openvip-devel — OpenVIP development is discussed here

Re: [OpenVIP-devel] What is coretypes.h?

[Vaclav S. <vs...@fa...>]

Hi,

wild guess: try GNU make (gmake), the .idl -> .h rules appear to use 
GNU syntax...

HTH,
Vaclav

Alexandre Jouravlev wrote:
> Hello, when trying to compile upf, I have the following problem:
>
> Making all in include
> Making all in upf
> make: don't know how to make coretypes.h. Stop
> *** Error code 1
> Stop in /usr/sK0T/OpenVIP/upf/include.
> *** Error code 1
>
> The system is FreeBSD, the sources are from todays CVS.
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by:
> Power Architecture Resource Center: Free content, downloads,
> discussions, and more. http://solutions.newsforge.com/ibmarch.tmpl
> _______________________________________________
> Openvip-devel mailing list
> Ope...@li...
> https://lists.sourceforge.net/lists/listinfo/openvip-devel

-- 
PGP key: 0x465264C9, available from http://pgp.mit.edu/

[OpenVIP-devel] What is coretypes.h?

[Alexandre J. <web...@ma...>]

Hello, when trying to compile upf, I have the following problem:

Making all in include
Making all in upf
make: don't know how to make coretypes.h. Stop
*** Error code 1
Stop in /usr/sK0T/OpenVIP/upf/include.
*** Error code 1

The system is FreeBSD, the sources are from todays CVS.

Re: [OpenVIP-devel] [PATCH] upf issues

[Vaclav S. <vac...@ma...>]

Hi,

thanks! Everything applied now.

Vaclav

=2D-=20
PGP key: 0x465264C9, available from http://pgp.mit.edu/

Re: [OpenVIP-devel] [PATCH] openvip issues

[<ton...@po...>]

Hello,

> > I suspect there's a bug in
> > timeline->network conversion when
> > using the overlay tracks on timeline. Or
> > does the same thing
> > happen if you use only VA, VFx, VB and AA,
> > AFx, AB tracks?
> 
> It just uses V0+1 and A0+1. To be exact, I'm
> just trying to  
> concatenate two streams here. Mplayer is able

OK, but V0+1 and A0+1 ARE overlay tracks. If you only want to
concatenate two files, place them both on VA+AA, the first file
followed by the second file.

The other two tracks, VB+AB are employed only if you want to join
two files using a transition (which is then placed on VFx/AFx). 

The remaining tracks (V0,V1,... and A0,A1,...) are so-called
overlay tracks. They were intended to insert things such as logos
over another video. The overlay tracks don't work properly yet,
they are not documented and you should not use them now.

> BTW: There are quite a few reasons for me to
> use and to look into openvip.
> I'm a big fan of python programs with C[++]
> backends. My personal
> favorite is the sip/PyQt Combo. Epecially
> sip4 is very promising,
> being able to wrap C++ libs without
> generating any Python stub, 
> lazy instancing and the like. 
> 
> Anyway, I understand your reasoning using wx,
> I've been there before ;-)

Thanks for posting your opinion and hope you enjoy hacking
OpenVIP :-)

Antonin Slavik






-- 
Chces kilo? Tak pripoj kamose pres VOLNY.
Vice na http://studentpartner.volny.cz/

Re: [OpenVIP-devel] [PATCH] openvip issues

[Hans-Peter J. <hp...@ur...>]

On Thursday 19 February 2004 16:25, Anton=EDn Slav=EDk wrote:
>
> libfftw3.so is the right one. But as I already said, the FFTW API
> has changed and I haven't  checked yet whether OpenVIP works with
> FFTW 3.

Will check this then.=20

> > Jep. I got around the above step, using this
> > script:
> >
> > --8<--
> > #! /bin/sh
> >
> > cd /home/hp/CVS/openvip.hp/bin
> >
> > export OPENVIP_HOME=3D$(pwd)
> > export LD_LIBRARY_PATH=3D$(pwd)
> >
> > python openvip-gui.py
> > -->8--
>
> That's correct, I forgot to tell you about the LD_LIBRARY_PATH. I
> have already added a note to the README file.

Cool.

> > The problematic xml is attached.
>
> This file seems odd to me. I suppose it's been converted from the
> timeline format - could you send us the original timeline,
> please?

No problem. Attached.

> I suspect there's a bug in timeline->network conversion when
> using the overlay tracks on timeline. Or does the same thing
> happen if you use only VA, VFx, VB and AA, AFx, AB tracks?

It just uses V0+1 and A0+1. To be exact, I'm just trying to =20
concatenate two streams here. Mplayer is able to play them=20
correctly, but I failed to concat them with any other tools=20
I have. The best result produces avicat, but Mplayer still=20
chokes with=20
"Bad interleaved AVI file detected, switching to -ni mode"
and looses audio on that gap.

Here's avitype's output:
<init> : Avifile RELEASE-0.7.38-030924-01:12-gcc version 3.3.1 (SuSE Linux)
<reader> : checking: take01.avi
<AVI reader> : MainHeader: MicroSecPerFrame=3D33366 MaxBytesPerSec=3D0
 PaddingGranularity=3D0 Flags=3D[ HAS_INDEX ] TotalFrames=3D3603
 InitialFrames=3D22 Streams=3D2 SuggestedBufferSize=3D32768 WxH=3D320x240
 Scale=3D0 Rate=3D0 Start=3D0 Length=3D0
<AVI Reader> : WARNING: fccHandler differs from biCompression!
<AVI reader> : StreamHeader: Type=3Dvids Handler=3DDX50 Flags=3D[ ]
 InitialFrames=3D0 Scale=3D100 Rate=3D2997 Start=3D0 Length=3D3603
 SuggestedBufferSize=3D1036800 Quality=3D10000 SampleSize=3D0 Rect l,r,t,b=
=3D0,0,0,0
<AVI reader> : StreamHeader: Type=3Dauds Handler=3D0x0 Flags=3D[ ]
 InitialFrames=3D22 Scale=3D512 Rate=3D11155 Start=3D0 Length=3D2619
 SuggestedBufferSize=3D11264 Quality=3D10000 SampleSize=3D512 Rect l,r,t,b=
=3D0,0,0,0
<AVI reader> : Reading index from offset: 19358492
<AVI reader> : Stream 0 vids : DX50 (0x30355844) 3603 chunks (14.07KB)
<AVI reader> : Stream 1 auds : MS ADPCM (0x2) 125 chunks (0.98KB)
<StreamCache> : Creating cache for file descriptor: 5

BTW: There are quite a few reasons for me to use and to look into openvip.
I'm a big fan of python programs with C[++] backends. My personal
favorite is the sip/PyQt Combo. Epecially sip4 is very promising,
being able to wrap C++ libs without generating any Python stub,=20
lazy instancing and the like.=20

Anyway, I understand your reasoning using wx, I've been there before ;-)

Pete

Re: [OpenVIP-devel] [PATCH] openvip issues

[<ton...@po...>]

Hello,

> Well, yes there are also fftw3 libs available
> here:
> 
> libdfftw.so
> libdrfftw.so
> libsfftw.so
> libsrfftw.so
> libfftw3.so
> libfftw3f.so
> 
> but ruled those out. The question is: which
> off the rest is the
> right one?

libfftw3.so is the right one. But as I already said, the FFTW API
has changed and I haven't  checked yet whether OpenVIP works with
FFTW 3.

> Jep. I got around the above step, using this
> script:
> 
> --8<--
> #! /bin/sh
> 
> cd /home/hp/CVS/openvip.hp/bin
> 
> export OPENVIP_HOME=$(pwd)
> export LD_LIBRARY_PATH=$(pwd)
> 
> python openvip-gui.py
> -->8--

That's correct, I forgot to tell you about the LD_LIBRARY_PATH. I
have already added a note to the README file.

> Thanks for the pointer. It seem to
> MemDataSource::Read the data just 
> fine, but something gets wrong in
> parseConnection() stage.
> 
> The problematic xml is attached.

This file seems odd to me. I suppose it's been converted from the
timeline format - could you send us the original timeline,
please? 

I suspect there's a bug in timeline->network conversion when
using the overlay tracks on timeline. Or does the same thing
happen if you use only VA, VFx, VB and AA, AFx, AB tracks?

Antonin Slavik





-- 
Chces kilo? Tak pripoj kamose pres VOLNY.
Vice na http://studentpartner.volny.cz/

Re: [OpenVIP-devel] [PATCH] openvip issues

[Hans-Peter J. <hp...@ur...>]

Hi Antonin,

thanks for the response.

On Wednesday 18 February 2004 15:25, Anton=EDn Slav=EDk wrote:
> Hello Peter,
>
> thanks for reporting the issues.
>
> > Linking
> > errors due to fftw
> > tricked me a bit: found lib{d,dr,s,sr}fftw
> > libs here, but no libfftw.
>
> Do you have FFTW 3 installed? OpenVIP has been written to link

Well, yes there are also fftw3 libs available here:

libdfftw.so
libdrfftw.so
libsfftw.so
libsrfftw.so
libfftw3.so
libfftw3f.so

but ruled those out. The question is: which off the rest is the
right one?

> against FFTW 2.x. This can be achieved using -lfftw, while FFTW 3
> changed it to -lfftw3. I'll fix the Jamfile to link against FFTW
> 3, but I have to read FFTW docs first since it is not 100%
> backward compatible with FFTW 2.x. Anyway, all FFTW plugins are
> optional so you can safely leave them out.
>
> > I had to copy libopenvip_hl.so to
> > /usr/local/lib in order to get
> > openvip going.
>
> Did you set OPENVIP_HOME as described in the README file?

Jep. I got around the above step, using this script:

=2D-8<--
#! /bin/sh

cd /home/hp/CVS/openvip.hp/bin

export OPENVIP_HOME=3D$(pwd)
export LD_LIBRARY_PATH=3D$(pwd)

python openvip-gui.py
=2D->8--

I remember those problems with Phil Thompson's sip/PyQt stuff.

Phil used to fix them with ld's --soname option, but I don't have any=20
idea, how to tell bjam such.

> ldd libopenvip_hl.so=20
        libupf.so.0 =3D> /usr/local/lib/libupf.so.0 (0x4011e000)
        librt.so.1 =3D> /lib/librt.so.1 (0x40120000)
        libstdc++.so.5 =3D> /usr/lib/libstdc++.so.5 (0x40133000)
        libm.so.6 =3D> /lib/libm.so.6 (0x401f3000)
        libgcc_s.so.1 =3D> /lib/libgcc_s.so.1 (0x40216000)
        libpthread.so.0 =3D> /lib/libpthread.so.0 (0x4021e000)
        libc.so.6 =3D> /lib/libc.so.6 (0x40272000)
        libdl.so.2 =3D> /lib/libdl.so.2 (0x403a8000)
        /lib/ld-linux.so.2 =3D> /lib/ld-linux.so.2 (0x80000000)


> > openvip.Error: failed to load XML data:
> > <?xml version=3D"1.0" ?>
> > <!DOCTYPE network PUBLIC
> > "-//OPENVIP//DTD Network Format V1.0//EN"
> > [...]
>
> If you have OPENVIP_HOME set correctly, then I have no idea what
> might be wrong. The load_network_from_string Python function is
> implemented using the NetworkFormatLoader::LoadFromString C++
> function, which can be found in
> openvip/src/core/networkloader.cpp. The error means that this
> function returned false - could you find out why?

Thanks for the pointer. It seem to MemDataSource::Read the data just=20
fine, but something gets wrong in parseConnection() stage.

The problematic xml is attached.

Cheers,
Pete

Re: [OpenVIP-devel] [PATCH] openvip issues

[<ton...@po...>]

Hello Peter,

thanks for reporting the issues.

> Linking
> errors due to fftw 
> tricked me a bit: found lib{d,dr,s,sr}fftw
> libs here, but no libfftw.

Do you have FFTW 3 installed? OpenVIP has been written to link
against FFTW 2.x. This can be achieved using -lfftw, while FFTW 3
changed it to -lfftw3. I'll fix the Jamfile to link against FFTW
3, but I have to read FFTW docs first since it is not 100%
backward compatible with FFTW 2.x. Anyway, all FFTW plugins are
optional so you can safely leave them out.  
> I had to copy libopenvip_hl.so to
> /usr/local/lib in order to get 
> openvip going.

Did you set OPENVIP_HOME as described in the README file?
 
> Now, loading some avi's seems fine, I can
> save the timeline, but
> trying to render them, I get:
> 
> Traceback (most recent call last):
> File
> "/home/hp/CVS/openvip.hp/bin/MainFrame.py",
> line 510, in OnRender
> background=False)
> File "/home/hp/CVS/openvip.hp/bin/render.py",
> line 94, in renderFromString
> r.run()
> File "/home/hp/CVS/openvip.hp/bin/render.py",
> line 52, in run
> self.job.run(self)
> File "/home/hp/CVS/openvip.hp/bin/render.py",
> line 61, in run
> task =
> globals.core.load_network_from_string(self.net)
> File
> "/home/hp/CVS/openvip.hp/bin/openvip.py",
> line 157, in load_network_from_string
> raise Error("failed to load XML data:\n%s" %
> xmldata)
> openvip.Error: failed to load XML data:
> <?xml version="1.0" ?>
> <!DOCTYPE network PUBLIC
> "-//OPENVIP//DTD Network Format V1.0//EN"
> [...]

If you have OPENVIP_HOME set correctly, then I have no idea what
might be wrong. The load_network_from_string Python function is
implemented using the NetworkFormatLoader::LoadFromString C++
function, which can be found in
openvip/src/core/networkloader.cpp. The error means that this
function returned false - could you find out why?

Regards,

Antonin Slavik









-- 
Chces kilo? Tak pripoj kamose pres VOLNY.
Vice na http://studentpartner.volny.cz/

[OpenVIP-devel] [PATCH] openvip issues

[Hans-Peter J. <hp...@ur...>]

Hi Vaclav et al.,

as promised, the openvip issues collection follow up ;-)

Building bjam was issue free(tm), IIRC.

Building openvip broke in several ways, where most of them could
be fixed by resolving dependencies. Linking errors due to fftw 
tricked me a bit: found lib{d,dr,s,sr}fftw libs here, but no libfftw.

I fixed that with:

--- openvip/Jamfile     2003-06-02 18:51:05.000000000 +0200
+++ openvip.hp/Jamfile  2004-02-16 20:26:06.000000000 +0100
@@ -220,13 +220,13 @@
 dll filter_bandpass : <template>plugin
         src/plugins/fftw_plugins/bandpass.cpp
     :
-        <find-library>fftw
+        <find-library>sfftw
     ;
 
 dll filter_equalizer : <template>plugin
         src/plugins/fftw_plugins/equalizer.cpp
     :
-        <find-library>fftw
+        <find-library>sfftw
     ;
 
 dll in_wav : <template>plugin

but don't know, if that one is the right one. Any ideas?

bjam didn't figured out the PYTHON_INCLUDES dir, I needed to 
specify it directly:

--- openvip/Jam-config  2003-02-14 20:10:37.000000000 +0100
+++ openvip.hp/Jam-config       2004-02-16 23:12:00.000000000 +0100
@@ -7,9 +7,10 @@
 if $(UNIX)
 {
 # Location where your copy of UPF is installed:
-UPF_PATH ?= /usr ;
+UPF_PATH ?= /usr/local ;
 PYTHON_ROOT ?= /usr ;
-PYTHON_VERSION ?= 2.2 ;
+PYTHON_INCLUDES ?= /usr/include/python ;
+PYTHON_VERSION ?= 2.3 ;
 }
 
 if $(NT)

After spitting lots of simple to scary warnings, build was successful.

I had to copy libopenvip_hl.so to /usr/local/lib in order to get 
openvip going.

Now, loading some avi's seems fine, I can save the timeline, but
trying to render them, I get:

Traceback (most recent call last):
  File "/home/hp/CVS/openvip.hp/bin/MainFrame.py", line 510, in OnRender
    background=False)
  File "/home/hp/CVS/openvip.hp/bin/render.py", line 94, in renderFromString
    r.run()
  File "/home/hp/CVS/openvip.hp/bin/render.py", line 52, in run
    self.job.run(self)
  File "/home/hp/CVS/openvip.hp/bin/render.py", line 61, in run
    task = globals.core.load_network_from_string(self.net)
  File "/home/hp/CVS/openvip.hp/bin/openvip.py", line 157, in load_network_from_string
    raise Error("failed to load XML data:\n%s" % xmldata)
openvip.Error: failed to load XML data:
<?xml version="1.0" ?>
<!DOCTYPE network PUBLIC
         "-//OPENVIP//DTD Network Format V1.0//EN"
[...]

I'm busted!

Any ideas on this one?

Cheers,
Pete

[OpenVIP-devel] [PATCH] upf issues

[Hans-Peter J. <hp...@ur...>]

Hi Vaclav & friends,

I just came across your project, and decided to give it a try.
Since the distributed packages didn't build for me (on SuSE 9.0),
I co'ed CVS.

Building upf needed this patch here:

--- upf/autogen.sh      2002-11-25 01:21:52.000000000 +0100
+++ upf.hp/autogen.sh   2004-02-16 18:56:25.000000000 +0100
@@ -1,13 +1,13 @@
 #!/bin/sh
 
 echo aclocal
-aclocal-1.6 -I admin
+aclocal -I admin
 echo autoconf
 autoconf
 echo autoheader
 autoheader
 echo automake
-automake-1.6 -a -c --foreign
+automake -a -c --foreign
 
 
 #libtoolize -c -f

and this file seems to be missing in CVS:

--- /dev/null   2003-09-23 19:59:22.000000000 +0200
+++ upf.hp/include/Makefile.am  2004-02-16 19:17:46.000000000 +0100
@@ -0,0 +1,2 @@
+
+SUBDIRS = upf


Thereafter build & install runs fine.

Lot's of Python 2.3 warnings

upf/tools/upfparser/typeinf.py:2{6789}: FutureWarning: int('0...', 0):
 sign will change in Python 2.4

could possibly be suppressed with:

--- upf/tools/upfparser/typeinf.py~     2003-02-04 00:22:48.000000000 +0100
+++ upf/tools/upfparser/typeinf.py      2004-02-16 22:06:05.000000000 +0100
@@ -23,10 +23,10 @@
 def encIID(IID, version):
     major=int(version.split('.')[0])
     minor=int(version.split('.')[1])
-    iid1=int("0x" + IID[0:8], 0)
-    iid2=int("0x" + IID[8:16], 0)
-    iid3=int("0x" + IID[16:24], 0)
-    iid4=int("0x" + IID[24:32], 0)
+    iid1=int("0x" + IID[0:8], 16)
+    iid2=int("0x" + IID[8:16], 16)
+    iid3=int("0x" + IID[16:24], 16)
+    iid4=int("0x" + IID[24:32], 16)
     return struct.pack('!LLLLHH', iid1, iid2, iid3, iid4, major, minor)
 
 

Since this is enough for a single mail, me thinks, I will follow up
with a openvip related one ;-)

Cheers,
Pete

Re: [OpenVIP-devel] upf ./Makefile.am

[Joel B. <jb...@la...>]

On Wed, 2004-02-04 at 15:54, Vaclav Slavik wrote:

> Joel Brauer wrote:
> > I just downloaded the cvs of upf and tried to run ./autogen.sh
> > which ended in
> > "Makefile.am:2: directory should not contain `/'"
> 
> Not really. It ended in writing out this message *and* generating 
> working Makefile.in. Or did it not? I was able to compile to the 
> library using it...


Originally, I thought this was the case as well, but when I tried to run
configure it died with:

    configure: creating ./config.status
    config.status: creating Makefile
    config.status: creating tools/Makefile
    config.status: creating tools/omniidl/Makefile
    config.status: creating include/Makefile
    config.status: error: cannot find input file: include/Makefile.in

Maybe it's unrelated, but I guess I assumed it had something to do with
the earlier error.

> 
> > Is there any reason it has to be in the tools dir?
> 
> Because it's a tool?
> 

Yeah.  That wasn't what I was asking.  I was asking if you minded if it
got moved up a level in the directory structure?

> Anyway, if you can fix it (preferably without moving lots of files 
> around), feel free to post a patch.
> 
> VS


Thanks,

-- 
Joel Brauer
La Sierra University
Programmer/Analyst
909-785-2308

-- this email is certified virus free! How? Because it didn't
-- come from any Micro$oft based platform or product.

PGP Key: 0x211149BA, available from http://pgp.mit.edu/

Re: [OpenVIP-devel] upf ./Makefile.am

[Vaclav S. <vac...@ma...>]

Joel Brauer wrote:
> I just downloaded the cvs of upf and tried to run ./autogen.sh
> which ended in
> "Makefile.am:2: directory should not contain `/'"

Not really. It ended in writing out this message *and* generating=20
working Makefile.in. Or did it not? I was able to compile to the=20
library using it...

> Is there any reason it has to be in the tools dir?

Because it's a tool?

Anyway, if you can fix it (preferably without moving lots of files=20
around), feel free to post a patch.

VS

=2D-=20
PGP key: 0x465264C9, available from http://pgp.mit.edu/

[OpenVIP-devel] upf ./Makefile.am

[Joel B. <jb...@la...>]

I just downloaded the cvs of upf and tried to run ./autogen.sh
which ended in
"Makefile.am:2: directory should not contain `/'"

Makefile.am line 2 contains

SUBDIRS = tools/omniidl include src tools tests examples

Apparently, the problem is that you can only recurse subdirs directly
with automake.  I don't know why.  Go figure.  But, The Automake docs
say you can't do it.  So, either omniidl needs to be moved out to a
second level dir, or possibly I might be able to write into the
Makefiles a conditional subdir that would cause the omniidl to be built
first and then not the next time around.  

The more I think about it, this seems like too much work for such a
simple problem.

Is there any reason it has to be in the tools dir?

I know that it was my comments previously that caused this change, so I
don't mean to be ungrateful.  I just want to find out what is/is not
acceptable regarding directory tree structure.

Thanks, 
-- 
Joel Brauer
La Sierra University
Programmer/Analyst
909-785-2308

-- this email is certified virus free! How? Because it didn't
-- come from any Micro$oft based platform or product.

PGP Key: 0x211149BA, available from http://pgp.mit.edu/

Re: [OpenVIP-devel] patch for upf/tools to compile under gcc3.3

[Vaclav S. <vac...@ma...>]

Vaclav Slavik wrote:
> > I also found that I had to compile things in a specific order.
>
> *Why* did you have to do it? You are certainly not supposed to.

I was able to reproduce it myself, so it is now fixed in CVS.

Thanks,
vaclav

=2D-=20
PGP key: 0x465264C9, available from http://pgp.mit.edu/

Re: [OpenVIP-devel] patch for upf/tools to compile under gcc3.3

[Joel B. <jb...@la...>]

Hey,

On Wed, 2004-01-28 at 15:25, Vaclav Slavik wrote:


> Joel Brauer wrote:
> > Here is the patch of the changes that I had to make in order to 
> > compile this under gcc 3.3.2.
> 
> Sorry, but the patch is clearly wrong -- it removes #ifdefed code and 
> so it is, after the patch is applied, *guaranteed* to fail to compile 
> on at least some platforms. The real fix would be to figure out why 
> is the code not used (i.e. what's wrong with the #defines and/or 
> autoconf tests results) and fix that.



Like I said I'm not too familiar with autoconf, so you're probably
right. I ought to track down where autoconf is going wrong, or possibly
add a #ifdefed to check for gcc >3.3.  Although I'm not sure how to do
that, so I'll have to look into it.

> > I also found that I had to compile things in a specific order.
> 
> *Why* did you have to do it? You are certainly not supposed to.
> 

Well If I didn't compile it this way.  It would die in mid compile.  It
would try to compile things that needed the idl generated .h files
before they had been generated.  And then try to compile parts that
required the upf library be already compiled in the src/ dir, but that
had not happened yet.  Have you compiled it on linux?  I don't see why
it would do things in the wrong order on my machine and not on others.

> > submitted a patch for that because I am not quite familiar with the
> > whole Autoconf Automake process.  But once I figure it out I'll
> > submit one.  Or if anyone knows how to get that in there
> > automagically then have at it :).  Essentially before I added it,
> > my Makefile was missing -L/usr/lib/python/config which caused the
> > compiler to say that -lpython2.2 was invalid.
> 
> You can simply set PYTHON_LIB to empty in the makefile, it's not 
> needed on ELF-based Unices.
> 

Are you saying that the -lpython2.2 is not needed? or that just the
-L/usr/lib/python/config is not needed?  The first I haven't tried, but
I know the second to be false.


> VS

-- 
Joel Brauer
La Sierra University
Programmer/Analyst
909-785-2308

-- this email is certified virus free! How? Because it didn't
-- come from any Micro$oft based platform or product.

PGP Key: 0x211149BA, available from http://pgp.mit.edu/

Re: [OpenVIP-devel] patch for upf/tools to compile under gcc3.3

[Vaclav S. <vac...@ma...>]

Joel Brauer wrote:
> =A0 =A0 One other thing. =A0The reason the patch is needed for gcc >3.3,
> is because the use of varargs.h is depricated in favor of stdarg.h.

Yep, I've seen that. The reason it didn't compile was that=20
configure.in was missing AC_FUNC_VPRINTF check and so the code that=20
used <stdarg.h> was never used.

VS

=2D-=20
PGP key: 0x465264C9, available from http://pgp.mit.edu/

Re: [OpenVIP-devel] patch for upf/tools to compile under gcc3.3

[Joel B. <jb...@la...>]

Oh,

    One other thing.  The reason the patch is needed for gcc >3.3, is
because the use of varargs.h is depricated in favor of stdarg.h.  I
probably incorrectly assumed that stdarg.h should also work with lower
versions of gcc.  Hence my removal of code refering to varargs.h 


-- 
Joel Brauer
La Sierra University
Programmer/Analyst
909-785-2308

-- this email is certified virus free! How? Because it didn't
-- come from any Micro$oft based platform or product.

PGP Key: 0x211149BA, available from http://pgp.mit.edu/

Re: [OpenVIP-devel] patch for upf/tools to compile under gcc3.3

[Vaclav S. <vac...@ma...>]

Hi,

Joel Brauer wrote:
> Here is the patch of the changes that I had to make in order to=20
> compile this under gcc 3.3.2.

Sorry, but the patch is clearly wrong -- it removes #ifdefed code and=20
so it is, after the patch is applied, *guaranteed* to fail to compile=20
on at least some platforms. The real fix would be to figure out why=20
is the code not used (i.e. what's wrong with the #defines and/or=20
autoconf tests results) and fix that.

> I also found that I had to compile things in a specific order.

*Why* did you have to do it? You are certainly not supposed to.

> submitted a patch for that because I am not quite familiar with the
> whole Autoconf Automake process.  But once I figure it out I'll
> submit one.  Or if anyone knows how to get that in there
> automagically then have at it :).  Essentially before I added it,
> my Makefile was missing -L/usr/lib/python/config which caused the
> compiler to say that -lpython2.2 was invalid.

You can simply set PYTHON_LIB to empty in the makefile, it's not=20
needed on ELF-based Unices.

VS

=2D-=20
PGP key: 0x465264C9, available from http://pgp.mit.edu/

[OpenVIP-devel] patch for upf/tools to compile under gcc3.3

[Joel B. <jb...@la...>]

Here is the patch of the changes that I had to make in order to compile
this under gcc 3.3.2.

I also found that I had to compile things in a specific order.  

./configure --prefix=/usr
cd tools/omniidl/
make
cd ../../include
make
cd ../src
make
cd ..
make
make install

I also had to put the output of python-config into src/python/Makefile
in order for it to compile.  I haven't submitted a patch for that
because I am not quite familiar with the whole Autoconf Automake
process.  But once I figure it out I'll submit one.  Or if anyone knows
how to get that in there automagically then have at it :).  Essentially
before I added it, my Makefile was missing -L/usr/lib/python/config
which caused the compiler to say that -lpython2.2 was invalid.  Hope
this helps.
-- 
Joel Brauer
La Sierra University
Programmer/Analyst
909-785-2308

-- this email is certified virus free! How? Because it didn't
-- come from any Micro$oft based platform or product.

PGP Key: 0x211149BA, available from http://pgp.mit.edu/

[OpenVIP-devel] novy ffmpeg, anyone?

[Vaclav S. <vac...@ma...>]

Cau,

zda se ze ve verzi ffmpeg-0.4.8 zapracovali na podpore win32:
  http://ffmpeg.sourceforge.net/ffmpeg-doc.html#SEC24
Tak kdyby to nekdo chtel vyzkouset a treba nahradit ty nase stare DLL=20
novymi... ;-)

Vasek

=2D-=20
PGP key: http://pgp.mit.edu:11371/pks/lookup?op=3Dget&search=3D0x465264C9

[OpenVIP-devel] nove stranky

[Vaclav S. <vac...@ma...>]

Cau,

Tonda udelal nove stranky, urcene uz pro verejnost, zalozil jsem i=20
mailing list openvip-users, tak kdo ma odvahu, tak se prihlaste ;-)=20
Prave vyplnuju zaznam do katalogu Freshmeatu...

Vasek

=2D-=20
PGP key: http://pgp.mit.edu:11371/pks/lookup?op=3Dget&search=3D0x465264C9

[OpenVIP-devel] Re: SW projekt OpenVIP - poznamky

[Tomas H. <Hol...@ks...>]

> Dobry den,
>
> zde je par postrehu a zkusenosti, ktere jsme nasbirali behem
> programovani OpenVIPu:

   Dekuji, i za vsechny, komu Vase zkusenosti pomohou.

> Dekujeme vam za vstricny pristup pred obhajobou i v jejim
> prubehu.

   ?
   Neni zac.

   Tomas Holan

[OpenVIP-devel] SW projekt OpenVIP - poznamky

[<ton...@po...>]

Dobry den,

zde je par postrehu a zkusenosti, ktere jsme nasbirali behem
programovani OpenVIPu:

---

1) Dobre zkusenosti se sdilenim zdrojovych kodu pomoci CVS,
repozitar na serveru sourceforge.net (navic prostor pro webove
stranky a dalsi soubory, mailing listy).

2) Boost Build - build system vhodny pro multiplatformni
aplikace, OpenVIP jsme vyvijeli soucasne v gcc, MSVC, mingw,
Borland C++. Problemy s rozchozenim bjamu pod Windows NT.

3) Python - interpretovany, proceduralni a objektove orientovany
jazyk, rychle se v nem programuje, k dispozici kvalitni tutorial,
nastroje pro generovani referencni dokumentace.

4) wxPython - Python verze multiplatformni knihovny wxWindows pro
vyvoj GUI, nativni vzhled aplikaci pod Windows i Unixem, kvalitni
dokumentace.
Drobne problemy s ruznym chovanim na ruznych platformach, ale je
to  primerena cena za to, ze aplikace je nativni.

5) wxGlade - vizualni editor GUI, generuje kod pro
wxWindows/wxPython, snadno ovladatelny, mensi problemy se
stabilitou.

6) Dokumentace v XML (DocBook), napsana v editoru XMLmind
(Windows i Unix, ve Windows mirne nestabilni). Z XML generujeme
skriptem na sourceforge.net online HTML dokumentaci - bez
problemu, dobre vypada. PDF dokumentaci jsme vyrobili pomoci
programu XMLmind FO Converter - problemy s velikosti obrazku a
tabulek.

7) Knihovna ffmpeg pro nacitani a ukladani videa. Velky vyber
kodeku a formatu, vysoka rychlost. Temer zadna dokumentace, je
treba studovat zdrojove kody. Nema podporu pro seekovani v
souborech (napr. na N-ty snimek), museli jsme naprogramovat sami.
Knihovna neni zcela stabilni (Windows?), stale se vyviji.

8) FFTW - velmi rychla multiplatformni knihovna pro vypocet
Fourierovy transformace, kvalitni dokumentace.

9) ImageMagick - multiplatformni knihovna pro nacitani, ukladani
a zpracovani statickych obrazku. Mnoho formatu, neprilis podrobna
dokumentace.

10) V pythonu jsme pouzili XML parser, ktery je soucasti Pythonu
2.2, ale nema  validaci proti DTD, takze nakonec pouzivame i
libxml2-python. libxml2  umi uplne vsechno, hlasi dobre chyby,
validuje, ma SAX i DOM 
interface, jenom je trosku velka.

11) UML diagramy v dokumentaci vygenerovane z existujicich IDL
souboru aplikaci Together. Diagramy dobre vypadaji, jen nekolik
rucnich uprav. Together je bohuzel komercni a neskutecne pomaly.

---

Pokud by vas zajimalo jeste neco dalsiho, dejte nam vedet. 

Dekujeme vam za vstricny pristup pred obhajobou i v jejim
prubehu.

Za vyvojare OpenVIPu

Antonin Slavik





-- 
Ziskejte kvalitu, kterou si zasluhujete. Za minimalni mesicni
poplatek vam nabizime Antivir, Antispam nebo dalsi kapacitu pro
vas Mailbox. Vice na: http://sluzby.volny.cz/product/postpaid/

Re: [OpenVIP-devel] Poznamky k openvipu - navrh

[Vaclav S. <vac...@ma...>]

Cau,

Anton=EDn Slav=EDk wrote:
> 2) Boost Build - build system vhodny pro multiplatformni
> aplikace, OpenVIP jsme vyvijeli soucasne v gcc, MSVC, mingw,
> Borland C++.

Drobne mouchy, ktere me iritovaly:
=2D problemy s rozchozenim bjamu na NT
=2D chyba behem kompilace ji neukonci, takze se da prehlednout
=2D na windows Ctrl-C neprerusi build, Ctrl-C-Ctrl-C....-Ctrl-C ano

> 4) wxPython - Python verze multiplatformni knihovny wxWindows pro
> vyvoj GUI, nativni vzhled aplikaci pod Windows i Unixem, kvalitni
> dokumentace.

=2E..drobne problemy s ruznym chovanim na ruznych platformach, ale je to=20
primerena cena za to, ze aplikace je nativni.

> 6) Dokumentace v XML (DocBook), napsana v editoru XMLmind
> (Windows i Unix, ve Windows mirne nestabilni). Z XML generujeme
> skriptem na sourceforge.net online HTML dokumentaci - bez
> problemu, dobre vypada. PDF dokumentaci jsme vyrobili pomoci
> programu XMLmind FO Converter - problemy s velikosti obrazku a
> tabulek.

Necekane problemy s tim vubec rozhodit konverzi do PDF.

> Mate treba nejake komentare k tem XML parserum, ktere jste
> pouzili?

V pythonu jsme pouzili parser, ktery je soucasti Pythonu 2.2, ale nema=20
validaci proti DTD, takze nakonec pouzivame i libxml2-python. libxml2=20
umi uplne vsechno, hlasi dobre chyby, validuje, ma SAX i DOM=20
interface, jenom je trosku velka.

HTH,
Vasek

=2D-=20
PGP key: http://pgp.mit.edu:11371/pks/lookup?op=3Dget&search=3D0x465264C9

[OpenVIP-devel] Poznamky k openvipu - navrh

[<ton...@po...>]

Hi,

zkusil jsem sepsat par poznamek k projektu, o ktere nas prosil
Holan. Muzete k tomu doplnit sve vlastni postrehy a pripadne
opravit veci, s kterymi nesouhlasite.

---

1) Dobre zkusenosti se sdilenim zdrojovych kodu pomoci CVS,
repozitar na serveru sourceforge.net (navic prostor pro webove
stranky a dalsi soubory, mailing listy).

2) Boost Build - build system vhodny pro multiplatformni
aplikace, OpenVIP jsme vyvijeli soucasne v gcc, MSVC, mingw,
Borland C++.

3) Python - interpretovany, proceduralni a objektove orientovany
jazyk, rychle se v nem programuje, k dispozici kvalitni tutorial,
nastroje pro generovani referencni dokumentace.

4) wxPython - Python verze multiplatformni knihovny wxWindows pro
vyvoj GUI, nativni vzhled aplikaci pod Windows i Unixem, kvalitni
dokumentace.

5) wxGlade - vizualni editor GUI, generuje kod pro
wxWindows/wxPython, snadno ovladatelny, mensi problemy se
stabilitou.

6) Dokumentace v XML (DocBook), napsana v editoru XMLmind
(Windows i Unix, ve Windows mirne nestabilni). Z XML generujeme
skriptem na sourceforge.net online HTML dokumentaci - bez
problemu, dobre vypada. PDF dokumentaci jsme vyrobili pomoci
programu XMLmind FO Converter - problemy s velikosti obrazku a
tabulek.

7) Knihovna ffmpeg pro nacitani a ukladani videa. Velky vyber
kodeku a formatu, vysoka rychlost. Temer zadna dokumentace, je
treba studovat zdrojove kody. Nema podporu pro seekovani v
souborech (napr. na N-ty snimek), museli jsme naprogramovat sami.
Knihovna neni zcela stabilni (Windows?), stale se vyviji.

8) FFTW - velmi rychla multiplatformni knihovna pro vypocet
Fourierovy transformace, kvalitni dokumentace.

9) ImageMagick - multiplatformni knihovna pro nacitani, ukladani
a zpracovani statickych obrazku. Mnoho formatu, neprilis podrobna
dokumentace.

---

Mate treba nejake komentare k tem XML parserum, ktere jste
pouzili?

Tonda




-- 
Ziskejte kvalitu, kterou si zasluhujete. Za minimalni mesicni
poplatek vam nabizime Antivir, Antispam nebo dalsi kapacitu pro
vas Mailbox. Vice na: http://sluzby.volny.cz/product/postpaid/