[PATCH] vorbis.h -> vorbis_data.c
It would be silly for me to send this as a patch, because it's an almost completely unreadable ~260kb patch. These are the changes to be made: 1. the attached patch, for Makefile and vorbis.c 2. vorbis.h would be changed to what is attached below 3. vorbis_data.c would be svn-cp'ed from vorbis.h, having the struct/typedefs removed, and a single line added: const float * ff_vorbis_vwin[8] = { vwin64, vwin128, vwin256, vwin512, vwin1024, vwin2048, vwin4096, vwin8192 }; I'll commit tommorrow, this is first step for me making vorbis_enc clean enough to commit to lavc... - ods15
Hi On Sat, Sep 30, 2006 at 06:09:02PM +0300, Oded Shimon wrote:
It would be silly for me to send this as a patch, because it's an almost completely unreadable ~260kb patch. These are the changes to be made:
1. the attached patch, for Makefile and vorbis.c
nothing is attached ...
2. vorbis.h would be changed to what is attached below 3. vorbis_data.c would be svn-cp'ed from vorbis.h, having the struct/typedefs removed, and a single line added:
const float * ff_vorbis_vwin[8] = { vwin64, vwin128, vwin256, vwin512, vwin1024, vwin2048, vwin4096, vwin8192 };
I'll commit tommorrow, this is first step for me making vorbis_enc clean enough to commit to lavc...
iam against it, moving stuff from 1 file to another leads to a 0k patch not a 260k one, diff the files relative to their correctly matching counterparts, that could also be an empty file if not a single line is matching [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB In the past you could go to a library and read, borrow or copy any book Today you'd get arrested for mere telling someone where the library is
On Sat, Sep 30, 2006 at 05:15:51PM +0200, Michael Niedermayer wrote:
Hi
On Sat, Sep 30, 2006 at 06:09:02PM +0300, Oded Shimon wrote:
It would be silly for me to send this as a patch, because it's an almost completely unreadable ~260kb patch. These are the changes to be made:
1. the attached patch, for Makefile and vorbis.c
nothing is attached ...
sorry :)
2. vorbis.h would be changed to what is attached below 3. vorbis_data.c would be svn-cp'ed from vorbis.h, having the struct/typedefs removed, and a single line added:
const float * ff_vorbis_vwin[8] = { vwin64, vwin128, vwin256, vwin512, vwin1024, vwin2048, vwin4096, vwin8192 };
I'll commit tommorrow, this is first step for me making vorbis_enc clean enough to commit to lavc...
iam against it, moving stuff from 1 file to another leads to a 0k patch not a 260k one, diff the files relative to their correctly matching counterparts, that could also be an empty file if not a single line is matching
ok, Here's the patch between (the current) vorbis.h, and vorbis_data.c , and yes it's a small patch. however, the patch between the current vorbis.h and the one I'm attaching here again, is an uninteresting 130kb one of all the tables being removed. So, again: 1. the attacked patch for makefile and vorbis.c 2. vorbis.h to what is attached below 3. vorbis_data.c from vorbis.h and the attached patch for it made 2 changes as Mans and Denes requested, the structs stay in vorbis.h - ods15
Hi On Sat, Sep 30, 2006 at 06:27:16PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 05:15:51PM +0200, Michael Niedermayer wrote:
Hi
On Sat, Sep 30, 2006 at 06:09:02PM +0300, Oded Shimon wrote:
It would be silly for me to send this as a patch, because it's an almost completely unreadable ~260kb patch. These are the changes to be made:
1. the attached patch, for Makefile and vorbis.c
nothing is attached ...
sorry :)
2. vorbis.h would be changed to what is attached below 3. vorbis_data.c would be svn-cp'ed from vorbis.h, having the struct/typedefs removed, and a single line added:
const float * ff_vorbis_vwin[8] = { vwin64, vwin128, vwin256, vwin512, vwin1024, vwin2048, vwin4096, vwin8192 };
I'll commit tommorrow, this is first step for me making vorbis_enc clean enough to commit to lavc...
iam against it, moving stuff from 1 file to another leads to a 0k patch not a 260k one, diff the files relative to their correctly matching counterparts, that could also be an empty file if not a single line is matching
ok, Here's the patch between (the current) vorbis.h, and vorbis_data.c , and yes it's a small patch.
however, the patch between the current vorbis.h and the one I'm attaching here again, is an uninteresting 130kb one of all the tables being removed.
So, again: 1. the attacked patch for makefile and vorbis.c 2. vorbis.h to what is attached below 3. vorbis_data.c from vorbis.h and the attached patch for it
made 2 changes as Mans and Denes requested, the structs stay in vorbis.h
looks ok [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB In the past you could go to a library and read, borrow or copy any book Today you'd get arrested for mere telling someone where the library is
On Sat, Sep 30, 2006 at 06:25:24PM +0200, Michael Niedermayer wrote:
On Sat, Sep 30, 2006 at 06:27:16PM +0300, Oded Shimon wrote:
made 2 changes as Mans and Denes requested, the structs stay in vorbis.h
looks ok
This is not the up-to-date patch, <20060930155533.GQ6927 at crate15> is. The structs are moved after all to vorbis.c . - ods15
Hi On Sat, Sep 30, 2006 at 07:28:30PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 06:25:24PM +0200, Michael Niedermayer wrote:
On Sat, Sep 30, 2006 at 06:27:16PM +0300, Oded Shimon wrote:
made 2 changes as Mans and Denes requested, the structs stay in vorbis.h
looks ok
This is not the up-to-date patch, <20060930155533.GQ6927 at crate15> is. The structs are moved after all to vorbis.c .
iam fine with that too but then vorbis.h has nothing in common anymore with the old vorbis.h (except the license header but that doesnt count) or is there something else? if not then the new vorbis.h should be added with svn add not as a difference to the old vorbis.h one way to actually do this (svn is a pile of shit ...) should be something like: svn mv vorbis.h vorbis_data.c change all vorbis.h -> vorbis_data.c references svn ci apply patch reverse the #include "vorbis_data.c" mess svn add vorbis.h svn ci [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB In the past you could go to a library and read, borrow or copy any book Today you'd get arrested for mere telling someone where the library is
On Sat, Sep 30, 2006 at 07:51:14PM +0200, Michael Niedermayer wrote:
Hi
On Sat, Sep 30, 2006 at 07:28:30PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 06:25:24PM +0200, Michael Niedermayer wrote:
On Sat, Sep 30, 2006 at 06:27:16PM +0300, Oded Shimon wrote:
made 2 changes as Mans and Denes requested, the structs stay in vorbis.h
looks ok
This is not the up-to-date patch, <20060930155533.GQ6927 at crate15> is. The structs are moved after all to vorbis.c .
iam fine with that too but then vorbis.h has nothing in common anymore with the old vorbis.h (except the license header but that doesnt count) or is there something else?
nope, nothing... I was reluctant about changing the license header :)
if not then the new vorbis.h should be added with svn add not as a difference to the old vorbis.h
one way to actually do this (svn is a pile of shit ...) should be something like:
svn mv vorbis.h vorbis_data.c change all vorbis.h -> vorbis_data.c references svn ci
apply patch reverse the #include "vorbis_data.c" mess svn add vorbis.h svn ci
is it possible to do 'svn mv' and 'svn add' in one go? svn mv vorbis.h vorbis_data.c svn add vorbis.h I'll check on a test repo... - ods15
On Sat, Sep 30, 2006 at 09:27:50PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 07:51:14PM +0200, Michael Niedermayer wrote:
Hi
On Sat, Sep 30, 2006 at 07:28:30PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 06:25:24PM +0200, Michael Niedermayer wrote:
On Sat, Sep 30, 2006 at 06:27:16PM +0300, Oded Shimon wrote:
made 2 changes as Mans and Denes requested, the structs stay in vorbis.h
looks ok
This is not the up-to-date patch, <20060930155533.GQ6927 at crate15> is. The structs are moved after all to vorbis.c .
iam fine with that too but then vorbis.h has nothing in common anymore with the old vorbis.h (except the license header but that doesnt count) or is there something else?
nope, nothing... I was reluctant about changing the license header :)
if not then the new vorbis.h should be added with svn add not as a difference to the old vorbis.h
one way to actually do this (svn is a pile of shit ...) should be something like:
svn mv vorbis.h vorbis_data.c change all vorbis.h -> vorbis_data.c references svn ci
apply patch reverse the #include "vorbis_data.c" mess svn add vorbis.h svn ci
is it possible to do 'svn mv' and 'svn add' in one go?
svn mv vorbis.h vorbis_data.c svn add vorbis.h
I'll check on a test repo...
Works fine, I'll use this. BTW, should there evn be a copyright header in vorbis_data.c ?.. - ods15
On Sat, Sep 30, 2006 at 09:30:41PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 09:27:50PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 07:51:14PM +0200, Michael Niedermayer wrote:
Hi
On Sat, Sep 30, 2006 at 07:28:30PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 06:25:24PM +0200, Michael Niedermayer wrote:
On Sat, Sep 30, 2006 at 06:27:16PM +0300, Oded Shimon wrote:
made 2 changes as Mans and Denes requested, the structs stay in vorbis.h
looks ok
This is not the up-to-date patch, <20060930155533.GQ6927 at crate15> is. The structs are moved after all to vorbis.c .
iam fine with that too but then vorbis.h has nothing in common anymore with the old vorbis.h (except the license header but that doesnt count) or is there something else?
nope, nothing... I was reluctant about changing the license header :)
if not then the new vorbis.h should be added with svn add not as a difference to the old vorbis.h
one way to actually do this (svn is a pile of shit ...) should be something like:
svn mv vorbis.h vorbis_data.c change all vorbis.h -> vorbis_data.c references svn ci
apply patch reverse the #include "vorbis_data.c" mess svn add vorbis.h svn ci
is it possible to do 'svn mv' and 'svn add' in one go?
svn mv vorbis.h vorbis_data.c svn add vorbis.h
I'll check on a test repo...
Works fine, I'll use this.
BTW, should there evn be a copyright header in vorbis_data.c ?..
Committed. I put my own name in vorbis.h .. I hope noone minds.. - ods15
Hi On Sat, Sep 30, 2006 at 09:30:41PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 09:27:50PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 07:51:14PM +0200, Michael Niedermayer wrote:
Hi
On Sat, Sep 30, 2006 at 07:28:30PM +0300, Oded Shimon wrote:
On Sat, Sep 30, 2006 at 06:25:24PM +0200, Michael Niedermayer wrote:
On Sat, Sep 30, 2006 at 06:27:16PM +0300, Oded Shimon wrote:
made 2 changes as Mans and Denes requested, the structs stay in vorbis.h
looks ok
This is not the up-to-date patch, <20060930155533.GQ6927 at crate15> is. The structs are moved after all to vorbis.c .
iam fine with that too but then vorbis.h has nothing in common anymore with the old vorbis.h (except the license header but that doesnt count) or is there something else?
nope, nothing... I was reluctant about changing the license header :)
if not then the new vorbis.h should be added with svn add not as a difference to the old vorbis.h
one way to actually do this (svn is a pile of shit ...) should be something like:
svn mv vorbis.h vorbis_data.c change all vorbis.h -> vorbis_data.c references svn ci
apply patch reverse the #include "vorbis_data.c" mess svn add vorbis.h svn ci
is it possible to do 'svn mv' and 'svn add' in one go?
svn mv vorbis.h vorbis_data.c svn add vorbis.h
I'll check on a test repo...
Works fine, I'll use this.
iam surprissed, last time i tried a few "intersecting" commands svn vommited at me (that was svn rm X + svn cp -r1234 svn://.../X X) [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB In the past you could go to a library and read, borrow or copy any book Today you'd get arrested for mere telling someone where the library is
On Sat, Sep 30, 2006 at 09:30:41PM +0300, Oded Shimon wrote:
BTW, should there evn be a copyright header in vorbis_data.c ?..
Yes, see the recent licensing thread. Diego
On Sat, Sep 30, 2006 at 06:09:02PM +0300, Oded Shimon wrote:
It would be silly for me to send this as a patch, because it's an almost completely unreadable ~260kb patch. These are the changes to be made:
1. the attached patch, for Makefile and vorbis.c 2. vorbis.h would be changed to what is attached below 3. vorbis_data.c would be svn-cp'ed from vorbis.h, having the struct/typedefs removed, and a single line added:
const float * ff_vorbis_vwin[8] = { vwin64, vwin128, vwin256, vwin512, vwin1024, vwin2048, vwin4096, vwin8192 };
I'll commit tommorrow, this is first step for me making vorbis_enc clean enough to commit to lavc...
- ods15
Oded Shimon <ods15 at ods15.dyndns.org> writes:
On Sat, Sep 30, 2006 at 06:09:02PM +0300, Oded Shimon wrote:
It would be silly for me to send this as a patch, because it's an almost completely unreadable ~260kb patch. These are the changes to be made:
1. the attached patch, for Makefile and vorbis.c 2. vorbis.h would be changed to what is attached below 3. vorbis_data.c would be svn-cp'ed from vorbis.h, having the struct/typedefs removed, and a single line added:
const float * ff_vorbis_vwin[8] = { vwin64, vwin128, vwin256, vwin512, vwin1024, vwin2048, vwin4096, vwin8192 };
I'll commit tommorrow, this is first step for me making vorbis_enc clean enough to commit to lavc...
- ods15
#ifndef _VORBIS_H #define _VORBIS_H
Symbols starting with _ are reserved. Call it something else, e.g. VORBIS_H. -- M?ns Rullg?rd mru at inprovide.com
Hi! szombat 30 szeptember 2006 17.16-kor Oded Shimon ezeket a bolcs gondolatokat fogalmazta meg:
On Sat, Sep 30, 2006 at 06:09:02PM +0300, Oded Shimon wrote:
It would be silly for me to send this as a patch, because it's an almost completely unreadable ~260kb patch. These are the changes to be made:
1. the attached patch, for Makefile and vorbis.c
I don't find moving all the struct definitions from vorbis.h to vorbis.c a particularly good idea, actually. Why should they not be in a header file, as it's normaly done ? bye Denes -- --- What kills me, doesn't make me stronger.
On Sat, Sep 30, 2006 at 06:30:39PM +0200, Balatoni Denes wrote:
Hi!
szombat 30 szeptember 2006 17.16-kor Oded Shimon ezeket a bolcs gondolatokat fogalmazta meg:
On Sat, Sep 30, 2006 at 06:09:02PM +0300, Oded Shimon wrote:
It would be silly for me to send this as a patch, because it's an almost completely unreadable ~260kb patch. These are the changes to be made:
1. the attached patch, for Makefile and vorbis.c
I don't find moving all the struct definitions from vorbis.h to vorbis.c a particularly good idea, actually. Why should they not be in a header file, as it's normaly done ?
2 problems I had with the structs being there. 1. since vorbis_enc also includes this header, the struct names might mix up together 2. for the typedefs you have to include bitstream.h and dsputil.h , which can bring confusion with bitstram.h weirdness :/ Is it really how it's normally done in ffmpeg? I see in most files the context is defined in the C file together... It's how it is in my vorbis encoder too. - ods15
Hi! szombat 30 szeptember 2006 17.31-kor Oded Shimon ezeket a bolcs gondolatokat fogalmazta meg:
2 problems I had with the structs being there. 1. since vorbis_enc also includes this header, the struct names might mix up together 2. for the typedefs you have to include bitstream.h and dsputil.h , which can bring confusion with bitstram.h weirdness :/
Is it really how it's normally done in ffmpeg? I see in most files the context is defined in the C file together... It's how it is in my vorbis encoder too.
Ok, you are right. Feel free to move those to vorbis.c than.
- ods15
bye Denes -- --- What kills me, doesn't make me stronger.
On Sat, Sep 30, 2006 at 06:57:14PM +0200, Balatoni Denes wrote:
Hi!
szombat 30 szeptember 2006 17.31-kor Oded Shimon ezeket a bolcs gondolatokat fogalmazta meg:
2 problems I had with the structs being there. 1. since vorbis_enc also includes this header, the struct names might mix up together 2. for the typedefs you have to include bitstream.h and dsputil.h , which can bring confusion with bitstram.h weirdness :/
Is it really how it's normally done in ffmpeg? I see in most files the context is defined in the C file together... It's how it is in my vorbis encoder too.
Ok, you are right. Feel free to move those to vorbis.c than.
Ok, keeping it simple, 2 attached patches. 1. makefile, vorbis.c, vorbis.h (130kb patch :) 2. vorbis.h -> vorbis_data.c will commit tommorrow ... (?) - ods15
Hi!
1. makefile, vorbis.c, vorbis.h (130kb patch :) 2. vorbis.h -> vorbis_data.c
btw. I might be a bit dense, but why not have one vorbis_data.h (note the .h), instead of vorbis.h and vorbis_data.c ? Is it faster to compile, or what?
- ods15
bye Denes
On Sat, Sep 30, 2006 at 07:11:48PM +0200, Balatoni Denes wrote:
Hi!
1. makefile, vorbis.c, vorbis.h (130kb patch :) 2. vorbis.h -> vorbis_data.c
btw. I might be a bit dense, but why not have one vorbis_data.h (note the .h), instead of vorbis.h and vorbis_data.c ? Is it faster to compile, or what?
vorbis_enc.c and vorbis.c share tables. if the tables are stored in a .h, they are duplicated needlessly in the output binary. - ods15
participants (5)
-
dbalatoni@interware.hu -
diego@biurrun.de -
michaelni@gmx.at -
mru@inprovide.com -
ods15@ods15.dyndns.org