us to remove the stubs from
win32/win32.c.
> >
> > 4) Update the docs to explain that some Win32: functions
are
> >    always defined for historical reasons, but
recommend that users
> >    should do a "use Win32" whenever they
use any of them.
> >
> > 5) Remove the Win32 module from libwin32 and
release it separately
> >    on CPAN to allow easier maintenance.

> On Mon, Feb 19, 2007 at 09:46:30PM -0800, Yitzchak
Scott-Thoennes wrote:
>> Just a heads up: I've had some time to work on
this.  I've got a
>> patch I'm testing for parts 2 and 3 below (Jan
already did part 1)
>> plus moving Win32API::File.
>>
>> As soon as everything passes on cygwin and
mingw-under-cygwin, I'll
>> post it for others to test on VC with both dmake
and nmake.
>
> Patch attached assumes win32/ext/Win32API has been
moved to ext/Win32API.

Forgot to mention, ext/Win32CORE/Win32CORE.c is mostly code
pulled out
of win32.c.  It should probably have a copyright statement,
but I'm not
sure what it should say.  win32.c itself has a very odd
copyright.

On Tue, 20 Feb 2007, Yitzchak Scott-Thoennes wrote:
>>> Just a heads up: I've had some time to work on
this. I've got a
>>> patch I'm testing for parts 2 and 3 below (Jan
already did part 1)
>>> plus moving Win32API::File.
>>>
>>> As soon as everything passes on cygwin and
mingw-under-cygwin, I'll
>>> post it for others to test on VC with both
dmake and nmake.
>>
>> Patch attached assumes win32/ext/Win32API has been
moved to
>> ext/Win32API.

Thanks for getting this done!  I hoped I would get around to
it
before Nicholas finished integrating things for 5.8.9, so
I'm
very glad you did this already.

I haven't actually tried the patch yet on my machine, but it
looks fine
to me from just reading it.  I'll try a build later today.

> Forgot to mention, ext/Win32CORE/Win32CORE.c is mostly
code pulled out
> of win32.c. It should probably have a copyright
statement, but I'm not
> sure what it should say. win32.c itself has a very odd
copyright.

I don't think there is anything in Win32CORE that dates back
to the hip
and Intergraph area.  I would suggest to put the standard
"copyright by
Larry Wall and others" copyright boilerplate in there.

Cheers,
-Jan

Jan Dubois wrote:
> On Tue, 20 Feb 2007, Yitzchak Scott-Thoennes wrote:
>>>> Just a heads up: I've had some time to work
on this. I've got a
>>>> patch I'm testing for parts 2 and 3 below
(Jan already did part 1)
>>>> plus moving Win32API::File.
>>>>
>>>> As soon as everything passes on cygwin and
mingw-under-cygwin, I'll
>>>> post it for others to test on VC with both
dmake and nmake.
>>>
>>> Patch attached assumes win32/ext/Win32API has
been moved to
>>> ext/Win32API.

Borland built fine too - 1 test failing with or without the
patch
(ext/POSIX/t/time.t).

> Thanks for getting this done!  I hoped I would get
around to it
> before Nicholas finished integrating things for 5.8.9,
so I'm
> very glad you did this already.
>
> I haven't actually tried the patch yet on my machine,
but it looks fine
> to me from just reading it.  I'll try a build later
today.

If you have ideas for what the Win32CORE.pm doc should say,
please
make suggestions.  I didn't touch what was there from the
cygwin
version.

Thank *you* for getting this rolling earlier.

>> Forgot to mention, ext/Win32CORE/Win32CORE.c is
mostly code pulled out
>> of win32.c. It should probably have a copyright
statement, but I'm not
>> sure what it should say. win32.c itself has a very
odd copyright.
>
> I don't think there is anything in Win32CORE that dates
back to the hip
> and Intergraph area.  I would suggest to put the
standard "copyright by
> Larry Wall and others" copyright boilerplate in
there.

OK.

On Tue, 20 Feb 2007, Yitzchak Scott-Thoennes wrote:
>> As soon as everything passes on cygwin and
mingw-under-cygwin, I'll
>> post it for others to test on VC with both dmake
and nmake.
>
> Patch attached assumes win32/ext/Win32API has been
moved to
> ext/Win32API.

I tried to test it, but I ran into other bleadperl issues:

        C:varbleadperlminiperl.exe
"-I....lib" "-I....lib"
-MExtUtils::Command -e rm_f re_comp.c
        C:varbleadperlminiperl.exe
"-I....lib" "-I....lib"
-MExtUtils::Command -e cp ....regcomp.c re_comp.c
        cl -c    -nologo -GF -W3 -MD -Zi -DNDEBUG -O1
-DWIN32 -D_CONSOLE -DNO_STRICT -DHAVE_DES_FCRYPT
-DPERL_IMPLICIT_CONTEXT
-DPERL_IMPLICIT_SYS -DUSE_PERLIO -DPERL_MSVCRT_READFIX -MD
-Zi -DNDEBUG -O1    -DVERSION="0.08" 
-DXS_VERSION="0.08"
"-I....libCORE"  -DPERL_EXT_RE_BUILD
-DPERL_EXT_RE_DEBUG -DPERL_EXT
re_comp.c(8005) : error C2143: syntax error : missing ';'
before 'type'
re_comp.c(8005) : error C2065: 're_debug_flags' : undeclared
identifier
NMAKE : fatal error U1077: 'cl' : return code '0x2'

I don't have time to look into this right now; I'll just try
again
tomorrow. 

Two notes on your patch though:

* It ended very abrupt, without a trailing newline.
* I think there are some tab/space mismatches happening. 
Are you using 4 char
  indented tabs?  The text inserted into win32/Win32.c had
TAB characters
  for 4 byte indents AFAICT.

Cheers,
-Jan

Jan Dubois wrote:
> On Tue, 20 Feb 2007, Yitzchak Scott-Thoennes wrote:
>>> As soon as everything passes on cygwin and
mingw-under-cygwin, I'll
>>> post it for others to test on VC with both
dmake and nmake.
>> Patch attached assumes win32/ext/Win32API has been
moved to
>> ext/Win32API.
> 
> I tried to test it, but I ran into other bleadperl
issues:
> 
>         C:varbleadperlminiperl.exe
"-I....lib" "-I....lib"
-MExtUtils::Command -e rm_f re_comp.c
>         C:varbleadperlminiperl.exe
"-I....lib" "-I....lib"
-MExtUtils::Command -e cp ....regcomp.c re_comp.c
>         cl -c    -nologo -GF -W3 -MD -Zi -DNDEBUG -O1
-DWIN32 -D_CONSOLE -DNO_STRICT -DHAVE_DES_FCRYPT
-DPERL_IMPLICIT_CONTEXT
> -DPERL_IMPLICIT_SYS -DUSE_PERLIO -DPERL_MSVCRT_READFIX
-MD -Zi -DNDEBUG -O1    -DVERSION="0.08" 
-DXS_VERSION="0.08"
> "-I....libCORE"  -DPERL_EXT_RE_BUILD
-DPERL_EXT_RE_DEBUG -DPERL_EXT
> re_comp.c(8005) : error C2143: syntax error : missing
';' before 'type'
> re_comp.c(8005) : error C2065: 're_debug_flags' :
undeclared identifier
> NMAKE : fatal error U1077: 'cl' : return code '0x2'
> 
> I don't have time to look into this right now; I'll
just try again
> tomorrow. 

Fixed in #30374 

(I haven't looked at Yitzchak's patch yet, but I'll try to
find some 
time in the next couple of days.)

-- 


------------------------------------------------
Radan Computational Ltd.

The information contained in this message and any files
transmitted with it are confidential and intended for the
addressee(s) only. If you have received this message in
error or there are any problems, please notify the sender
immediately. The unauthorized use, disclosure, copying or
alteration of this message is strictly forbidden. Note that
any views or opinions presented in this email are solely
those of the author and do not necessarily represent those
of Radan Computational Ltd. The recipient(s) of this message
should check it and any attached files for viruses: Radan
Computational will accept no liability for any damage caused
by any virus transmitted by this email.

Yitzchak Scott-Thoennes wrote:
> On Mon, Feb 19, 2007 at 09:46:30PM -0800, Yitzchak
Scott-Thoennes wrote:
>> Just a heads up: I've had some time to work on
this.  I've got a
>> patch I'm testing for parts 2 and 3 below (Jan
already did part 1)
>> plus moving Win32API::File.
>>
>> As soon as everything passes on cygwin and
mingw-under-cygwin, I'll
>> post it for others to test on VC with both dmake
and nmake.
> 
> Patch attached assumes win32/ext/Win32API has been
moved to ext/Win32API.

Thanks! Applied as #30379, together with appropriate 
integration/deletion to move Win32API/ and minor changes to
copyright 
and tabbing as suggested by Jan.

-- 


------------------------------------------------
Radan Computational Ltd.

The information contained in this message and any files
transmitted with it are confidential and intended for the
addressee(s) only. If you have received this message in
error or there are any problems, please notify the sender
immediately. The unauthorized use, disclosure, copying or
alteration of this message is strictly forbidden. Note that
any views or opinions presented in this email are solely
those of the author and do not necessarily represent those
of Radan Computational Ltd. The recipient(s) of this message
should check it and any attached files for viruses: Radan
Computational will accept no liability for any damage caused
by any virus transmitted by this email.

On Thu, 22 Feb 2007, Steve Hay wrote:
> Yitzchak Scott-Thoennes wrote:
>>> Just a heads up: I've had some time to work on
this. I've got a
>>> patch I'm testing for parts 2 and 3 below (Jan
already did part 1)
>>> plus moving Win32API::File.
>>>
>>> As soon as everything passes on cygwin and
mingw-under-cygwin, I'll
>>> post it for others to test on VC with both
dmake and nmake.
>>
>> Patch attached assumes win32/ext/Win32API has been
moved to
>> ext/Win32API.
>
> Thanks! Applied as #30379, together with appropriate
> integration/deletion to move Win32API/ and minor
changes to copyright
> and tabbing as suggested by Jan.

Works fine for me so far.

Cheers,
-Jan

On 2/22/07, Steve Hay <steve.hayuk.radan.com> wrote:
> Yitzchak Scott-Thoennes wrote:

> > Patch attached assumes win32/ext/Win32API has been
moved to ext/Win32API.
>
> Thanks! Applied as #30379, together with appropriate
> integration/deletion to move Win32API/

Hmm.  I get the following on VMS with blead30382:

ext/Win32API/File/t/file....................................
FAILED at test 1
ext/Win32API/File/t/tie.....................................
FAILED at test 1
ext/Win32CORE/t/win32core...................................
FAILED--expected
1 tests, saw 0

Do the tests check to see what OS they are running under
and/or
whether the extensions they depend on are built?

Craig Berry wrote:
> On 2/22/07, Steve Hay <steve.hayuk.radan.com> wrote:
>> Yitzchak Scott-Thoennes wrote:
> 
>> > Patch attached assumes win32/ext/Win32API has
been moved to 
>> ext/Win32API.
>>
>> Thanks! Applied as #30379, together with
appropriate
>> integration/deletion to move Win32API/
> 
> Hmm.  I get the following on VMS with blead30382:
> 
>
ext/Win32API/File/t/file....................................
FAILED at 
> test 1
>
ext/Win32API/File/t/tie.....................................
FAILED at 
> test 1
>
ext/Win32CORE/t/win32core...................................
FAILED--expected 
> 
> 1 tests, saw 0
> 
> Do the tests check to see what OS they are running
under and/or
> whether the extensions they depend on are built?

Sorry for the hassle. Do you need something like #29572?

http://public.activestate.com/cgi-bin/perlbrowse/p/29572


-- 


------------------------------------------------
Radan Computational Ltd.

The information contained in this message and any files
transmitted with it are confidential and intended for the
addressee(s) only. If you have received this message in
error or there are any problems, please notify the sender
immediately. The unauthorized use, disclosure, copying or
alteration of this message is strictly forbidden. Note that
any views or opinions presented in this email are solely
those of the author and do not necessarily represent those
of Radan Computational Ltd. The recipient(s) of this message
should check it and any attached files for viruses: Radan
Computational will accept no liability for any damage caused
by any virus transmitted by this email.