From: Frantisek D. <va...@us...> - 2003-12-31 18:13:59
|
Hi team, xine is compiled by cygwin quite nicely. I'll commit necessary patches to Makefiles. I have some questions/requests about other patches: 1) cygwin_dvdnav.diff - There differs types in header files and .c files. It's external project, can somebody please commit the patch into libdvdnav CVS and xine CVS? 2) cygwin_goom.diff - The same problem. I sent the patch somewhere into goom project some months a go (yes, at the time I tried cygwin too), but I don't know if the target place was right :-) and I haven't checked if somebody used the patch. 3) cygwin_vcd.diff - '-lwinmm' was required for compiling VCD plugin. The patch looks and works fine, if nobody will want some different solution, I'll commit this into xine CVS. 4) I can't link src/post/deinterlace/plugins/greedy2frame_template.c - unresolved symbols. I don't know how to compile it with the assembler code inside. I attached compilation errors, symbols of the libdeinterlaceplugins.a under Cygwin and the symbols under Linux (where it is OK). Thanks for support. Cheers, Frantisek |
From: Miguel F. <mi...@ce...> - 2003-12-31 18:35:10
|
On Wed, 2003-12-31 at 13:14, Frantisek Dvorak wrote: > Hi team, > > xine is compiled by cygwin quite nicely. I'll commit necessary patches > to Makefiles. cool! ;-) > 4) I can't link src/post/deinterlace/plugins/greedy2frame_template.c - > unresolved symbols. I don't know how to compile it with the assembler > code inside. I attached compilation errors, symbols of the > libdeinterlaceplugins.a under Cygwin and the symbols under Linux (where > it is OK). this file is not compiled directly, it is included from greedy2frame.c with some defines set to select a particular optimization (mmx, sse, etc). it looks like the static variables in there are not being defined, as if MASKS_DEFINED was already defined beforehand. maybe you should try removing the line: #if !defined(MASKS_DEFINED) (and the closing #endif) to see if it fixes the problem. i have no idea why it would fail with cygwin thought. regards, Miguel |
From: Frantisek D. <va...@us...> - 2003-12-31 18:49:27
|
V St, 31. 12. 2003 v 22:43, Miguel Freitas p=C3=AD=C5=A1e: > > 4) I can't link src/post/deinterlace/plugins/greedy2frame_template.c= - > > unresolved symbols. I don't know how to compile it with the assembler > > code inside. I attached compilation errors, symbols of the > > libdeinterlaceplugins.a under Cygwin and the symbols under Linux (whe= re > > it is OK). >=20 > this file is not compiled directly, it is included from greedy2frame.c > with some defines set to select a particular optimization (mmx, sse, > etc). >=20 > it looks like the static variables in there are not being defined, as i= f > MASKS_DEFINED was already defined beforehand. maybe you should try > removing the line: >=20 > #if !defined(MASKS_DEFINED) > (and the closing #endif) >=20 > to see if it fixes the problem. i have no idea why it would fail with > cygwin thought. >=20 I already tried it. And there are warnings inside the ifdef. :-) Also I tried to remove 'static', 'const', change types to 'int32_t'... Library tools must do some additional things for exporting data variables under Windows... Frantisek |
From: Frantisek D. <va...@us...> - 2004-01-01 18:20:22
Attachments:
asm.c
|
V St, 31. 12. 2003 v 22:43, Miguel Freitas p=C3=AD=C5=A1e: > > 4) I can't link src/post/deinterlace/plugins/greedy2frame_template.c= - > > unresolved symbols. I don't know how to compile it with the assembler > > code inside. I attached compilation errors, symbols of the > > libdeinterlaceplugins.a under Cygwin and the symbols under Linux (whe= re > > it is OK). >=20 > to see if it fixes the problem. i have no idea why it would fail with > cygwin thought. >=20 I attached minimalized file, which I can compile under Linux but not under Cygwin. It's due to assembler...? Cheers, Frantisek |
From: Miguel F. <mi...@ce...> - 2004-01-02 20:52:43
|
On Thu, 2004-01-01 at 13:20, Frantisek Dvorak wrote: > I attached minimalized file, which I can compile under Linux but not > under Cygwin. It's due to assembler...? yes, sort of... looks like a name mangling problem: cygwin prepends a '_' to the variable names so they don't match the asm. as a small contribution to the win32 port i added something to cvs that should fix your compilation ;-) regards, Miguel |
From: Frantisek D. <va...@us...> - 2004-01-02 22:37:28
|
V So, 03. 01. 2004 v 01:01, Miguel Freitas p=C3=AD=C5=A1e: > On Thu, 2004-01-01 at 13:20, Frantisek Dvorak wrote: > > I attached minimalized file, which I can compile under Linux but not > > under Cygwin. It's due to assembler...? >=20 > yes, sort of... looks like a name mangling problem: cygwin prepends a > '_' to the variable names so they don't match the asm. >=20 > as a small contribution to the win32 port i added something to cvs that > should fix your compilation ;-) >=20 Oh thanks, I'll try. :-) |
From: Miguel F. <mi...@ce...> - 2003-12-31 18:43:38
|
On Wed, 2003-12-31 at 13:14, Frantisek Dvorak wrote: > Hi team, > > xine is compiled by cygwin quite nicely. I'll commit necessary patches > to Makefiles. btw, if you care, last time i tried compiling libxine for cygwin it seemed that a good plan for getting xine under windows fast would be: - check with vlc (or maybe mplayer?) about how they do about directx headers. i heard we are supposed to get a couple of headers from a M$ sdk into cygwin in order to compile the directx drivers. - code a very simple frontend for windows using one of these RAD apps. MSVC or Kylix should work fine, to do more like a prof of concept thing. - check mingw. i never tried it, but people say it might be better than cygwin in our case. regards, Miguel |
From: Frantisek D. <va...@us...> - 2003-12-31 18:58:50
|
V St, 31. 12. 2003 v 22:51, Miguel Freitas p=C3=AD=C5=A1e: > btw, if you care, last time i tried compiling libxine for cygwin it > seemed that a good plan for getting xine under windows fast would be: >=20 > - check with vlc (or maybe mplayer?) about how they do about directx > headers. i heard we are supposed to get a couple of headers from a M$ > sdk into cygwin in order to compile the directx drivers.=20 >=20 I'll try. I already have sources of mplayer. > - code a very simple frontend for windows using one of these RAD apps. > MSVC or Kylix should work fine, to do more like a prof of concept thing. >=20 Some RAD tool is very good idea. But I also peep at gxine. :-) > - check mingw. i never tried it, but people say it might be better than > cygwin in our case. >=20 I have it - mingw and msys. I haven't worked on/downloaded support for threads yet. Cheers, Frantisek |
From: Gildas B. <gb...@al...> - 2004-01-02 13:28:02
|
On Wednesday 31 December 2003 22:51, Miguel Freitas wrote: > > btw, if you care, last time i tried compiling libxine for cygwin it > seemed that a good plan for getting xine under windows fast would be: > > - check with vlc (or maybe mplayer?) about how they do about directx > headers. i heard we are supposed to get a couple of headers from a M$ > sdk into cygwin in order to compile the directx drivers. > Here is the package we use for VLC: http://download.videolan.org/pub/videolan/contrib/win32-dx7headers.tgz AFAIR it was originally found on Peter Puck's web site (which is apparently down). > > - check mingw. i never tried it, but people say it might be better than > cygwin in our case. > mingw is not really better than cygwin. Compiler wise they are almost identical if you use the -mno-cygwin (no posix emulation layer) with cygwin. Mingw is certainly lighter to download/install but cygwin provides a whole unix environment so should be easier to use in your case. -- Gildas |
From: Miguel F. <mi...@ce...> - 2004-01-02 21:08:19
|
Hi Gildas, On Fri, 2004-01-02 at 08:29, Gildas Bazin wrote: > Here is the package we use for VLC: > http://download.videolan.org/pub/videolan/contrib/win32-dx7headers.tgz > AFAIR it was originally found on Peter Puck's web site (which is apparently > down). > > > > > - check mingw. i never tried it, but people say it might be better than > > cygwin in our case. > > > > mingw is not really better than cygwin. Compiler wise they are almost > identical if you use the -mno-cygwin (no posix emulation layer) with > cygwin. Mingw is certainly lighter to download/install but cygwin provides > a whole unix environment so should be easier to use in your case. wow! thanks a lot for your comments! :) i don't use windows machines myself, but i will keep that in mind just in case i give xine/win32 another go... regards, Miguel |
From: Frantisek D. <va...@us...> - 2004-01-02 22:37:21
Attachments:
win32_visual.diff
|
Hi Miguel, V St, 31. 12. 2003 v 22:51, Miguel Freitas p=C3=AD=C5=A1e: > btw, if you care, last time i tried compiling libxine for cygwin it > seemed that a good plan for getting xine under windows fast would be: >=20 > - check with vlc (or maybe mplayer?) about how they do about directx > headers. i heard we are supposed to get a couple of headers from a M$ > sdk into cygwin in order to compile the directx drivers.=20 >=20 > - code a very simple frontend for windows using one of these RAD apps. > MSVC or Kylix should work fine, to do more like a prof of concept thing. >=20 Because I don't plan code some new frontend, I quick adjusted MSVC port f= or cygwin. It's little unmaintained, but good for testing :-). Is there v= olunteer for work on it or make new powerful frontend? Also I have question about win32_visual_t structure. Should it be inside = xine.h? Now it's inside 'video_out/video_out_win32.h', but this file isn'= t installed by 'make install' (see win32_visual.diff). Cheers, Frantisek PS: Both - MSVC ports and cygwin port quite work (but there remains big = work yet and indeed some not-comitted changes). |
From: Miguel F. <mi...@ce...> - 2004-01-02 23:49:10
|
On Fri, 2004-01-02 at 17:37, Frantisek Dvorak wrote: > Also I have question about win32_visual_t structure. Should it be > inside xine.h? yes. please commit... ;) regards, Miguel |
From: Frantisek D. <va...@us...> - 2004-01-02 22:47:29
|
Hi, V P=C3=A1, 02. 01. 2004 v 14:29, Gildas Bazin p=C3=AD=C5=A1e: > Here is the package we use for VLC: > http://download.videolan.org/pub/videolan/contrib/win32-dx7headers.tgz > AFAIR it was originally found on Peter Puck's web site (which is appare= ntly=20 > down). >=20 What about licensing? Can it be directly included in CVS? Many thanks for info. Cheers Frantisek |
From: Miguel F. <mi...@ce...> - 2004-01-02 23:07:10
|
On Fri, 2004-01-02 at 17:47, Frantisek Dvorak wrote: > Hi, >=20 > V P=E1, 02. 01. 2004 v 14:29, Gildas Bazin p=ED=B9e: > > Here is the package we use for VLC: > > http://download.videolan.org/pub/videolan/contrib/win32-dx7headers.tgz > > AFAIR it was originally found on Peter Puck's web site (which is appare= ntly=20 > > down). > >=20 >=20 > What about licensing? Can it be directly included in CVS? I guess not, i see Microsoft copyrights everywhere in these files.=20 While i'm still guessing, maybe it would be possible to "reimplement" only the portions of these headers we need. Like creating the declaration of the used functions and a few constants/defines so it could be compiled out of the box... regards, Miguel |
From: Frantisek D. <va...@us...> - 2004-01-03 20:30:36
|
V So, 03. 01. 2004 v 03:15, Miguel Freitas p=C3=AD=C5=A1e: > On Fri, 2004-01-02 at 17:47, Frantisek Dvorak wrote: > > Hi, > >=20 > > V P=C3=A1, 02. 01. 2004 v 14:29, Gildas Bazin p=C3=AD=C5=A1e: > > > Here is the package we use for VLC: > > > http://download.videolan.org/pub/videolan/contrib/win32-dx7headers.= tgz > > > AFAIR it was originally found on Peter Puck's web site (which is ap= parently=20 > > > down). > > >=20 > >=20 > > What about licensing? Can it be directly included in CVS? >=20 > I guess not, i see Microsoft copyrights everywhere in these files.=20 > While i'm still guessing, maybe it would be possible to "reimplement" > only the portions of these headers we need. Like creating the > declaration of the used functions and a few constants/defines so it > could be compiled out of the box... >=20 Good idea. Btw. patch "win32_visual" commited. And your changes into greedy helped, now I know why the MANGLE in the ffmpeg header files. :-) Cheers, Frantisek |
From: James Courtier-D. <Ja...@su...> - 2003-12-31 21:53:38
|
Frantisek Dvorak wrote: > Hi team, > > xine is compiled by cygwin quite nicely. I'll commit necessary patches > to Makefiles. > > I have some questions/requests about other patches: > > 1) cygwin_dvdnav.diff - There differs types in header files and .c > files. It's external project, can somebody please commit the patch into > libdvdnav CVS and xine CVS? > This has been checked into libdvdnav CVS by me now. You can test xine-lib with the external libdvdnav by building xine-lib with "./configure --with-external-dvdnav" If you can get libdvdcss to compile and work under cygwin, that would also be nice. Cheers James |
From: Frantisek D. <va...@us...> - 2004-01-01 18:20:28
|
V St, 31. 12. 2003 v 22:59, James Courtier-Dutton p=C3=AD=C5=A1e: > Frantisek Dvorak wrote: > > Hi team, > >=20 > > xine is compiled by cygwin quite nicely. I'll commit necessary patche= s > > to Makefiles. > >=20 > > I have some questions/requests about other patches: > >=20 > > 1) cygwin_dvdnav.diff - There differs types in header files and .c > > files. It's external project, can somebody please commit the patch in= to > > libdvdnav CVS and xine CVS? > >=20 > This has been checked into libdvdnav CVS by me now. >=20 > You can test xine-lib with the external libdvdnav by building xine-lib=20 > with "./configure --with-external-dvdnav" >=20 Thanks, I'll try it. > If you can get libdvdcss to compile and work under cygwin, that would=20 > also be nice. >=20 I can't use libdvdcss in my country, but I'll try compile it under cygwin. :-) Cheers, Frantisek |
From: Frantisek D. <va...@us...> - 2004-01-02 22:37:16
Attachments:
msvc_dvdnav.diff
cygwin_dvdnav.log.bz2
|
Hi James, V St, 31. 12. 2003 v 22:59, James Courtier-Dutton p=C3=AD=C5=A1e: > >=20 > > 1) cygwin_dvdnav.diff - There differs types in header files and .c > > files. It's external project, can somebody please commit the patch in= to > > libdvdnav CVS and xine CVS? > >=20 > This has been checked into libdvdnav CVS by me now. >=20 > You can test xine-lib with the external libdvdnav by building xine-lib=20 > with "./configure --with-external-dvdnav" >=20 I had problems with compilation of separate libdvdnav under cygwin. Somet= hing is bad with libtool - result is .libs/libdvdnav.a which takes 8 byte= s. When I manually removed '-rpath /usr/lib/locale', it was made right .a= file, but still wasn't made shared version. I attached build log of cygw= in, if there is some libtool guru here, who know how to fix it (cygwin_dv= dnav.log.bz2). And my patch broke MSVC compilation - missing header. I send the fix (msv= c_dvdnav.diff). Cheers, Frantisek |
From: Michael R. <mr...@us...> - 2004-01-12 19:40:16
|
Hi Frantisek, > I had problems with compilation of separate libdvdnav under cygwin. > Something is bad with libtool - result is .libs/libdvdnav.a which > takes 8 bytes. When I manually removed '-rpath /usr/lib/locale', it > was made right .a file, but still wasn't made shared version. I > attached build log of cygwin, if there is some libtool guru here, who > know how to fix it (cygwin_dvdnav.log.bz2). > > And my patch broke MSVC compilation - missing header. I send the fix > (msvc_dvdnav.diff). I forgot to follow this thread. Just wanted to notify you that I made some changes to the build system of libdvdnav. It would be really cool, of you could repeat your windows comilation tests (cygwin, msvc, any other magic you try) and tell me, how it goes and if some patches are still needed. Michael -- printk(KERN_WARNING "Multi-volume CD somehow got mounted.\n"); 2.2.16 /usr/src/linux/fs/isofs/inode.c |
From: Michael R. <mr...@us...> - 2004-01-01 13:31:04
|
Hi Frantisek, > xine is compiled by cygwin quite nicely. I'll commit necessary > patches to Makefiles. Thanks for your effort. > I have some questions/requests about other patches: > > 1) cygwin_dvdnav.diff - There differs types in header files and .c > files. It's external project, can somebody please commit the patch > into libdvdnav CVS and xine CVS? Just being curious: Why did the original int's have to be replaced with int32_t? Michael -- if (user_specified) /* Didn't work, but the user is convinced this is the * place. */ 2.4.0-test2 /usr/src/linux/drivers/parport/parport_pc.c |
From: James Courtier-D. <Ja...@su...> - 2004-01-01 14:31:30
|
Michael Roitzsch wrote: > Hi Frantisek, > > >>xine is compiled by cygwin quite nicely. I'll commit necessary >>patches to Makefiles. > > > Thanks for your effort. > > >>I have some questions/requests about other patches: >> >> 1) cygwin_dvdnav.diff - There differs types in header files and .c >>files. It's external project, can somebody please commit the patch >>into libdvdnav CVS and xine CVS? > > > Just being curious: Why did the original int's have to be replaced with > int32_t? > > Michael > Michael, There was a mismatch which linux gcc did not pick up on. I.E. .h file had int, .c file had int32_t Now they both have int32_t This change should not affect the ABI. Cheers James |
From: Michael R. <mr...@us...> - 2004-01-01 14:55:31
|
Hi James, > There was a mismatch which linux gcc did not pick up on. > I.E. .h file had int, .c file had int32_t > Now they both have int32_t Ahh, I see. Thanks for clarifying. > This change should not affect the ABI. Except for machines, where int is 64 bit. Michael -- #ifdef STUPIDLY_TRUST_BROKEN_PCMD_ENA_BIT 2.4.0-test2 /usr/src/linux/drivers/ide/cmd640.c |