| View previous topic :: View next topic |
| Author |
Message |
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3869
Location: Rasi, Finland
|
 Posted: Fri Mar 23, 2018 9:20 pm Post subject: |
 |
|
Libav, no matter how cleanly coded, is the prime of example how not to do things within open source community.
_________________
..: Zucca :..
| My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
| Quote: | | I am NaN! I am a man! |
Last edited by Zucca on Sat Mar 24, 2018 7:22 am; edited 1 time in total |
|
| Back to top |
|
 |
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Fri Mar 23, 2018 9:45 pm Post subject: |
|
|
It's the best example of how not to do a fork 
For how not to react to a fork, see Apache OpenOffice.
For how not to treat downstream developers, see Gtk+3.
For how not to treat anyone, see vapier. |
|
| Back to top |
|
|
Tony0945
Watchman
Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA
|
| Posted: Fri Mar 23, 2018 9:52 pm Post subject: |
|
|
| asturm wrote: | | You don't always have to try that hard. |
The hard part was restraining myself. As soon as I saw the name I edited /etc/portage/package.mask and forked portage-2.3.24 into /usr/local/portage/sys-apps/portage |
|
| Back to top |
|
|
krinn
Watchman
Joined: 02 May 2003
Posts: 7470
|
| Posted: Fri Mar 23, 2018 9:54 pm Post subject: |
|
|
| asturm wrote: | | krinn, on the other hand Portage should not be in the business of supporting out-of-tree packages just because some users refuse a dependency based on vague feelings. You should not base your critique of the fork on that particular bug though, it has nothing to do with it. |
stop also only thinking about out of tree package as it lower portage importance to only in tree package
adding a dep not use is bad, and worst that dep was added also for non gentoo users and non opnrc users, because portage is use on non gentoo (edit: got the name gentoo prefix), which that dep force them to install openrc or systemd
my critique  is base on his comment in the bug (not the bug itself) and the readme, which both are quiet clear about his attitude with portage dev.
and my critique on his tree is base on the validity of his comment on the bug, as not only he didn't express his disagreement nicely, but his jugement on the issue was wrong, which gave me a good level on how valuable will be his tree considering this bug was so simple to understand and fix. |
|
| Back to top |
|
|
Fitzcarraldo
Advocate
Joined: 30 Aug 2008
Posts: 2056
Location: United Kingdom
|
| Posted: Sat Mar 24, 2018 1:42 am Post subject: |
|
|
| Tony0945 wrote: | | As soon as I saw the name I edited /etc/portage/package.mask and forked portage-2.3.24 into /usr/local/portage/sys-apps/portage |
You have seen the new virtual/package-manager, right?
| Code: | $ eix virtual/package-manager
[I] virtual/package-manager
Available versions: 1
Installed versions: 1(01:07:22 23/03/18)
Description: Virtual for the package manager |
| Code: | $ cat /usr/portage/virtual/package-manager/package-manager-1.ebuild
# Copyright 1999-2018 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2
EAPI=5
DESCRIPTION="Virtual for the package manager"
SLOT="0"
KEYWORDS="alpha amd64 arm arm64 hppa ia64 m68k ~mips ppc ppc64 s390 sh sparc x86 ~ppc-aix ~x64-cygwin ~amd64-fbsd ~x86-fbsd ~amd64-linux ~arm-linux ~x86-linux ~ppc-macos ~x64-macos ~x86-macos ~m68k-mint ~sparc-solaris ~sparc64-solaris ~x64-solaris ~x86-solaris"
RDEPEND="
|| (
sys-apps/portage
sys-apps/portage-mgorny
sys-apps/paludis
sys-apps/pkgcore
)" |
_________________
Clevo W230SS: amd64, VIDEO_CARDS="intel modesetting nvidia".
Compal NBLB2: ~amd64, xf86-video-ati. Dual boot Win 7 Pro 64-bit.
OpenRC systemd-utils[udev] elogind KDE on both.
My blog |
|
| Back to top |
|
|
Tony0945
Watchman
Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA
|
| Posted: Sat Mar 24, 2018 2:53 am Post subject: |
|
|
| Fitzcarraldo wrote: | | You have seen the new virtual/package-manager, right? |
Do we really need everything virtualized? Doesn't just add to the already lengthy dependency calculations. |
|
| Back to top |
|
|
John R. Graham
Administrator
Joined: 08 Mar 2005
Posts: 10691
Location: Somewhere over Atlanta, Georgia
|
| Posted: Sat Mar 24, 2018 3:57 am Post subject: |
|
|
Virtuals like that add virtually (couldn't resist) no time to the dependency calculations because they have minimal depth.
- John
_________________
I can confirm that I have received between 0 and 499 National Security Letters. |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6069
Location: Removed by Neddy
|
| Posted: Sat Mar 24, 2018 9:07 am Post subject: |
|
|
| John R. Graham wrote: | Virtuals like that add virtually (couldn't resist) no time to the dependency calculations because they have minimal depth.
- John |
Exactly, now the specific contributor to the virtual may increase dependency calculations but that is package specific.
...
This can go one of two way and both ways have occured in Gentoo's history.
1) A positive contribution. There are issues and maintaining historic capability seems redundant. If a scrum could occur on certain functions to get working to then be back ported then fantastic. Pkgcore was like this... a separate creation to speed aspects up (snakeoil) and ports stuff back. There was good relations between pkgcore devs and portage dev's.
What makes me doubt this is the wording in the mission statement Why not branch on the portage tree WHY fork like this...
2) A negative contribution. If the devo in question cannot articulate the problem and thus "shows how bad it is" in code to make a point not to resolve the situation then this will go sour very quickly. Paludis started out as a dependency QA resolver to throw back at dev's as how bad they are. This just creates toxic behaviour and well.. the cabal of paludis devo's and what they did is well known. If we are heading that way again this needs to be nipped in the bud.
Also mgorny seem to have a selective memory...
| Quote: | | However, its goal is to follow the example set by projects such as libav and libressl. They never became mainstream but they made the respective original projects |
LibAV was disgusting.. hostile pollution of the namespace while breaking API and ABI compatibility, breaking security. THAT is not how you do a fork, THAT is not a good way to improve another program.
Forks can rejuvenate development (look at vim and neovim) but if he views libav as a good example, he needs to be very careful about what is "good"
_________________
#define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0; |
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Sat Mar 24, 2018 9:40 am Post subject: |
|
|
| Naib wrote: | | Why not branch on the portage tree WHY fork like this... |
Doing it under a different project name makes it idiot-proof from the sort of idiot who goes looking for ways to shoot himself in the foot so he can whine about it, such as installing an experimental package manager over his only working one. You know, people like, uhh... miroR. That guy was a real headcase.
Putting it in a different repo entirely might be insurance against brain-dead bots like that one that auto-merged the zlib subslot patch that wasn't actually approved a while back, wasting days of everyone's time on pointless rebuilds...
Or it could just be a euphemistic "eff you" gesture to other devs, but I'll pretend it's not that until I can't any more. libav left no room for doubt in that regard. |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6069
Location: Removed by Neddy
|
| Posted: Sat Mar 24, 2018 11:40 am Post subject: |
|
|
| Ant P. wrote: |
Doing it under a different project name makes it idiot-proof from the sort of idiot who goes looking for ways to shoot himself in the foot so he can whine about it, such as installing an experimental package manager over his only working one. You know, people like, uhh... miroR. That guy was a real headcase.
|
The one thing this universe is great at making is better idiots
_________________
#define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0; |
|
| Back to top |
|
|
Tony0945
Watchman
Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA
|
| Posted: Sat Mar 24, 2018 1:30 pm Post subject: |
|
|
| Naib wrote: | | The one thing this universe is great at making is better idiots |
That's something we can all agree with! At least "bigger" if not "better", |
|
| Back to top |
|
|
krinn
Watchman
Joined: 02 May 2003
Posts: 7470
|
| Posted: Sat Mar 24, 2018 2:07 pm Post subject: |
|
|
| Tony0945 wrote: | | Naib wrote: | | The one thing this universe is great at making is better idiots |
That's something we can all agree with! At least "bigger" if not "better", |
Better said like that?
| Quote: | “Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.”
― Albert Einstein |
|
|
| Back to top |
|
|
Tony0945
Watchman
Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA
|
| Posted: Sat Mar 24, 2018 2:27 pm Post subject: |
|
|
| krinn wrote: | | Quote: | “Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.”
― Albert Einstein |
|
Having just left a political blog, Einstein's quote hits home! |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 22992
|
| Posted: Sat Mar 24, 2018 4:42 pm Post subject: |
|
|
| tld wrote: | | The reason I refused to to within 1,000 miles of that one was their decision to use the same fucking library names as ffmpeg...just to force everyone to make a choice. |
I can see two good technical reasons for reusing library names, though I recognize that there are some good reasons against it, too. First, if the fork is expected to be drop-in compatible (say, MySQL vs MariaDB), reusing names lets the system administrator switch implementations without needing to rebuild programs or modify scripts. Second, if the upstream (in this case, ffmpeg) build system is bad enough, fixing the names could be a huge pain.
On the other hand, and to your point, if the fork is not drop-in compatible, whether due to intentionally making incompatible changes or even merely failing to follow upstream's changes, reusing the names makes a huge mess. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6780
|
| Posted: Sat Mar 24, 2018 5:27 pm Post subject: |
|
|
| Hu wrote: | | First, if the fork is expected to be drop-in compatible |
In case of libav it was declared intention of the fork that this will not be the case: Functionality should be dropped, other implemented differently.
| Quote: | | build system is bad enough, fixing the names could be a huge pain. |
A build system where it is a serious technical problem to change the name of produced library files or of installation paths under /usr/include still has to be inventend.
It was in fact intentional (perhaps even declared explicitly, I do not remember) to force users to decide in order to harm the original project. |
|
| Back to top |
|
|
krinn
Watchman
Joined: 02 May 2003
Posts: 7470
|
| Posted: Sat Mar 24, 2018 10:02 pm Post subject: |
|
|
You are memory short ; it was intentional and more than just using same library name, they have tried to take over the ffmpeg name, as they were owning the server administration... because ffmpeg name is well know and a famous tool.
What they have tried to do was never seen, and will certainly not be seen any more, because it could only be range in the category: totally pure asshole action!
I don't have better than this at hand https://www.reddit.com/r/linux/comments/vvdxn/the_ffmpeglibav_situation/ |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6069
Location: Removed by Neddy
|
| Posted: Sat Mar 24, 2018 10:14 pm Post subject: |
|
|
| krinn wrote: | You are memory short ; it was intentional and more than just using same library name, they have tried to take over the ffmpeg name, as they were owning the server administration... because ffmpeg name is well know and a famous tool.
What they have tried to do was never seen, and will certainly not be seen any more, because it could only be range in the category: totally pure asshole action!
I don't have better than this at hand https://www.reddit.com/r/linux/comments/vvdxn/the_ffmpeglibav_situation/ |
pretty much. Go find the thread here in gentoo chat ( here it is: https://forums.gentoo.org/viewtopic-t-1010096-highlight-libav.html ) where we the users pushed back because some asshole gentoo developer decided to push politics (make libav the default in an obnoxious way in gentoo) over quality...
It links to loads of security issues with libav, asshatery namespace pollution, breakage with other applications etc...
Best bit is the same gentoo developer has cited libav as why forks are good... Any other developer and this fork would have been viewed as engaging but this one? I question his motives, question his morals, question his real intent
--edit--
http://blog.pkh.me/p/13-the-ffmpeg-libav-situation.html
another good summary. Basically any developer who cites libAV as why forks are good needs to step back and REALLY think whether how they act are in the best interest of opensource in general because personally I would not want anything they do near my stuff. Now there are two VERY questionable gentoo developers... time to look at archlinux again
_________________
#define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0;
Last edited by Naib on Sat Mar 24, 2018 10:23 pm; edited 3 times in total |
|
| Back to top |
|
|
mike155
Advocate
Joined: 17 Sep 2010
Posts: 4438
Location: Frankfurt, Germany
|
| Posted: Sat Mar 24, 2018 10:15 pm Post subject: |
|
|
| I would like to install both versions of portage (traditional portage and mgorny portage) in parallel. Then I could easily compare them. |
|
| Back to top |
|
|
blopsalot
Apprentice
Joined: 28 Jan 2017
Posts: 231
|
| Posted: Mon Mar 26, 2018 6:00 pm Post subject: |
|
|
Regarding libav, for someone that was involved heavily in digital video during that time frame, I think it is fair to add the Michael Niedermayer was being unreasonable and it took all that craziness for him to realize it. He finally commited multithreading after they did and he continued to backport everything they added even though it was rejected previously. The repercussions on the average end user were unfortunate, but the libav people were not hacks, they were just pushing forward and they were MUCH friendlier downstream. ffmpeg was too slow for HD without mt patches. That is only one example of cool stuff he was not including. now there's no reason not to use ffmpeg, but it did serve a purpose.
As far as this fork goes, I like it, repoman is back included, he doesn't actually force usage of gemato outside of the ebuild (i use git repo), its smaller, its faster, no breakage so far.
Thank you mgorny for all your work but please leave more stuff configurable through use flags. We Gentoo users that you havent heard from much in the last 8 years like control. 
and free the ml!!! |
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Mon Mar 26, 2018 7:16 pm Post subject: |
|
|
| blopsalot wrote: | | but the libav people were not hacks, they were just pushing forward and they were MUCH friendlier downstream. |
Sending legal threats over a logo they had no rights to is not my idea of a friendly downstream. |
|
| Back to top |
|
|
blopsalot
Apprentice
Joined: 28 Jan 2017
Posts: 231
|
| Posted: Mon Mar 26, 2018 7:19 pm Post subject: |
|
|
| Ant P. wrote: | | blopsalot wrote: | | but the libav people were not hacks, they were just pushing forward and they were MUCH friendlier downstream. |
Sending legal threats over a logo they had no rights to is not my idea of a friendly downstream. |
Yeah it was crazy. I do not agree with their actions but from a technical standpoint, that group of people had proven code that made interfacing with the libraries easier and much faster. |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6069
Location: Removed by Neddy
|
| Posted: Mon Mar 26, 2018 7:49 pm Post subject: |
|
|
| blopsalot wrote: | | Ant P. wrote: | | blopsalot wrote: | | but the libav people were not hacks, they were just pushing forward and they were MUCH friendlier downstream. |
Sending legal threats over a logo they had no rights to is not my idea of a friendly downstream. |
Yeah it was crazy. I do not agree with their actions but from a technical standpoint, that group of people had proven code that made interfacing with the libraries easier and much faster. |
at the expense of security and compatibility with other applications. It was messed up, bad form, not the way to operate....
All they had todo was not pollute the namespace then it would be those that WANT to use said libs problem....
Those that advocate libAV as a forking methodology are a threat to FOSS
_________________
#define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0; |
|
| Back to top |
|
|
asturm
Developer
Joined: 05 Apr 2007
Posts: 9325
|
| Posted: Mon Mar 26, 2018 8:21 pm Post subject: |
|
|
| blopsalot wrote: | | Thank you mgorny for all your work but please leave more stuff configurable through use flags. We Gentoo users that you havent heard from much in the last 8 years like control. |
he's forked Portage, not the gentoo ebuild repository. |
|
| Back to top |
|
|
blopsalot
Apprentice
Joined: 28 Jan 2017
Posts: 231
|
| Posted: Mon Mar 26, 2018 8:28 pm Post subject: |
|
|
| Naib wrote: | | blopsalot wrote: | | Ant P. wrote: | | blopsalot wrote: | | but the libav people were not hacks, they were just pushing forward and they were MUCH friendlier downstream. |
Sending legal threats over a logo they had no rights to is not my idea of a friendly downstream. |
Yeah it was crazy. I do not agree with their actions but from a technical standpoint, that group of people had proven code that made interfacing with the libraries easier and much faster. |
at the expense of security and compatibility with other applications. It was messed up, bad form, not the way to operate....
All they had todo was not pollute the namespace then it would be those that WANT to use said libs problem....
Those that advocate libAV as a forking methodology are a threat to FOSS |
Fabrice Bellard is a threat to FOSS? that's funny. there was no methodology. they had a falling out and it was messy. Michael backported their patches. ffmpeg and libav coexist as colloborative projects still today. ffmpeg is amazing because of it all. happy ending.
| asturm wrote: | | blopsalot wrote: | | Thank you mgorny for all your work but please leave more stuff configurable through use flags. We Gentoo users that you havent heard from much in the last 8 years like control. |
he's forked Portage, not the gentoo ebuild repository. |
i was wondering why it wasnt compiling. |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6069
Location: Removed by Neddy
|
| Posted: Mon Mar 26, 2018 8:51 pm Post subject: |
|
|
| blopsalot wrote: | | Naib wrote: | | blopsalot wrote: | | Ant P. wrote: | | blopsalot wrote: | | but the libav people were not hacks, they were just pushing forward and they were MUCH friendlier downstream. |
Sending legal threats over a logo they had no rights to is not my idea of a friendly downstream. |
Yeah it was crazy. I do not agree with their actions but from a technical standpoint, that group of people had proven code that made interfacing with the libraries easier and much faster. |
at the expense of security and compatibility with other applications. It was messed up, bad form, not the way to operate....
All they had todo was not pollute the namespace then it would be those that WANT to use said libs problem....
Those that advocate libAV as a forking methodology are a threat to FOSS |
Fabrice Bellard is a threat to FOSS? that's funny. there was no methodology. they had a falling out and it was messy. Michael backported their patches. ffmpeg and libav coexist as colloborative projects still today. ffmpeg is amazing because of it all. happy ending.
| asturm wrote: | | blopsalot wrote: | | Thank you mgorny for all your work but please leave more stuff configurable through use flags. We Gentoo users that you havent heard from much in the last 8 years like control. |
he's forked Portage, not the gentoo ebuild repository. |
i was wondering why it wasnt compiling. |
Did they or did they not attempt a hostile takeover of the namespace
_________________
#define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0; |
|
| Back to top |
|
|
|
Gentoo Chat
