Mailing List Archive: MythTV: Dev

Re: Ticket #4497: Win32 builds broken after r15435

Index | Next | Previous | View Threaded

jmodi at vt

Jan 20, 2008, 8:06 PM

Post #1 of 24 (3799 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435

On Jan 19, 2008 9:32 PM, MythTV <mythtv [at] cvs> wrote:

> #4497: Win32 builds broken after r15435
>
> ------------------------------------------+---------------------------------
> Reporter: Jay Modi <jaymode [at] gmail> | Owner: ijr
> Type: patch | Status: new
> Priority: minor | Milestone: unknown
> Component: mythtv | Version: unknown
> Severity: medium | Mlocked: 0
>
> ------------------------------------------+---------------------------------
> The Win32 builds seem to have been partially broken by r15435. The first
> issue I noticed is that the typedef on line 89 needs a semicolon.
> However,
> this leads to even more problems, dealing with multiple definitions. The
> cure for this is to remove the typedef and then include compat.h but this
> causes linking errors. These linking errors are due to some other issue,
> as I have removed the entire r15435 changes from my compat.h and get the
> same errors.
>
> I have attached a patch file to add the include.
>
> --
> Ticket URL: <http://svn.mythtv.org/trac/ticket/4497>
> MythTV <http://svn.mythtv.org/trac>
> MythTV
>

I wrote this ticket up last night and have tried again today with a clean
subversion checkout and no patches applied. Now the weird thing is that I
got a new error today. This error comes in a file that has not been changed
in years according to trac. Is anyone else (the few of us) that are building
with MinGW experiencing similar issues on the HEAD revision of svn.

Using this configure command:
./configure --prefix=/usr --disable-dbox2 --disable-hdhomerun --disable-dvb
-
-disable-ivtv --disable-iptv --disable-joystick-menu --disable-xvmc-vld
--disab
le-x11 --disable-xvmc --enable-directx --enable-memalign-hack
--compile-type=de
bug

When I run make, this is the new error:
gcc -c -march=i586 -g -Wall -Wno-switch -Wdisabled-optimization
-Wpointer-arith -Wredundant-decls -fPIC -DPIC -mthreads -w -fexceptions
-frtti -DUSING_MINGW -DUSING_WINAUDIO -DUSING_D3D -DUSING_DIRECTX
-DUSING_DIRECTX -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr\"
-DLIBDIR=\"/usr/lib\" -DHAVE_AV_CONFIG_H -D_LARGEFILE_SOURCE -DQT_DLL
-DQT_THREAD_SUPPORT -I"/usr/include" -I".." -I"../.." -I"/qt-3.3.x-p8/include"
-I"C:/msys/1.0/qt-3.3.x-p8/mkspecs/win32-g++" -o samplerate.o samplerate.c
In file included from samplerate.c:27:
float_cast.h: In function `lrint':
float_cast.h:90: error: `_asm' undeclared (first use in this function)
float_cast.h:90: error: (Each undeclared identifier is reported only once
float_cast.h:90: error: for each function it appears in.)
float_cast.h:90: error: syntax error before '{' token
float_cast.h: In function `lrintf':
float_cast.h:102: error: `_asm' undeclared (first use in this function)
float_cast.h:102: error: syntax error before '{' token
make[2]: *** [samplerate.o] Error 1
make[2]: Leaving directory `/c/mythtv/mythtv/libs/libmythsamplerate'
make[1]: *** [sub-libmythsamplerate] Error 2
make[1]: Leaving directory `/c/mythtv/mythtv/libs'
make: *** [sub-libs] Error 2

davidbuzz at gmail

Jan 20, 2008, 9:46 PM

Post #2 of 24 (3722 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

Yes, I'm experiencing similar isues on win32 with trunk. r15433 works for
the main mythtv (but even this is broken for mythplugins, but htat's a minor
issue presently for me compared to the main mythtv).

Buzz.
On Jan 21, 2008 2:06 PM, Jay Modi <jmodi [at] vt> wrote:

> On Jan 19, 2008 9:32 PM, MythTV <mythtv [at] cvs> wrote:
>
> > #4497: Win32 builds broken after r15435
> >
> > ------------------------------------------+---------------------------------
> > Reporter: Jay Modi <jaymode [at] gmail> | Owner: ijr
> > Type: patch | Status: new
> > Priority: minor | Milestone: unknown
> > Component: mythtv | Version: unknown
> > Severity: medium | Mlocked: 0
> >
> > ------------------------------------------+---------------------------------
> > The Win32 builds seem to have been partially broken by r15435. The
> > first
> > issue I noticed is that the typedef on line 89 needs a semicolon.
> > However,
> > this leads to even more problems, dealing with multiple definitions.
> > The
> > cure for this is to remove the typedef and then include compat.h but
> > this
> > causes linking errors. These linking errors are due to some other
> > issue,
> > as I have removed the entire r15435 changes from my compat.h and get
> > the
> > same errors.
> >
> > I have attached a patch file to add the include.
> >
> > --
> > Ticket URL: < http://svn.mythtv.org/trac/ticket/4497>
> > MythTV <http://svn.mythtv.org/trac>
> > MythTV
> >
>
>
>
> I wrote this ticket up last night and have tried again today with a clean
> subversion checkout and no patches applied. Now the weird thing is that I
> got a new error today. This error comes in a file that has not been changed
> in years according to trac. Is anyone else (the few of us) that are building
> with MinGW experiencing similar issues on the HEAD revision of svn.
>
> Using this configure command:
> ./configure --prefix=/usr --disable-dbox2 --disable-hdhomerun
> --disable-dvb -
> -disable-ivtv --disable-iptv --disable-joystick-menu --disable-xvmc-vld
> --disab
> le-x11 --disable-xvmc --enable-directx --enable-memalign-hack
> --compile-type=de
> bug
>
> When I run make, this is the new error:
> gcc -c -march=i586 -g -Wall -Wno-switch -Wdisabled-optimization
> -Wpointer-arith -Wredundant-decls -fPIC -DPIC -mthreads -w -fexceptions
> -frtti -DUSING_MINGW -DUSING_WINAUDIO -DUSING_D3D -DUSING_DIRECTX
> -DUSING_DIRECTX -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr\"
> -DLIBDIR=\"/usr/lib\" -DHAVE_AV_CONFIG_H -D_LARGEFILE_SOURCE -DQT_DLL
> -DQT_THREAD_SUPPORT -I"/usr/include" -I".." -I"../.." -I"/qt- 3.3.x-p8/include"
> -I"C:/msys/1.0/qt-3.3.x-p8/mkspecs/win32-g++" -o samplerate.o samplerate.c
> In file included from samplerate.c:27:
> float_cast.h: In function `lrint':
> float_cast.h:90: error: `_asm' undeclared (first use in this function)
> float_cast.h:90: error: (Each undeclared identifier is reported only once
> float_cast.h:90: error: for each function it appears in.)
> float_cast.h:90: error: syntax error before '{' token
> float_cast.h: In function `lrintf':
> float_cast.h:102: error: `_asm' undeclared (first use in this function)
> float_cast.h:102: error: syntax error before '{' token
> make[2]: *** [samplerate.o] Error 1
> make[2]: Leaving directory `/c/mythtv/mythtv/libs/libmythsamplerate'
> make[1]: *** [sub-libmythsamplerate] Error 2
> make[1]: Leaving directory `/c/mythtv/mythtv/libs'
> make: *** [sub-libs] Error 2
>
>
> _______________________________________________
> mythtv-dev mailing list
> mythtv-dev [at] mythtv
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
>
>

davidbuzz at gmail

Jan 20, 2008, 9:56 PM

Post #3 of 24 (3731 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

AFAIK, the current patch-set that is needed for a successful win32 build
consists of:

* r15433 or earlier

* qt.patch for QT from here: http://tanas.ca/qt.patch

* backend.patch.gz from trac #4392
http://svn.mythtv.org/trac/raw-attachment/ticket/4392/backend.patch.gz

* importicons_windows_2.diff from trac #3334
http://svn.mythtv.org/trac/raw-attachment/ticket/3334/importicons_windows_2.diff

* touch mythtv/config/config.pro (which seems to be missing from SVN?)

* mythtv/libs/libmyth/mythverbose.h doesn't get installed to the /include
folder, so mythplugins build fails unless fixed manually.

* and mythplugins build fails anyway with:
make[2]: Entering directory `/C/mythtv/mythplugins/mythvideo/mythvideo'
g++ -c -march=k8 -g -Wall -Wno-switch -Wpointer-arith -Wredundant-decls
-Wno-non
-virtual-dtor -D__STDC_CONSTANT_MACROS -mthreads -Wall -g -fexceptions
-frtti -
DUNICODE -D_GNU_SOURCE -DPREFIX=\"/usr\" -DMMX -Di386 -DUSING_DIRECTX
-D_FILE_OF
FSET_BITS=64 -DUSING_MINGW -DQT_DLL -DQT_PLUGIN -DQT_THREAD_SUPPORT
-I"/usr/incl
ude" -I"/usr/include" -I"/qt-3.3.x-p8/include" -I"C:/MSys/1.0/qt-3.3.x-p8
/mkspec
s/win32-g++" -o main.o main.cpp
In file included from C:/MSys/1.0/include/mythtv/mythcontext.h:6,
from main.cpp:16:
In file included from C:/MSys/1.0/include/mythtv/dialogbox.h:7,
from dvdripbox.h:20,
from main.cpp:35:
C:/MSys/1.0/include/mythtv/compat.h: In function `const char* dlerror()':
C:/MSys/1.0/include/mythtv/compat.h:206: error: cannot convert `char*' to
`WCHAR
*' for argument `5' to `DWORD FormatMessageW(DWORD, const void*, DWORD,
DWORD, W
CHAR*, DWORD, char**)'
C:/MSys/1.0/include/mythtv/compat.h:208: warning: int format, DWORD arg (arg
4)
make[2]: *** [main.o] Error 1
make[2]: Leaving directory `/C/mythtv/mythplugins/mythvideo/mythvideo'
make[1]: *** [sub-mythvideo] Error 2
make[1]: Leaving directory `/C/mythtv/mythplugins/mythvideo'
make: *** [sub-mythvideo] Error 2

Buzz

On Jan 21, 2008 2:06 PM, Jay Modi <jmodi [at] vt> wrote:

> On Jan 19, 2008 9:32 PM, MythTV <mythtv [at] cvs> wrote:
>
> > #4497: Win32 builds broken after r15435
> >
> > ------------------------------------------+---------------------------------
> > Reporter: Jay Modi <jaymode [at] gmail> | Owner: ijr
> > Type: patch | Status: new
> > Priority: minor | Milestone: unknown
> > Component: mythtv | Version: unknown
> > Severity: medium | Mlocked: 0
> >
> > ------------------------------------------+---------------------------------
> > The Win32 builds seem to have been partially broken by r15435. The
> > first
> > issue I noticed is that the typedef on line 89 needs a semicolon.
> > However,
> > this leads to even more problems, dealing with multiple definitions.
> > The
> > cure for this is to remove the typedef and then include compat.h but
> > this
> > causes linking errors. These linking errors are due to some other
> > issue,
> > as I have removed the entire r15435 changes from my compat.h and get
> > the
> > same errors.
> >
> > I have attached a patch file to add the include.
> >
> > --
> > Ticket URL: < http://svn.mythtv.org/trac/ticket/4497>
> > MythTV <http://svn.mythtv.org/trac>
> > MythTV
> >
>
>
>
> I wrote this ticket up last night and have tried again today with a clean
> subversion checkout and no patches applied. Now the weird thing is that I
> got a new error today. This error comes in a file that has not been changed
> in years according to trac. Is anyone else (the few of us) that are building
> with MinGW experiencing similar issues on the HEAD revision of svn.
>
> Using this configure command:
> ./configure --prefix=/usr --disable-dbox2 --disable-hdhomerun
> --disable-dvb -
> -disable-ivtv --disable-iptv --disable-joystick-menu --disable-xvmc-vld
> --disab
> le-x11 --disable-xvmc --enable-directx --enable-memalign-hack
> --compile-type=de
> bug
>
> When I run make, this is the new error:
> gcc -c -march=i586 -g -Wall -Wno-switch -Wdisabled-optimization
> -Wpointer-arith -Wredundant-decls -fPIC -DPIC -mthreads -w -fexceptions
> -frtti -DUSING_MINGW -DUSING_WINAUDIO -DUSING_D3D -DUSING_DIRECTX
> -DUSING_DIRECTX -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr\"
> -DLIBDIR=\"/usr/lib\" -DHAVE_AV_CONFIG_H -D_LARGEFILE_SOURCE -DQT_DLL
> -DQT_THREAD_SUPPORT -I"/usr/include" -I".." -I"../.." -I"/qt- 3.3.x-p8/include"
> -I"C:/msys/1.0/qt-3.3.x-p8/mkspecs/win32-g++" -o samplerate.o samplerate.c
> In file included from samplerate.c:27:
> float_cast.h: In function `lrint':
> float_cast.h:90: error: `_asm' undeclared (first use in this function)
> float_cast.h:90: error: (Each undeclared identifier is reported only once
> float_cast.h:90: error: for each function it appears in.)
> float_cast.h:90: error: syntax error before '{' token
> float_cast.h: In function `lrintf':
> float_cast.h:102: error: `_asm' undeclared (first use in this function)
> float_cast.h:102: error: syntax error before '{' token
> make[2]: *** [samplerate.o] Error 1
> make[2]: Leaving directory `/c/mythtv/mythtv/libs/libmythsamplerate'
> make[1]: *** [sub-libmythsamplerate] Error 2
> make[1]: Leaving directory `/c/mythtv/mythtv/libs'
> make: *** [sub-libs] Error 2
>
>
> _______________________________________________
> mythtv-dev mailing list
> mythtv-dev [at] mythtv
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
>
>

davidbuzz at gmail

Jan 21, 2008, 3:54 AM

Post #4 of 24 (3708 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

FYI,
after compiling about 20 revisions tonite, I have the following brief for
near-HEAD Win32 builds :

------------------------------------------
15434 -> is the last version to build successfully (with the known requisite
win32 patches)
------------------------------------------
15435 -> 15477 all give this same compile error:
In file included from uitypes.h:19,
from managedlist.h:9,
from managedlist.cpp:4:
util.h:89: error: `LONGLONG' does not name a type
make[2]: *** [managedlist.o] Error 1
make[2]: Leaving directory `/c/mythtv/mythtv/libs/libmyth'
make[1]: *** [sub-libmyth] Error 2
make[1]: Leaving directory `/c/mythtv/mythtv/libs'
make: *** [sub-libs] Error 2
------------------------------------------
15478 ->15514(HEAD) all give this same compile problem:
In file included from httprequest.cpp:37:
../libmyth/util.h: At global scope:
../libmyth/util.h:90: error: expected initializer before "inline"
In file included from ../libmyth/mythcontext.h:6,
from httprequest.cpp:38:
C:/msys/1.0/qt-3.3.x-p8/include/qpixmap.h:249: warning: 'void
bitBlt(QPaintDevice*, int, int, const QPaintDevice*, int, int, int, int,
Qt::RasterOp, bool)' redeclared without dllimport attribute after being
referenced with dllimport linkage
make[2]: *** [httprequest.o] Error 1
make[2]: Leaving directory `/c/mythtv/mythtv/libs/libmythupnp'
make[1]: *** [sub-libmythupnp] Error 2
make[1]: Leaving directory `/c/mythtv/mythtv/libs'
make: *** [sub-libs] Error 2
------------------------------------------

On Jan 21, 2008 3:56 PM, buzz <davidbuzz [at] gmail> wrote:

> AFAIK, the current patch-set that is needed for a successful win32 build
> consists of:
>
> * r15433 or earlier
>
> * qt.patch for QT from here: http://tanas.ca/qt.patch
>
> * backend.patch.gz from trac #4392
> http://svn.mythtv.org/trac/raw-attachment/ticket/4392/backend.patch.gz
>
> * importicons_windows_2.diff from trac #3334 http://svn.mythtv.org/trac/raw-attachment/ticket/3334/importicons_windows_2.diff
>
>
> * touch mythtv/config/config.pro (which seems to be missing from SVN?)
>
> * mythtv/libs/libmyth/mythverbose.h doesn't get installed to the /include
> folder, so mythplugins build fails unless fixed manually.
>
> * and mythplugins build fails anyway with:
> make[2]: Entering directory `/C/mythtv/mythplugins/mythvideo/mythvideo'
> g++ -c -march=k8 -g -Wall -Wno-switch -Wpointer-arith -Wredundant-decls
> -Wno-non
> -virtual-dtor -D__STDC_CONSTANT_MACROS -mthreads -Wall -g -fexceptions
> -frtti -
> DUNICODE -D_GNU_SOURCE -DPREFIX=\"/usr\" -DMMX -Di386 -DUSING_DIRECTX
> -D_FILE_OF
> FSET_BITS=64 -DUSING_MINGW -DQT_DLL -DQT_PLUGIN -DQT_THREAD_SUPPORT
> -I"/usr/incl
> ude" -I"/usr/include" -I"/qt- 3.3.x-p8/include" -I"C:/MSys/1.0/qt-3.3.x-p8
> /mkspec
> s/win32-g++" -o main.o main.cpp
> In file included from C:/MSys/1.0/include/mythtv/mythcontext.h:6,
> from main.cpp:16:
> In file included from C:/MSys/1.0/include/mythtv/dialogbox.h:7,
> from dvdripbox.h:20,
> from main.cpp:35:
> C:/MSys/1.0/include/mythtv/compat.h: In function `const char* dlerror()':
> C:/MSys/1.0/include/mythtv/compat.h:206: error: cannot convert `char*' to
> `WCHAR
> *' for argument `5' to `DWORD FormatMessageW(DWORD, const void*, DWORD,
> DWORD, W
> CHAR*, DWORD, char**)'
> C:/MSys/1.0/include/mythtv/compat.h:208: warning: int format, DWORD arg
> (arg 4)
> make[2]: *** [ main.o] Error 1
> make[2]: Leaving directory `/C/mythtv/mythplugins/mythvideo/mythvideo'
> make[1]: *** [sub-mythvideo] Error 2
> make[1]: Leaving directory `/C/mythtv/mythplugins/mythvideo'
> make: *** [sub-mythvideo] Error 2
>
>
> Buzz
>
>
>
> On Jan 21, 2008 2:06 PM, Jay Modi <jmodi [at] vt> wrote:
>
> > On Jan 19, 2008 9:32 PM, MythTV <mythtv [at] cvs > wrote:
> >
> > > #4497: Win32 builds broken after r15435
> > > ------------------------------------------+---------------------------------
> > >
> > > Reporter: Jay Modi <jaymode [at] gmail> | Owner: ijr
> > > Type: patch | Status: new
> > > Priority: minor | Milestone: unknown
> > > Component: mythtv | Version: unknown
> > > Severity: medium | Mlocked: 0
> > >
> > > ------------------------------------------+---------------------------------
> > > The Win32 builds seem to have been partially broken by r15435. The
> > > first
> > > issue I noticed is that the typedef on line 89 needs a semicolon.
> > > However,
> > > this leads to even more problems, dealing with multiple definitions.
> > > The
> > > cure for this is to remove the typedef and then include compat.h but
> > > this
> > > causes linking errors. These linking errors are due to some other
> > > issue,
> > > as I have removed the entire r15435 changes from my compat.h and get
> > > the
> > > same errors.
> > >
> > > I have attached a patch file to add the include.
> > >
> > > --
> > > Ticket URL: < http://svn.mythtv.org/trac/ticket/4497>
> > > MythTV <http://svn.mythtv.org/trac >
> > > MythTV
> > >
> >
> >
> >
> > I wrote this ticket up last night and have tried again today with a
> > clean subversion checkout and no patches applied. Now the weird thing is
> > that I got a new error today. This error comes in a file that has not been
> > changed in years according to trac. Is anyone else (the few of us) that are
> > building with MinGW experiencing similar issues on the HEAD revision of svn.
> >
> >
> > Using this configure command:
> > ./configure --prefix=/usr --disable-dbox2 --disable-hdhomerun
> > --disable-dvb -
> > -disable-ivtv --disable-iptv --disable-joystick-menu --disable-xvmc-vld
> > --disab
> > le-x11 --disable-xvmc --enable-directx --enable-memalign-hack
> > --compile-type=de
> > bug
> >
> > When I run make, this is the new error:
> > gcc -c -march=i586 -g -Wall -Wno-switch -Wdisabled-optimization
> > -Wpointer-arith -Wredundant-decls -fPIC -DPIC -mthreads -w -fexceptions
> > -frtti -DUSING_MINGW -DUSING_WINAUDIO -DUSING_D3D -DUSING_DIRECTX
> > -DUSING_DIRECTX -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr\"
> > -DLIBDIR=\"/usr/lib\" -DHAVE_AV_CONFIG_H -D_LARGEFILE_SOURCE -DQT_DLL
> > -DQT_THREAD_SUPPORT -I"/usr/include" -I".." -I"../.." -I"/qt- 3.3.x-p8/include"
> > -I"C:/msys/1.0/qt-3.3.x-p8/mkspecs/win32-g++" -o samplerate.o
> > samplerate.c
> > In file included from samplerate.c:27:
> > float_cast.h: In function `lrint':
> > float_cast.h:90: error: `_asm' undeclared (first use in this function)
> > float_cast.h:90: error: (Each undeclared identifier is reported only
> > once
> > float_cast.h:90: error: for each function it appears in.)
> > float_cast.h:90: error: syntax error before '{' token
> > float_cast.h: In function `lrintf':
> > float_cast.h:102: error: `_asm' undeclared (first use in this function)
> > float_cast.h:102: error: syntax error before '{' token
> > make[2]: *** [samplerate.o] Error 1
> > make[2]: Leaving directory `/c/mythtv/mythtv/libs/libmythsamplerate'
> > make[1]: *** [sub-libmythsamplerate] Error 2
> > make[1]: Leaving directory `/c/mythtv/mythtv/libs'
> > make: *** [sub-libs] Error 2
> >
> >
> > _______________________________________________
> > mythtv-dev mailing list
> > mythtv-dev [at] mythtv
> > http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
> >
> >
>

jmodi at vt

Jan 21, 2008, 1:29 PM

Post #5 of 24 (3707 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> ------------------------------------------
> 15434 -> is the last version to build successfully (with the known
> requisite win32 patches)
> ------------------------------------------
>
I can confirm that 15434 builds correctly for me also.

>
> 15435 -> 15477 all give this same compile error:
> In file included from uitypes.h:19,
> from managedlist.h:9,
> from managedlist.cpp:4:
> util.h:89: error: `LONGLONG' does not name a type
> make[2]: *** [managedlist.o] Error 1
> make[2]: Leaving directory `/c/mythtv/mythtv/libs/libmyth'
> make[1]: *** [sub-libmyth] Error 2
> make[1]: Leaving directory `/c/mythtv/mythtv/libs'
> make: *** [sub-libs] Error 2
>
15435 builds correctly for me as does 15477 with my remoteutil.patch. I did
not get the above error. Unsure about why that is.

> ------------------------------------------
> 15478 ->15514(HEAD) all give this same compile problem:
> In file included from httprequest.cpp:37:
> ../libmyth/util.h: At global scope:
> ../libmyth/util.h:90: error: expected initializer before "inline"
> In file included from ../libmyth/mythcontext.h:6,
> from httprequest.cpp:38:
> C:/msys/1.0/qt-3.3.x-p8/include/qpixmap.h:249: warning: 'void
> bitBlt(QPaintDevice*, int, int, const QPaintDevice*, int, int, int, int,
> Qt::RasterOp, bool)' redeclared without dllimport attribute after being
> referenced with dllimport linkage
> make[2]: *** [httprequest.o] Error 1
> make[2]: Leaving directory `/c/mythtv/mythtv/libs/libmythupnp'
> make[1]: *** [sub-libmythupnp] Error 2
> make[1]: Leaving directory `/c/mythtv/mythtv/libs'
> make: *** [sub-libs] Error 2
> ------------------------------------------
>

r15478 brings in the "as-needed" linker options and breaks the win32 builds.

g++ -Wl,-enable-auto-import -mthreads -Wl,-subsystem,windows -shared
-Wl,--export-all-symbols -o libmythupnp-0.20.dll object_ld_script
moc_ssdpcache.o -L"/qt-3.3.x-p8\lib" -L"/mingw/lib" -L"/mingw/lib" -lqtmain
-lfreetype -lmp3lame -lqt-mt -lkernel32 -luser32 -lgdi32 -lcomdlg32
-ladvapi32 -lshell32 -lole32 -loleaut32 -luuid -limm32 -lwinmm -lws2_32
-lwinspool
./httprequest.o: In function `ZN11HTTPRequest14SetRequestTypeERK7QString':
c:/mythtv/mythtv/libs/libmythupnp/httprequest.cpp:152: undefined reference
to `print_verbose_messages'
c:/mythtv/mythtv/libs/libmythupnp/httprequest.cpp:152: undefined reference
to `MythContext::verbose_mutex'
c:/mythtv/mythtv/libs/libmythupnp/httprequest.cpp:152: undefined reference
to `MythContext::verbose_mutex'
./httprequest.o: In function `ZN11HTTPRequest12SendResponseEv':
c:/mythtv/mythtv/libs/libmythupnp/httprequest.cpp:204: undefined reference
to `print_verbose_messages'

These messages continue on for a long time.

Now what is puzzling to me is that I am not getting the above errors that I
originally was getting last night about the typedef and expecting an
initializer. I definitely cleaned it up but the only thing I did differently
was using msys to do all of this instead on the Win32 script. I am going to
clean out the mythtv directory and start over again just to see if I can
reproduce this behavoir.

Attachments:

remoteutil.patch (0.32 KB)

nigel at ind

Jan 21, 2008, 3:40 PM

Post #6 of 24 (3724 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> r15478 brings in the "as-needed" linker options and breaks the
> win32 builds.
>
> g++ -Wl,-enable-auto-import -mthreads -Wl,-subsystem,windows -
> shared -Wl,--export-all-symbols -o libmythupnp-0.20.dll
> object_ld_script moc_ssdpcache.o -L"/qt-3.3.x-p8\lib" -L"/mingw/
> lib" -L"/mingw/lib" -lqtmain -lfreetype -lmp3lame -lqt-mt -
> lkernel32 -luser32 -lgdi32 -lcomdlg32 -ladvapi32 -lshell32 -lole32 -
> loleaut32 -luuid -limm32 -lwinmm -lws2_32 -lwinspool
> ./httprequest.o: In function
> `ZN11HTTPRequest14SetRequestTypeERK7QString':
> c:/mythtv/mythtv/libs/libmythupnp/httprequest.cpp:152: undefined
> reference to `print_verbose_messages'
> c:/mythtv/mythtv/libs/libmythupnp/httprequest.cpp:152: undefined
> reference to `MythContext::verbose_mutex'
> c:/mythtv/mythtv/libs/libmythupnp/httprequest.cpp:152: undefined
> reference to `MythContext::verbose_mutex'
> ./httprequest.o: In function `ZN11HTTPRequest12SendResponseEv':
> c:/mythtv/mythtv/libs/libmythupnp/httprequest.cpp:204: undefined
> reference to `print_verbose_messages'

Yes. Sorry about that.

If it is any consolation, I also broke
the Mac OS X builds with that checkin.

--
Nigel Pearson, nigel [at] ind|Well, I own the hotel
Telstra Net. Eng., Sydney, Australia | and I also live in it.
Office: 9202 3900 Fax: 9261 3912 |My life is very much
Mobile: 0408 664435 Home: 9792 6998 | like Monopoly

_______________________________________________
mythtv-dev mailing list
mythtv-dev [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev

jaymode at gmail

Jan 21, 2008, 4:07 PM

Post #7 of 24 (3727 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> Now what is puzzling to me is that I am not getting the above errors that
> I originally was getting last night about the typedef and expecting an
> initializer. I definitely cleaned it up but the only thing I did differently
> was using msys to do all of this instead on the Win32 script. I am going to
> clean out the mythtv directory and start over again just to see if I can
> reproduce this behavior.
>

I deleted the previous mythtv folder and rechecked out r15481, the last
revision dealing with the as-needed options and still do not get the errors
about the typedef or from remoteutil.cpp. I patched with the backend and
import icons patches only. The linking errors are still present.

I checked out HEAD and still no issues with util.h anymore. I was using the
win32 perl script, when I first got those errors(now I am not). Buzz what
were you using? I will try out the latest Win32 script.

davidbuzz at gmail

Jan 21, 2008, 4:19 PM

Post #8 of 24 (3711 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

Jay,
my initial compile issue manifested by using the win32- script, but I did
the divide-and-conquer compiliations in msys with the same results. It
shouldn't make any difference (and didn't to me), becuase the win32-packager
script actually executes the shell commands in a msys bash shell, so it's
the same basic setup either way.

FYI - there are still some minor tweeks needed with the win32-packager
script, so be warned that a "from scratch" setup isn't entirely hands-off
just yet, but it's not doing anything wrong AFAIK, it just "drops out" when
it can't figure out what to do next, which I believe is the right thing to
do.

Buzz.
On Jan 22, 2008 10:07 AM, Jay Modi <jaymode [at] gmail> wrote:

>
> Now what is puzzling to me is that I am not getting the above errors that
> > I originally was getting last night about the typedef and expecting an
> > initializer. I definitely cleaned it up but the only thing I did differently
> > was using msys to do all of this instead on the Win32 script. I am going to
> > clean out the mythtv directory and start over again just to see if I can
> > reproduce this behavior.
> >
>
> I deleted the previous mythtv folder and rechecked out r15481, the last
> revision dealing with the as-needed options and still do not get the errors
> about the typedef or from remoteutil.cpp. I patched with the backend and
> import icons patches only. The linking errors are still present.
>
> I checked out HEAD and still no issues with util.h anymore. I was using
> the win32 perl script, when I first got those errors(now I am not). Buzz
> what were you using? I will try out the latest Win32 script.
>
>
>

jaymode at gmail

Jan 21, 2008, 4:56 PM

Post #9 of 24 (3710 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

>
> Jay,
> my initial compile issue manifested by using the win32- script, but I did
> the divide-and-conquer compiliations in msys with the same results. It
> shouldn't make any difference (and didn't to me), becuase the win32-packager
> script actually executes the shell commands in a msys bash shell, so it's
> the same basic setup either way.
>

Ok well I think I may have figured it out. In the configure step can you
take out --cpu=k8 and then retry? That seems to be what is causing the
issues for me. I think this flag should not be in the script because not
everyone is running an AMD processor. I am using an Intel Pentium M and a
Pentium D.

> FYI - there are still some minor tweeks needed with the win32-packager
> script, so be warned that a "from scratch" setup isn't entirely hands-off
> just yet, but it's not doing anything wrong AFAIK, it just "drops out" when
> it can't figure out what to do next, which I believe is the right thing to
> do.
>

I agree that it should drop out when something unexpected happens.

jaymode at gmail

Jan 21, 2008, 6:47 PM

Post #10 of 24 (3700 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> Yes. Sorry about that.
>
> If it is any consolation, I also broke
> the Mac OS X builds with that checkin.
>

Any idea on how we could fix this issue? I don't know too much about this
linking and linker commands.

nigel at ind

Jan 21, 2008, 7:15 PM

Post #11 of 24 (3703 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> Any idea on how we could fix this issue?

Create a libs/libmyth/libmyth.def containing all the symbols
from that library that libmythui and libmythupnp require,
and add rules in the latter two directories' .pro files.
Something like:

mingw {
target.path = $${PREFIX}/bin
implib.target = libmyth-bootstrap.a
implib.commands = test -e libmyth-bootstrap.a || dlltool --input-
def ../libmyth/libmyth.def --dllname libmyth-$${LIBVERSION}.dll --
output-lib libmyth-bootstrap.a -k
QMAKE_EXTRA_WIN_TARGETS += implib
POST_TARGETDEPS += libmyth-bootstrap.a
}

The existing commands like that can probably
also be removed from libs/libmyth/libmyth.pro

I'm trying to setup a Win32 build env now,
but being a Mac and Unix head means that
this is an exercise in cursing and grumbling.
You might be able to fix & test faster than me.

--
Nigel Pearson, nigel [at] ind| I've been talking to-
Telstra Net. Eng., Sydney, Australia | the ship's computer.
Office: 9202 3900 Fax: 9261 3912 | ... It hates me!
Mobile: 0408 664435 Home: 9792 6998 | Marvin

_______________________________________________
mythtv-dev mailing list
mythtv-dev [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev

davidbuzz at gmail

Jan 21, 2008, 7:28 PM

Post #12 of 24 (3711 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

>
> I'm trying to setup a Win32 build env now,
> but being a Mac and Unix head means that
> this is an exercise in cursing and grumbling.
> You might be able to fix & test faster than me.
>
Seriously consider using the
http://svn.mythtv.org/svn/trunk/mythtv/contrib/Win32/win32-packager.pl to
get the guts of the system "up and running", it's much easier. (many hours
less work!).

you need:
http://downloads.activestate.com/ActivePerl/Windows/5.8/ActivePerl-5.8.8.822-MSWin32-x86-280952.msi

and http://www.mythtv.org/wiki/index.php/Windows_Port

Buzz.

jmodi at vt

Jan 21, 2008, 7:54 PM

Post #13 of 24 (3721 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

Seriously consider using the
http://svn.mythtv.org/svn/trunk/mythtv/contrib/Win32/win32-packager.pl
> to get the guts of the system "up and running", it's much easier.
> (many hours less work!).
>
> you need: http://downloads.activestate.com/ActivePerl/Windows/5.8/ActivePerl-5.8.8.822-MSWin32-x86-280952.msi
>
>
> and http://www.mythtv.org/wiki/index.php/Windows_Port
>
> Buzz.
>

Agreed. The script works very well. I have done it manually and it is pretty
painstaking.

nigel at ind

Jan 21, 2008, 7:55 PM

Post #14 of 24 (3712 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> Seriously consider using the http://svn.mythtv.org/svn/trunk/mythtv/
> contrib/Win32/win32-packager.pl to get the guts of the system "up
> and running", it's much easier. (many hours less work!).
>
> you need: http://downloads.activestate.com/ActivePerl/Windows/5.8/
> ActivePerl-5.8.8.822-MSWin32-x86-280952.msi
>
> and http://www.mythtv.org/wiki/index.php/Windows_Port

And a new VMWare volume with enough space on it,
and a proxy workaround
(my work firewall requires authentication)
and a caffeine hit,
and more hours in the day
...

--
Nigel Pearson, nigel [at] ind|The weak point of
Telstra Net. Eng., Sydney, Australia | the modern car is
Office: 9202 3900 Fax: 9261 3912 | the squidgy organic
Mobile: 0408 664435 Home: 9792 6998 | bit behind the wheel.
_______________________________________________
mythtv-dev mailing list
mythtv-dev [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev

davidbuzz at gmail

Jan 21, 2008, 8:05 PM

Post #15 of 24 (3719 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

Hmm, yes, I see your point, feel free to ask me question off-line if you
need help. :-)

On Jan 22, 2008 1:55 PM, Nigel Pearson <nigel [at] ind> wrote:

> > Seriously consider using the http://svn.mythtv.org/svn/trunk/mythtv/
> > contrib/Win32/win32-packager.pl to get the guts of the system "up
> > and running", it's much easier. (many hours less work!).
> >
> > you need: http://downloads.activestate.com/ActivePerl/Windows/5.8/
> > ActivePerl-5.8.8.822-MSWin32-x86-280952.msi
> >
> > and http://www.mythtv.org/wiki/index.php/Windows_Port
>
> And a new VMWare volume with enough space on it,
> and a proxy workaround
> (my work firewall requires authentication)
> and a caffeine hit,
> and more hours in the day
> ...
>
> --
> Nigel Pearson, nigel [at] ind|The weak point of
> Telstra Net. Eng., Sydney, Australia | the modern car is
> Office: 9202 3900 Fax: 9261 3912 | the squidgy organic
> Mobile: 0408 664435 Home: 9792 6998 | bit behind the wheel.
>

davidbuzz at gmail

Jan 21, 2008, 8:07 PM

Post #16 of 24 (3707 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

Would a free tool to automatically generate the .def file/s help?

(like this one: http://www.objmedia.demon.co.uk/freeSoftware/def32.html )

I haven't figured out the magical incantactions to make it play nice, but it
looks to have potential.

Buzz.

On Jan 22, 2008 1:15 PM, Nigel Pearson <nigel [at] ind> wrote:

> > Any idea on how we could fix this issue?
>
> Create a libs/libmyth/libmyth.def containing all the symbols
> from that library that libmythui and libmythupnp require,
> and add rules in the latter two directories' .pro files.
> Something like:
>
> mingw {
> target.path = $${PREFIX}/bin
> implib.target = libmyth-bootstrap.a
> implib.commands = test -e libmyth-bootstrap.a || dlltool --input-
> def ../libmyth/libmyth.def --dllname libmyth-$${LIBVERSION}.dll --
> output-lib libmyth-bootstrap.a -k
> QMAKE_EXTRA_WIN_TARGETS += implib
> POST_TARGETDEPS += libmyth-bootstrap.a
> }
>
>
>
> The existing commands like that can probably
> also be removed from libs/libmyth/libmyth.pro
>
>
>
>
> I'm trying to setup a Win32 build env now,
> but being a Mac and Unix head means that
> this is an exercise in cursing and grumbling.
> You might be able to fix & test faster than me.
>
> --
> Nigel Pearson, nigel [at] ind| I've been talking to-
> Telstra Net. Eng., Sydney, Australia | the ship's computer.
> Office: 9202 3900 Fax: 9261 3912 | ... It hates me!
> Mobile: 0408 664435 Home: 9792 6998 | Marvin
>
>
> _______________________________________________
> mythtv-dev mailing list
> mythtv-dev [at] mythtv
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
>

andrei at tanas

Jan 21, 2008, 9:21 PM

Post #17 of 24 (3710 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

<6664ae760801202006n56387586wae4022acd3959d6f [at] mail>
<705088780801202156o6bd75235v8969a30407160461 [at] mail>
<705088780801210354i6038ef47ie89939db25049dd3 [at] mail>
<6664ae760801211329m2f37876ej37cc40928d7f6b1a [at] mail>
<1850C2AD-419B-435B-B87D-EE42749E988D [at] ind>
<6664ae760801211847i1d2034dax8738b01f874cbe0e [at] mail>
<5CEEE766-B08F-4E4E-9E32-0FBEA5F42BE1 [at] ind> <705088780801212007w3290f5a8hd3e591877a44b260 [at] mail>
Message-ID: <95eb2f2719c83a7ef2f4ef9a7b3449fd [at] localhos>
X-Sender: andrei [at] tanas
User-Agent: RoundCube Webmail/0.1b
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 8bit

First of all, I'll try to bring patches for the current head up to speed
tomorrow. As far as I understand one of the old issues came back, so it
shouldn't bee too hard to fix this.

On Tue, 22 Jan 2008 14:07:23 +1000, buzz <davidbuzz [at] gmail> wrote:
> Would a free tool to automatically generate the .def file/s help?
>
> (like this one: http://www.objmedia.demon.co.uk/freeSoftware/def32.html )
>
> I haven't figured out the magical incantactions to make it play nice, but
> it
> looks to have potential.

It's the "chicken and egg" situation. You can't generate .def files using
this tool (or other tools) because in theory, you don't have any .dll's yet
(you do from previous builds but that's not the point).

This is why this "bootstrapping" - creating of import libraries from .def
files was invented in the first place, and this is how I solved initial
circular dependencies.

> On Jan 22, 2008 1:15 PM, Nigel Pearson <nigel [at] ind> wrote:
>
>> I'm trying to setup a Win32 build env now,
>> but being a Mac and Unix head means that
>> this is an exercise in cursing and grumbling.
>> You might be able to fix & test faster than me.

Nigel, if you need help setting up Windows environment, please feel free to
e-mail me directly.

_______________________________________________
mythtv-dev mailing list
mythtv-dev [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev

davidbuzz at gmail

Jan 21, 2008, 9:50 PM

Post #18 of 24 (3703 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

Andrei,

> It's the "chicken and egg" situation. You can't generate .def files using
> this tool (or other tools) because in theory, you don't have any .dll's
> yet
> (you do from previous builds but that's not the point).
>
<hopefully not too nieve a question>
Well actually, I think that is the point... Since we do have dll's being
successfully built (for some SVN revisions), why not just create a .def file
automatically from these? (using dll2def or pexports?) If you want to be
thorough, you could re-create the .def file after a successful .dll is
built, and emit some meaningful warning/error message if it's different than
the one used the first-time-round?

wouldn't that work?
<end question>

>
>
> This is why this "bootstrapping" - creating of import libraries from .def
> files was invented in the first place, and this is how I solved initial
> circular dependencies.

Yes, and we all win32-ers thankyou for it! :-)

Buzz.

andrei at tanas

Jan 21, 2008, 9:55 PM

Post #19 of 24 (3703 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> Ok well I think I may have figured it out. In the configure step can you take out --cpu=k8 and
> then retry? That seems to be what is causing the issues for me. I think this flag
> should not be in the script because not everyone is running an AMD processor. I am
> using an Intel Pentium M and a Pentium D.

Cpu type autodetection is broken in configure script, when run in MinGW. If you don't use cpu=k8, you will get perfectly compatible i686 build, however, without MMX or SSE support. You can expect video codecs, colorspace converters etc to run several times slower.
"K8" is suitable for Pentium 4 and Core Duo, not just AMD CPUs. If you are using Pentium M, try to find out which CPU setting you should use, but don't just disable it.

Sent from BlackBerry�
_______________________________________________
mythtv-dev mailing list
mythtv-dev [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev

davidbuzz at gmail

Jan 21, 2008, 9:59 PM

Post #20 of 24 (3702 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

I never had a problem with cpu=k8 on Pentium D or on "Core 2 Duo". Although
I couldn't call it "fast", I haven't done proper comparisons. ('Core 2
Duo' is aka Centrino, aka X86_FAMILY_6_MODEL_15_0)

On Jan 22, 2008 3:55 PM, Andrei Tanas <andrei [at] tanas> wrote:

> > Ok well I think I may have figured it out. In the configure step can you
> take out --cpu=k8 and
> > then retry? That seems to be what is causing the issues for me. I think
> this flag
> > should not be in the script because not everyone is running an AMD
> processor. I am
> > using an Intel Pentium M and a Pentium D.
>
> Cpu type autodetection is broken in configure script, when run in MinGW.
> If you don't use cpu=k8, you will get perfectly compatible i686 build,
> however, without MMX or SSE support. You can expect video codecs, colorspace
> converters etc to run several times slower.
> "K8" is suitable for Pentium 4 and Core Duo, not just AMD CPUs. If you are
> using Pentium M, try to find out which CPU setting you should use, but don't
> just disable it.
>
> Sent from BlackBerry(R)
> _______________________________________________
> mythtv-dev mailing list
> mythtv-dev [at] mythtv
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
>

davidbuzz at gmail

Jan 21, 2008, 10:08 PM

Post #21 of 24 (3698 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

I autocreated a libmyth.def file for testing, and I wondered if its
acceptable (I have no idea about these things)?
its here: http://davidbuzz.googlepages.com/libmyth.def

On Jan 22, 2008 1:15 PM, Nigel Pearson <nigel [at] ind> wrote:

> > Any idea on how we could fix this issue?
>
> Create a libs/libmyth/libmyth.def containing all the symbols
> from that library that libmythui and libmythupnp require,
> and add rules in the latter two directories' .pro files.
> Something like:
>
> mingw {
> target.path = $${PREFIX}/bin
> implib.target = libmyth-bootstrap.a
> implib.commands = test -e libmyth-bootstrap.a || dlltool --input-
> def ../libmyth/libmyth.def --dllname libmyth-$${LIBVERSION}.dll --
> output-lib libmyth-bootstrap.a -k
> QMAKE_EXTRA_WIN_TARGETS += implib
> POST_TARGETDEPS += libmyth-bootstrap.a
> }
>
>
>
> The existing commands like that can probably
> also be removed from libs/libmyth/libmyth.pro
>
>
>
>
> I'm trying to setup a Win32 build env now,
> but being a Mac and Unix head means that
> this is an exercise in cursing and grumbling.
> You might be able to fix & test faster than me.
>
> --
> Nigel Pearson, nigel [at] ind| I've been talking to-
> Telstra Net. Eng., Sydney, Australia | the ship's computer.
> Office: 9202 3900 Fax: 9261 3912 | ... It hates me!
> Mobile: 0408 664435 Home: 9792 6998 | Marvin
>
>
> _______________________________________________
> mythtv-dev mailing list
> mythtv-dev [at] mythtv
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
>

jaymode at gmail

Jan 22, 2008, 4:12 AM

Post #22 of 24 (3696 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> Cpu type autodetection is broken in configure script, when run in MinGW.
> If you don't use cpu=k8, you will get perfectly compatible i686 build,
> however, without MMX or SSE support. You can expect video codecs, colorspace
> converters etc to run several times slower.
> "K8" is suitable for Pentium 4 and Core Duo, not just AMD CPUs. If you are
> using Pentium M, try to find out which CPU setting you should use, but don't
> just disable it.
>

The cpu=k8 is what had broken the builds probably due to some of the mmx/sse
code being brought in; nothing was ever wrong with the binaries when I had
used it. I will try it out using the pentium-m flag. I suspect this will
still break it because it will bring in both MMX and SSE support, which is
where some of the issues were.

http://gentoo-wiki.com/Safe_Cflags
is a good place to find out what --cpu= you can use. Just take the correct
-march= for your processor. In my case it was just --cpu=pentium-m

jerrymr at gmail

Jan 22, 2008, 6:10 AM

Post #23 of 24 (3686 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

On Jan 22, 2008 12:50 AM, buzz <davidbuzz [at] gmail> wrote:

> Andrei,
>
> > It's the "chicken and egg" situation. You can't generate .def files
> > using
> > this tool (or other tools) because in theory, you don't have any .dll's
> > yet
> > (you do from previous builds but that's not the point).
> >
> <hopefully not too nieve a question>
> Well actually, I think that is the point... Since we do have dll's being
> successfully built (for some SVN revisions), why not just create a .def file
> automatically from these? (using dll2def or pexports?) If you want to be
> thorough, you could re-create the .def file after a successful .dll is
> built, and emit some meaningful warning/error message if it's different than
> the one used the first-time-round?
>
> wouldn't that work?
> <end question>
>

No, because any time the signature for any exported function changes, you'd
have a .def file generated from a previous version (without the change),
trying to use it to link code with the change.

Suppose in version 1 you had function int foo(int a). You generate a .def
file from that. Then in version 2, foo gets a second parameter: int foo(int
a, bool b). Version 2 is not yet compiled, so you use the .def file
generated from V1, and it's not going to link.

-Jerry

>

andrei at tanas

Jan 22, 2008, 7:37 AM

Post #24 of 24 (3690 views)
Permalink

Re: Ticket #4497: Win32 builds broken after r15435 [In reply to]

> First of all, I'll try to bring patches for the current head up to
> speed
> tomorrow. As far as I understand one of the old issues came back, so it
> shouldn't bee too hard to fix this.

See ticket 4516 (http://svn.mythtv.org/trac/ticket/4516) for the patch to
build 15533 on MinGW.

_______________________________________________
mythtv-dev mailing list
mythtv-dev [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads