[pacman-dev] Time for changes
Ok, I think it's about time I picked up pacman developement, seeing as we really have no one else doing it. I've been busy, so I was really hoping someone else would volunteer, but what the hell. If you frugalware guys could send me a full diff (here or offlist) of your repo compared to the trunk, that'd be great. I don't like applying patches without verifying them, so I'm going to step through each hunk (this is why I originally requested them as the seperate patches sent to the ML, but I can deal with it either way). One thing i plan on doing right off the bat: I want to enable a side-by-side install with the current pacman, so that we can get some real testing by those (like me) who don't want to overwrite the stable pacman. This is just informational, but it shouldn't affect the frugalware guys. To the furgalware team: you have more experience with this new pacman that I do... could you give me a brief "what's changed" - functionality wise. I mean, when actually using pacman3, what is done differently? What future directions are we going for? I'd like to get this stable as soon as possible, but once that's done, I have a handful of nifty features I wanted to add (those on the dev list: I want to add filesystem hooks first, then some more advanced output logging). There are also some potential changes to gensync and friends, depending on what direction we go with some svn changes... but that's off topic. Anyway, let me know. Aaron Griffin - phrakture -
On Thu, Sep 28, 2006 at 01:48:28PM -0500, Aaron Griffin wrote:
Ok, I think it's about time I picked up pacman developement, seeing as we really have no one else doing it. I've been busy, so I was really hoping someone else would volunteer, but what the hell.
I'll get your back when I can, Aaron. I've merged most of the libarchive stuff in, but it's not committed yet. For some reason, the static linking fails. Dynamic one works fine. I tested this with FW's pacman and I had the same error, so it's likely something on my system, not in the pacman build setup. - J
On 9/28/06, Judd Vinet <jvinet@zeroflux.org> wrote:
On Thu, Sep 28, 2006 at 01:48:28PM -0500, Aaron Griffin wrote:
Ok, I think it's about time I picked up pacman developement, seeing as we really have no one else doing it. I've been busy, so I was really hoping someone else would volunteer, but what the hell.
I'll get your back when I can, Aaron. I've merged most of the libarchive stuff in, but it's not committed yet. For some reason, the static linking fails. Dynamic one works fine. I tested this with FW's pacman and I had the same error, so it's likely something on my system, not in the pacman build setup.
While you're available... do you seriously like sw=2 ts=2 ? I was going to bring that up and see if it offended anyone to switch to 4/4 - though I guess if you just use noet, and remove those settings, it'd be up to the editor (modelines override editor settings).
On Thu, Sep 28, 2006 at 01:48:28PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
If you frugalware guys could send me a full diff (here or offlist) of your repo compared to the trunk, that'd be great. I don't like applying patches without verifying them, so I'm going to step through each hunk (this is why I originally requested them as the seperate patches sent to the ML, but I can deal with it either way).
i'll send one soon. yes, separate small patches are nice, but most splitted patch sent to this list is already outdated, so i think reviewing the big patch will be less work for you
One thing i plan on doing right off the bat: I want to enable a side-by-side install with the current pacman, so that we can get some real testing by those (like me) who don't want to overwrite the stable pacman. This is just informational, but it shouldn't affect the frugalware guys.
that's exactly what we did before we switched to pacman3 (i've just noticed i still have a pacman2 static binary on my system :) )
To the furgalware team: you have more experience with this new pacman that I do... could you give me a brief "what's changed" - functionality wise. I mean, when actually using pacman3, what is done differently?
huh, lots of stuff. i think the best is to read both the cvs log and our changelog. unfortunelly, in the early days i haven't included in the relevant cvs logs in my "synced with upstream" patches. but i'll include a changelog in the header of the big patch
What future directions are we going for? I'd like to get this stable as soon as possible, but once that's done, I have a handful of nifty features I wanted to add (those on the dev list: I want to add filesystem hooks first, then some more advanced output logging).
when i first saw pacman, i had two big issues and two small ones: - if foo conflicted with bar, and foo was installed pacman -S bar did not work (that patch is already merged to the cvs) - insufficed error messages (mostly done by Aurel) the not-so-urgent ones: - the download code was in pacman, not in the library (this is not merged yet) - the config parser was in pacman, not in the library (this is not merged yet) these were the main issues and all is solved in our tree, so i don't have big plans, users always have enough small feature requests :)
There are also some potential changes to gensync and friends, depending on what direction we go with some svn changes... but that's off topic.
there isn't any cvs-specific code in gensync, afaik. in our tree there is a oneliner reference to darcs to ignore the _darcs dir when searching for FrugalBuilds i hope i can send you the big patch in 1-2 days udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
2006/9/28, Judd Vinet <jvinet@zeroflux.org>:
On Thu, Sep 28, 2006 at 01:48:28PM -0500, Aaron Griffin wrote:
Ok, I think it's about time I picked up pacman developement, seeing as we really have no one else doing it. I've been busy, so I was really hoping someone else would volunteer, but what the hell.
I'll get your back when I can, Aaron. I've merged most of the libarchive stuff in, but it's not committed yet. For some reason, the static linking fails. Dynamic one works fine. I tested this with FW's pacman and I had the same error, so it's likely something on my system, not in the pacman build setup.
Finally! :-D Thanks, guys! BTW, it's offtopic a bit, but I'm just curious - where is Aurelien Foret? I know that he discontinued Pacman 3 development, but where is he? I heard nothing about him for more than half of the year. -- Roman Kyrylych (Роман Кирилич)
On 9/28/06, VMiklos <vmiklos@frugalware.org> wrote:
On Thu, Sep 28, 2006 at 01:48:28PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
One thing i plan on doing right off the bat: I want to enable a side-by-side install with the current pacman, so that we can get some real testing by those (like me) who don't want to overwrite the stable pacman. This is just informational, but it shouldn't affect the frugalware guys.
that's exactly what we did before we switched to pacman3 (i've just noticed i still have a pacman2 static binary on my system :) )
Yeah, though I'd probably reverse that and, basically add a "3" to everything - pacman3, /etc/pacman3.d/, /var/lib/pacman3 etc etc - I don't want to mess up existing databases/configs at all.
huh, lots of stuff. i think the best is to read both the cvs log and our changelog. unfortunelly, in the early days i haven't included in the relevant cvs logs in my "synced with upstream" patches. but i'll include a changelog in the header of the big patch
Ok, I don't have a link to your web view (darcs IIRC), can you send that, so I can take a peek - it's probably on your front page somewhere, but I'm at work and rather busy.
when i first saw pacman, i had two big issues and two small ones: - if foo conflicted with bar, and foo was installed pacman -S bar did not work (that patch is already merged to the cvs)
I'm not sure I understand this one. If foo and bar conflict, bar shouldn't be installed if foo is installed... can you explain this case more?
these were the main issues and all is solved in our tree, so i don't have big plans, users always have enough small feature requests :)
I have a few big-ish ideas... mainly a few ideas to fix up repetative .install files - in short, some sort of file-matching config and commands to run when something is matched - if the /opt/mono/docs dir has been changed in this pacman run, rerun monodoc generation (right now, monodoc reruns about 5 times on a decent mono upgrade) - same with, say, xorg font dirs. This will remove quite alot of repetative install files (depmod for modules, all the font stuff for fonts, all the gnome stuff, vim doc regeneration for vim plugins, etc etc).
There are also some potential changes to gensync and friends, depending on what direction we go with some svn changes... but that's off topic.
there isn't any cvs-specific code in gensync, afaik. in our tree there is a oneliner reference to darcs to ignore the _darcs dir when searching for FrugalBuilds
It's not with relation to cvs or svn really, but some ideas on how things can be changed _if_ we go about reorganizing things in an svn repo. It's still rather up in the air right now.
i hope i can send you the big patch in 1-2 days
Awesome, thanks alot.
On Thu, Sep 28, 2006 at 04:00:43PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
Ok, I don't have a link to your web view (darcs IIRC), can you send that, so I can take a peek - it's probably on your front page somewhere, but I'm at work and rather busy.
http://darcs.frugalware.org/darcsweb/darcsweb.cgi?r=pacman
when i first saw pacman, i had two big issues and two small ones: - if foo conflicted with bar, and foo was installed pacman -S bar did not work (that patch is already merged to the cvs)
I'm not sure I understand this one. If foo and bar conflict, bar shouldn't be installed if foo is installed... can you explain this case more?
pacman2 offered to remove foo and installed bar. nowadays pacamn3 does the same
This will remove quite alot of repetative install files (depmod for modules, all the font stuff for fonts, all the gnome stuff, vim doc regeneration for vim plugins, etc etc).
the ability to create such hooks is a great idea, imho :) udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On Thu, Sep 28, 2006 at 10:44:54PM +0200, VMiklos <vmiklos@frugalware.org> wrote:
i hope i can send you the big patch in 1-2 days
here is a raw patch: http://frugalware.org/~vmiklos/patches/pacman-proposed/pacman-20060929-fruga... current problems: 1) since cvs uses that stupid context diff, the real changes to the manpages won't be shown since they are renamed (it was necessary because they can't be generated if we want to translate them) 2) i haven't excluded the po files, so the real code changes aren't separated from the translations 3) of course there maybe be some changes what you don't want to merge (for example having the archs() array in FrugalBuilds is a good example, i think since you (why the hell?) started to use arch() _after_ we started to use archs()..) if you want me to exclude pactest/po files/whatever from the diff, then just ask udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
2006. 09. 28, csütörtök keltezéssel 12.09-kor Judd Vinet ezt írta:
On Thu, Sep 28, 2006 at 01:48:28PM -0500, Aaron Griffin wrote:
Ok, I think it's about time I picked up pacman developement, seeing as we really have no one else doing it. I've been busy, so I was really hoping someone else would volunteer, but what the hell.
I'll get your back when I can, Aaron. I've merged most of the libarchive stuff in, but it's not committed yet. For some reason, the static linking fails. Dynamic one works fine. I tested this with FW's pacman and I had the same error, so it's likely something on my system, not in the pacman build setup.
Nice to hear this :) I like libarchive patch and i hope you can deal with it. Once that done, you can apply many patches that depends on libarchive patchset. Like the progressbar, package-counter, etc. Huhm any error about this? :) Maybe some autotools* braindead problem and maybe i can solve it :) An idea for this. BE sure that libarchive NOT compiled with ACL support. libarchive does only checks and if ACL available in system then it pulls it into itself and links with -lacl . pacman's autotools* now cant handle this, because there is no way to discover that libarchive was compiled with ACL or not. I talked about the author of libarchive i sent some patches about ACL disable function, but no reply until this. So be sure that libarchive not compiled with ACL support. Or if yes, then rewrite configure.ac part and put a -lacl into CFLAGS. Another: I noticed when you build with --disable-static (in configure) then it wants to do pacman.static (libalpm.static not created) as i remember. I think this bug still exists. But it just a cosmetics to autotools* :) Regards Christian Hamar alias krix
On Fri, Sep 29, 2006 at 08:27:55AM +0200, Christian Hamar <krics@gds.hu> wrote:
Nice to hear this :) I like libarchive patch and i hope you can deal with it.
2006-09-28 22:51 judd * configure.ac, lib/libalpm/add.c, lib/libalpm/alpm.h, lib/libalpm/error.c, lib/libalpm/package.c, lib/libalpm/sync.c, lib/libalpm/util.c, lib/libalpm/util.h: removed libtar support in favour of libarchive :) udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On Thu, 28 Sep 2006 21:22:37 +0200, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 9/28/06, Judd Vinet <jvinet@zeroflux.org> wrote:
On Thu, Sep 28, 2006 at 01:48:28PM -0500, Aaron Griffin wrote:
Ok, I think it's about time I picked up pacman developement, seeing as we really have no one else doing it. I've been busy, so I was really hoping someone else would volunteer, but what the hell.
I'll get your back when I can, Aaron. I've merged most of the libarchive stuff in, but it's not committed yet. For some reason, the static linking fails. Dynamic one works fine. I tested this with FW's pacman and I had the same error, so it's likely something on my system, not in the pacman build setup.
While you're available... do you seriously like sw=2 ts=2 ? I was going to bring that up and see if it offended anyone to switch to 4/4 - though I guess if you just use noet, and remove those settings, it'd be up to the editor (modelines override editor settings).
Good to see some action here again :) I wanted to see if the slow operation with XFS/JFS? could be improved. So i made a script that finds packages without using pacman. It's called pacfind. or whatever you want to call it :) (i think someone alreayd have made a pacfind script? #!/bin/sh if [ "$1" = "-Sd" -a -n "$2" ] then cd /var/lib/pacman/ grep -r -A 1 DESC *|grep -v local/|egrep "desc-.*$2.*"|sed 's/.*desc-/^[[1;33m&\n^[[1;37m/'|sed 's/\/desc-//' # grep -A 1 DESC */*/desc didn't work for some reason :/ elif [ "$1" = "-Ss" -a -n "$2" ] then cd /var/lib/ find pacman -name "*$2*"|grep -ve /local -e '.*testing$' -e '.*AqD$' -e '.*brain0$' -e '.*community$' -e '.*current$' -e '.*extra$' -e '.*karsten$' -e '.*unstable$'|xargs grep -r -A 1 DESC|grep -v DESC -v |grep -v -- --|sed 's|^pacman/||'|sed 's/.*desc-/^[[1;33m&\n^[[1;37m/'|sed 's/\/desc-//' fi # the escape codes won't work ofc. # The -Sd search packgae descriptions # The -Ss search only packagenames Results: searching for packagename takes 15-16 sec now which used to take 1.04m with pacman -Ss -Sd takes 56 sec but is really quite faster than pacman's 1.04m because it starts sooner and only spits out the info slower because of coloring, i suppose. So there's definitively some ground to be covered for pacman and I really hope that pacman3 won't have to be db dependent <:) Also my script could probably be improved by some of you bash gurus :) and also have you guys seen this?: http://bbs.archlinux.org/viewtopic.php?t=22872&highlight=pacman+poll :) me out! -- Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
On Fri, Sep 29, 2006 at 04:36:10PM +0200, kfs1@online.no wrote:
I wanted to see if the slow operation with XFS/JFS? could be improved. So i made a script that finds packages without using pacman.
It's called pacfind. or whatever you want to call it :) (i think someone alreayd have made a pacfind script?
unfortunatelly it will only work with the 'files' backend, if we switch to an other backend (to improve pacman's speed) then it won't work at all udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On Fri, 29 Sep 2006 17:12:13 +0200, VMiklos <vmiklos@frugalware.org> wrote:
On Fri, Sep 29, 2006 at 04:36:10PM +0200, kfs1@online.no wrote:
I wanted to see if the slow operation with XFS/JFS? could be improved. So i made a script that finds packages without using pacman.
It's called pacfind. or whatever you want to call it :) (i think someone alreayd have made a pacfind script?
unfortunatelly it will only work with the 'files' backend, if we switch to an other backend (to improve pacman's speed) then it won't work at all
that's what i'm saying... did you read the rest of the email? Anyway, I suppose it doesn't matter much since i'm talking to the head of frugalware pacman not arch. -- Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
Ok, I've been merging things piece by piece. Judd, I'm still waiting for write access to the cvs repo (maybe Xentac can do it too?) - maybe you did this and I just don't know about it 8). On a side note, something which is rather important, I noticed on quite a few of these files you guys had removed judd's name in the copyright, and added your own. If I'm not mistaken, this is a rather severe GPL violation - something along the lines of "permission is granted as long as this copyright remains in tact". Please fix this. Thanks, Aaron
On Tue, Oct 03, 2006 at 04:29:53PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On a side note, something which is rather important, I noticed on quite a few of these files you guys had removed judd's name in the copyright, and added your own. If I'm not mistaken, this is a rather severe GPL violation - something along the lines of "permission is granted as long as this copyright remains in tact". Please fix this.
you can find a script called copyright.py in the patch, that was used to fix up the copyrights. as far as i know Judd's name was only removed where the author of the file was Aurelien or me (like libalpm/server.c) if you think somewhere it was removed where it should not be, then probably the cvs logs are incorrect but feel free to came up with a list of "invalid" removes adding proper copyright lines are a lot of time and when (according to the cvs logs) judd never touched a file, i don't think he should be in the copyright list. see the kernel source, you can't find Linus' name in every file and of course the basic purpose of the while copyright update is to appropriate Aurelien's and others' work, not hiding Judd's one, i hope this is clear :) udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On 10/3/06, VMiklos <vmiklos@frugalware.org> wrote:
On Tue, Oct 03, 2006 at 04:29:53PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On a side note, something which is rather important, I noticed on quite a few of these files you guys had removed judd's name in the copyright, and added your own. If I'm not mistaken, this is a rather severe GPL violation - something along the lines of "permission is granted as long as this copyright remains in tact". Please fix this.
you can find a script called copyright.py in the patch, that was used to fix up the copyrights. as far as i know Judd's name was only removed where the author of the file was Aurelien or me (like libalpm/server.c)
Does not matter. The GPL explicitly states that the copyright lines MUST be maintained. They were not.
if you think somewhere it was removed where it should not be, then probably the cvs logs are incorrect but feel free to came up with a list of "invalid" removes
It should never be removed, according to the GPL.
adding proper copyright lines are a lot of time and when (according to the cvs logs) judd never touched a file, i don't think he should be in the copyright list. see the kernel source, you can't find Linus' name in every file
Again, this does not matter. They were there originally. It is not about who did what work, it is about changing and explicitly violating the copyright. Frankly, I find it rather rude that you guys would go through and say "whoops, no judd here". It's not something you do lightly, as it required you to explicitly go through CVS logs to check. Even then, there's alot of changes on files that have no other changes (at least in this diff) beyond just the copyright notices. I know people take this stuff lightly, but in reality this is a crucial point of the GPL.
On Tue, Oct 03, 2006 at 06:02:51PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
Again, this does not matter. They were there originally. It is not about who did what work, it is about changing and explicitly violating the copyright.
i am not a lawyer. i don't know how legal saying "i the author of something" when "i am definitely not". when Aurelien created a new file from scratch, he added a "copyright 2002-2006 Judd" line, which is - afaik - not valid, since: 1) the author of the file is Aurelien 2) the file was nowhere in 2002
Frankly, I find it rather rude that you guys would go through and say "whoops, no judd here". It's not something you do lightly, as it required you to explicitly go through CVS logs to check.
the copyrights were not valid, since they stated the only copyright holder is Judd which is not true. so we had to fix them and the easiest way was to fix them according to the cvs logs
Even then, there's alot of changes on files that have no other changes (at least in this diff) beyond just the copyright notices.
I know people take this stuff lightly, but in reality this is a crucial point of the GPL.
okay, i hope we can discuss this and finally the invalid copyrights can be fixed in the cvs. if you don't like our script then fix them by hand: the current situation is not clear, since there are more than one people in the AUTHORS file, but the copyrights in the source files states the only copyright holder is Judd and again, please show where are the files where Judd's line is removed http://darcs.frugalware.org/darcsweb/darcsweb.cgi?r=pacman;a=darcs_commitdif...; ^ the only file here is server.h, to where _i_ was who added Judd's copyright line since i've just copy&pasted the header from an other file udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On 10/3/06, VMiklos <vmiklos@frugalware.org> wrote:
the copyrights were not valid, since they stated the only copyright holder is Judd which is not true. so we had to fix them and the easiest way was to fix them according to the cvs logs
Using cvs is 100% not viable. There are 100 cases I can think of where judd would not be logged in cvs, yet still had wrote some of it... for instance (please read the first case a few times): * He was the author of pacman 2, on which this code is based, thus satisfying the GPL * He emailed the files to someone else * The files were moved from one repository to another, and changes were not preserved * The originals were made outside of cvs, and the first check in was done by another user I can go on and on. Simple fact of the matter is that you cannot use CVS logs to do this. Hey, none of you guys have your names listed in our cvs, can I remove those? It's still just rather rude to go about and do these things. I know you guys are trying to contribute and all, but this is rather inane. And besides, you talk about "invalid copyrights".... says who? It is completely within legal limits for an author to not be a copyright holder. To me, i see it as disrespectful to Judd more than anything else... I mean, if someone sent me a patch for something I originally wrote that had around 100 lines of things like this: - * Copyright (c) 2002-2006 by Judd Vinet <jvinet@zeroflux.org> + * Copyright (c) 2005 by Judd Vinet <jvinet@zeroflux.org> and -COPYRIGHT_HOLDER = Judd Vinet +COPYRIGHT_HOLDER = Yoyodyne, Inc. I would be rather offended. In fact, it's not even me and it offends the crap out of me.
Hi! I'm not a lawyer, nor I'm strong in interpreting GPL terms, but "one copyright holder - many authors" is common thing in many projects. For example in every source file of SDL (except in those that are just included from other projects) there are the following lines: SDL - Simple DirectMedia Layer Copyright (C) 1997-2006 Sam Lantinga And it doesn't matter that some file didn't exist in 1997, and that most of audio code was (and continues to be) written by Ryan Gordon. There are also other contributors, i.e. maintainers of ports for other platforms. But there is a file named CREDITS, where _all_ contributions and their authors are listed, with many thanks to them. And this is just fine, because in other way there wolud be 10-20 lines of copyright info in many files. As for generating any copyrights info from CVS - then you forget about those who has no acces to CVS but contributed some patches. -- Roman Kyrylych (Роман Кирилич)
On Wed, Oct 04, 2006 at 01:31:53AM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 10/3/06, VMiklos <vmiklos@frugalware.org> wrote: Using cvs is 100% not viable. There are 100 cases I can think of where judd would not be logged in cvs, yet still had wrote some of it... for instance (please read the first case a few times): * He was the author of pacman 2, on which this code is based, thus satisfying the GPL
okay, we need facts. what about alpm.c? i _do_ think it was nowhere in 2002
* He emailed the files to someone else * The files were moved from one repository to another, and changes were not preserved * The originals were made outside of cvs, and the first check in was done by another user
that's the situation with Aurelien, not with Judd. i don't know if Judd is the author of alpm.c or not. as you can see i haven't removed his name even originally the whole file may be writted by Aurelien
I can go on and on. Simple fact of the matter is that you cannot use CVS logs to do this. Hey, none of you guys have your names listed in our cvs, can I remove those?
1) if not listed, then what to remove? 2) see the cvs logs, see the AUTHORS file. if that makes you happy, then feel free to remove me, ask Christian and Aurelien about their lines
It's still just rather rude to go about and do these things. I know you guys are trying to contribute and all, but this is rather inane.
stop. it was me who updated the copyright lines, this issue has nothing with Christian or any other developer
And besides, you talk about "invalid copyrights".... says who? It is completely within legal limits for an author to not be a copyright holder.
okay, i'm still not a lawyer. who decides if the author is the copyright holder is the author or not?
-COPYRIGHT_HOLDER = Judd Vinet +COPYRIGHT_HOLDER = Yoyodyne, Inc.
feel free to flame me, but this is already checked in by Judd: $ find . -type f |xargs grep 'COPYRIGHT_HOLDER =' ./lib/libalpm/po/Makevars:COPYRIGHT_HOLDER = Judd Vinet ./src/pacman/po/Makevars:COPYRIGHT_HOLDER = Yoyodyne, Inc. and of course you can see, that's the default, so the story imho is the following: 1) when krix added the files to our tree, he fogot to change the default copyright holder 2) he sent here the patch 3) Judd corrected the copyright line 4) we never noticed he modified the patch we sent him conclusion: we never changed "Judd Vinet" to "Yoyodyne, Inc."
I would be rather offended. In fact, it's not even me and it offends the crap out of me.
yes, that's strange. i could say don't talk about this at all, because it's Judd and my business, definitely not your one maybe you don't see, but i still would like to be constructive. let's try to turn this endless flame thread to some positive discussion and let's try to find out a solution 1) if you have enough time, then please fix the copyright lines in the cvs by hand, and when their are fixed, i'm sure the difference between the arch tree and in our tree will be minimal and i'll change the headers in our tree so that there will be no difference 2) please understand that the current state in the cvs is invalid. if you have time, then search for the first pacman release that featured pacman's -T option. even in our tree the copyright says pacman_deptest() is from 2002, which is obviously not true. same for alpm.h and so on 3) if you think the copyright lines in the cvs are valid, then please have a look at the kernel's source, there is zero "copyright 1991-2006 Linus Torvalds" line 4) in other words, i've sent here a patch to modify those lines, you said the patch is not ok, then feel free to improve it. and of course you have the right to reject the patch, but please consider the ideas mentioned above before doing so udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On Wed, Oct 04, 2006 at 12:13:02PM +0300, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
I'm not a lawyer, nor I'm strong in interpreting GPL terms
nor me :)
but "one copyright holder - many authors" is common thing in many projects. For example in every source file of SDL (except in those that are just included from other projects) there are the following lines: SDL - Simple DirectMedia Layer Copyright (C) 1997-2006 Sam Lantinga And it doesn't matter that some file didn't exist in 1997, and that most of audio code was (and continues to be) written by Ryan Gordon. There are also other contributors, i.e. maintainers of ports for other platforms. But there is a file named CREDITS, where _all_ contributions and their authors are listed, with many thanks to them. And this is just fine, because in other way there wolud be 10-20 lines of copyright info in many files.
good question. i took my ideas from the kernel source, for example: $ grep -c Linus /usr/src/linux/drivers/net/wireless/ipw2200.c 0 so where Linus is not an author, there he isn't listed at all. the situation is the same with MPlayer: $ grep -c Gereoffy libvo/vo_png.c 0 of course maybe all the kernel & MPlayer devs are crazy, but i don't think so :)
As for generating any copyrights info from CVS - then you forget about those who has no acces to CVS but contributed some patches.
the whole issue is about this: Christian and me have no write access to the cvs and our contributions are pushed by others and if they have a good day then they include our names in the cvs log, if they don't have, then they don't even do so udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
2006/10/4, VMiklos <vmiklos@frugalware.org>:
good question. i took my ideas from the kernel source, for example:
$ grep -c Linus /usr/src/linux/drivers/net/wireless/ipw2200.c 0
so where Linus is not an author, there he isn't listed at all. the situation is the same with MPlayer:
$ grep -c Gereoffy libvo/vo_png.c 0
of course maybe all the kernel & MPlayer devs are crazy, but i don't think so :)
Then it seems that the style of inclusion of copyright info is different in different projects. :-/ ...as always in OSS world. :-)
As for generating any copyrights info from CVS - then you forget about those who has no acces to CVS but contributed some patches.
the whole issue is about this: Christian and me have no write access to the cvs and our contributions are pushed by others and if they have a good day then they include our names in the cvs log, if they don't have, then they don't even do so
Oh, didn't know about that. Come on, guys! Why not just relax and first come to decission about what copyright info style should be in Pacman - like in SDL or like in kernel & MPlayer? In the first case there is standard copyright info for entire project, that repeats in _every_ file. Example: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. In the second case every file has its own copyright info. Example main.c: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. some_functions.c: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. SuperProjectX. SomeFunctions. Copyright (C) 1999-2004, good guy. Some megacool feature support - contriubuted by another good guy. Then if you come to conclusion - there will be no misunderstandings about what to include and in which form. I don't think any of devs are very bothered with copyrights. There just should be standard policy about this. -- Roman Kyrylych (Роман Кирилич)
Roman Kyrylych wrote:
2006/10/4, VMiklos <vmiklos@frugalware.org>:
<snip>
Come on, guys! Why not just relax and first come to decission about what copyright info style should be in Pacman - like in SDL or like in kernel & MPlayer?
In the first case there is standard copyright info for entire project, that repeats in _every_ file. Example: SuperProjectX. Copyright (C) 1997-2006, SomeCompany.
In the second case every file has its own copyright info. Example main.c: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. some_functions.c: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. SuperProjectX. SomeFunctions. Copyright (C) 1999-2004, good guy. Some megacool feature support - contriubuted by another good guy.
Then if you come to conclusion - there will be no misunderstandings about what to include and in which form. I don't think any of devs are very bothered with copyrights. There just should be standard policy about this.
exactly! IMOHO... this is a *very* small disctration. I've been lurking here for a while under a huge rock, waiting for activity on pacman-3, and now when Aaron is ready to get things cranking to a release state, we're going to be side tracked by politics? meh! We're way cooler than this I believe. Like Roman observed, there are _clearly_ two different approaches to things, and while Aurielen and Vmiklos took one approach, Aaron expects another approach. IMOHO, Aaron, since you're right now incharge of the project, pick one Method and lets us live with it and move on to getting pacman out the door, its been waaaay too long in development already i think (though the method you pick does not have to be the one you really like, considering the work vmiklos and co have done, in the end, both methods work, various projects use either). /me silently goes back to lurking under said HugeRock(tm)
On Wed, Oct 04, 2006 at 01:11:50PM +0300, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
Come on, guys! Why not just relax and first come to decission about what copyright info style should be in Pacman - like in SDL or like in kernel & MPlayer?
In the first case there is standard copyright info for entire project, that repeats in _every_ file. Example: SuperProjectX. Copyright (C) 1997-2006, SomeCompany.
In the second case every file has its own copyright info. Example main.c: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. some_functions.c: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. SuperProjectX. SomeFunctions. Copyright (C) 1999-2004, good guy. Some megacool feature support - contriubuted by another good guy.
Then if you come to conclusion - there will be no misunderstandings about what to include and in which form. I don't think any of devs are very bothered with copyrights. There just should be standard policy about this.
if the first case means something like "Copyright (C) 2002-2006, The Pacman Team.", then it sounds ok to me Judd, what is your opinion? :) udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On 10/4/06, Essien Ita Essien <essiene@datavibe.net> wrote:
Roman Kyrylych wrote:
2006/10/4, VMiklos <vmiklos@frugalware.org>:
<snip>
Come on, guys! Why not just relax and first come to decission about what copyright info style should be in Pacman - like in SDL or like in kernel & MPlayer?
In the first case there is standard copyright info for entire project, that repeats in _every_ file. Example: SuperProjectX. Copyright (C) 1997-2006, SomeCompany.
In the second case every file has its own copyright info. Example main.c: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. some_functions.c: SuperProjectX. Copyright (C) 1997-2006, SomeCompany. SuperProjectX. SomeFunctions. Copyright (C) 1999-2004, good guy. Some megacool feature support - contriubuted by another good guy.
Then if you come to conclusion - there will be no misunderstandings about what to include and in which form. I don't think any of devs are very bothered with copyrights. There just should be standard policy about this.
exactly!
IMOHO... this is a *very* small disctration. I've been lurking here for a while under a huge rock, waiting for activity on pacman-3, and now when Aaron is ready to get things cranking to a release state, we're going to be side tracked by politics? meh! We're way cooler than this I believe.
Like Roman observed, there are _clearly_ two different approaches to things, and while Aurielen and Vmiklos took one approach, Aaron expects another approach.
IMOHO, Aaron, since you're right now incharge of the project, pick one Method and lets us live with it and move on to getting pacman out the door, its been waaaay too long in development already i think (though the method you pick does not have to be the one you really like, considering the work vmiklos and co have done, in the end, both methods work, various projects use either).
+1. I could not say anything better, i make our words mine.
/me silently goes back to lurking under said HugeRock(tm)
/me too :o)
_______________________________________________ pacman-dev mailing list pacman-dev@archlinux.org http://www.archlinux.org/mailman/listinfo/pacman-dev
On 10/4/06, Douglas Andrade <dsandrade@gmail.com> wrote:
On 10/4/06, Essien Ita Essien <essiene@datavibe.net> wrote:
IMOHO, Aaron, since you're right now incharge of the project, pick one Method and lets us live with it and move on to getting pacman out the door, its been waaaay too long in development already i think (though the method you pick does not have to be the one you really like, considering the work vmiklos and co have done, in the end, both methods work, various projects use either).
+1. I could not say anything better, i make our words mine.
Well, here's the way I see it. As with anything copyrighted, one can hold the copyright without being a direct author of a given chunk of code. This is deemed a "contribution" of code to the original copyright author. Since 2002 Judd has held, under the GPL, the copyright for pacman, regardless of changes. I believe it's viable for contributing authors to copyright their work as well. I feel that it's far more direct to copyright the entire program to judd by default (yes, from 2002 as it is based on the original pacman). Sure additional copyright info can be added, but judd should remain in place. As for the rationale of people commenting on me bringing this up. Regardless of _why_ you use free software, the protections given by the FSF and the GPL are VITAL. As for why this was brought up when development has been delayed so long - I have received a 34 thousand line diff file, of which a large chunk is copyright changes. I doubt anyone would feel comfortable simpl applying to 34 thousand line patch without thinking to _any_ code, let alone code they're supposed to be responsible for. One can claim it's easy to "just fix it", but honestly, it's just tedious and not inherantly easy. As for the claim that CVS is invalid, I am going to claim that, at this point, it is not. pacman 3 is a "derivative work" based on the original pacman. The copyright from 2002 still holds. I will not be revoking Judd's copyright, as I believe that is improper.
I agree with Aaron that Judd is the copyright holder for the pacman project. Even if you forked the code into another project, the Judd Vinet copyright notice would have to remain atop every file. I would suggest something like a "Portions Copyright" line or "contributing authors" line if you are really concerned with getting credit for your work. No rule states that you cannot have multiple parts of a file under different copyrights, however, this is messy and I wouldn't want to deal with it. My $0.02.. On 10/4/06, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 10/4/06, Douglas Andrade <dsandrade@gmail.com> wrote:
On 10/4/06, Essien Ita Essien <essiene@datavibe.net> wrote:
IMOHO, Aaron, since you're right now incharge of the project, pick one Method and lets us live with it and move on to getting pacman out the door, its been waaaay too long in development already i think (though the method you pick does not have to be the one you really like, considering the work vmiklos and co have done, in the end, both methods work, various projects use either).
+1. I could not say anything better, i make our words mine.
Well, here's the way I see it. As with anything copyrighted, one can hold the copyright without being a direct author of a given chunk of code. This is deemed a "contribution" of code to the original copyright author.
Since 2002 Judd has held, under the GPL, the copyright for pacman, regardless of changes.
I believe it's viable for contributing authors to copyright their work as well. I feel that it's far more direct to copyright the entire program to judd by default (yes, from 2002 as it is based on the original pacman). Sure additional copyright info can be added, but judd should remain in place.
As for the rationale of people commenting on me bringing this up. Regardless of _why_ you use free software, the protections given by the FSF and the GPL are VITAL. As for why this was brought up when development has been delayed so long - I have received a 34 thousand line diff file, of which a large chunk is copyright changes. I doubt anyone would feel comfortable simpl applying to 34 thousand line patch without thinking to _any_ code, let alone code they're supposed to be responsible for. One can claim it's easy to "just fix it", but honestly, it's just tedious and not inherantly easy.
As for the claim that CVS is invalid, I am going to claim that, at this point, it is not. pacman 3 is a "derivative work" based on the original pacman. The copyright from 2002 still holds. I will not be revoking Judd's copyright, as I believe that is improper.
_______________________________________________ pacman-dev mailing list pacman-dev@archlinux.org http://www.archlinux.org/mailman/listinfo/pacman-dev
On Wed, 4 Oct 2006 09:56:41 -0500 "Aaron Griffin" <aaronmgriffin@gmail.com> wrote:
On 10/4/06, Douglas Andrade <dsandrade@gmail.com> wrote:
On 10/4/06, Essien Ita Essien <essiene@datavibe.net> wrote:
IMOHO, Aaron, since you're right now incharge of the project, pick one Method and lets us live with it and move on to getting pacman out the door, its been waaaay too long in development already i think (though the method you pick does not have to be the one you really like, considering the work vmiklos and co have done, in the end, both methods work, various projects use either).
+1. I could not say anything better, i make our words mine.
Well, here's the way I see it. As with anything copyrighted, one can hold the copyright without being a direct author of a given chunk of code. This is deemed a "contribution" of code to the original copyright author.
Since 2002 Judd has held, under the GPL, the copyright for pacman, regardless of changes.
I believe it's viable for contributing authors to copyright their work as well. I feel that it's far more direct to copyright the entire program to judd by default (yes, from 2002 as it is based on the original pacman). Sure additional copyright info can be added, but judd should remain in place.
As for the rationale of people commenting on me bringing this up. Regardless of _why_ you use free software, the protections given by the FSF and the GPL are VITAL. As for why this was brought up when development has been delayed so long - I have received a 34 thousand line diff file, of which a large chunk is copyright changes. I doubt anyone would feel comfortable simpl applying to 34 thousand line patch without thinking to _any_ code, let alone code they're supposed to be responsible for. One can claim it's easy to "just fix it", but honestly, it's just tedious and not inherantly easy.
As for the claim that CVS is invalid, I am going to claim that, at this point, it is not. pacman 3 is a "derivative work" based on the original pacman. The copyright from 2002 still holds. I will not be revoking Judd's copyright, as I believe that is improper.
I'm going to weigh in on this one too. Hopefully I can add a little bit more perspective to try and clear things up. It's not about who thinks who has a copyright, it's about what it says. When Aurelien created the file, he probably refactored/copied a bunch of code from the existing pacman source, so he also included the copyright notice (albeit adjusted to add himself). Judd owned the original copyright and Aurelien added his copyrighted material to also be considered under the GPL umbrella. It is not anyone's place to go in and remove the copyright lines of anyone else unless a) they can prove that that person's copyright was falsely added (ie. that person owns no copyright over anything in the file) or b) they are the person and they know they don't own any copyright to anything in that file. Option b is a really easy one, option a is super hard. The reason that Linus' name isn't in the ipw2200 driver because no one took any of Linus' copyrighted works and used them. Linus himself probably also didn't add any code to that file when it was created. But if someone went in and removed Linus' name from a file that he did own a copyright to, would someone raise a stink? It's not about the final state or who was right or wrong, it's about the meaning of the change. What does it mean to take away someone's copyright claim to something? Jason
On Wed, 4 Oct 2006 11:40:40 +0200 VMiklos <vmiklos@frugalware.org> wrote:
okay, we need facts. what about alpm.c? i _do_ think it was nowhere in 2002
When I see a copyright in a source file that is part of a the entire source code for an autonomous/whole project, I see it as a copyright representing the WHOLE project, not that individual file. Following that logic, if I write a new source file (eg, db.c or whatever) in 2004, I still add the 2002-2004 (C) to it, since the pacman project itself began in 2002. Approaching this idea from another angle... imagine if all of pacman was in a single .c file. Then it would make perfect sense to keep the copyright from 2002, since that's when the file began? Files are just a way of organizing the code -- the project itself is what I'm considering when I write down those dates. I think some projects have a more structured modularity to them that necessitates a more distributed copyright system. I have reservations about moving copyrights over to something like "The Pacman Team", because then the code is officially governed by a body of people, and a body which has no real rules or structure at this time. What if there was a decision to be made that could only be done so by the copyright holder, and "The Pacman Team" was divided on it? Can an ephemeral collection of people legally own a copyright like this without some sort of offical registration of members? I'm assuming the member list of the pacman-dev ML does not constitute the pacman team. But if you look at the ChangeLog for pacman2 (not even including pacman3) there are a lot of fly-by contributors that could feel put out if they didn't have some rights to their contributed code.
2) see the cvs logs, see the AUTHORS file. if that makes you happy, then feel free to remove me, ask Christian and Aurelien about their lines
The AUTHORS (or better yet, CREDITS) file should be used to list all major contributors to a project. That's there so we don't have to amend the copyright headers of all the source files everytime someone makes a contribution. If we amended the copyright headers, then pacman.c would probably have about 30 names at the top of it.
stop. it was me who updated the copyright lines, this issue has nothing with Christian or any other developer
And besides, you talk about "invalid copyrights".... says who? It is completely within legal limits for an author to not be a copyright holder.
Definitely. Some projects require that any patches be submitted with a little email stub that basically says "I give up any rights to this work to the original copyright holder" just to avoid issues like this. The author of the patch is still known, but he/she does not retain copyright rights to it any longer.
-COPYRIGHT_HOLDER = Judd Vinet +COPYRIGHT_HOLDER = Yoyodyne, Inc.
feel free to flame me, but this is already checked in by Judd: $ find . -type f |xargs grep 'COPYRIGHT_HOLDER =' ./lib/libalpm/po/Makevars:COPYRIGHT_HOLDER = Judd Vinet ./src/pacman/po/Makevars:COPYRIGHT_HOLDER = Yoyodyne, Inc.
and of course you can see, that's the default, so the story imho is the following: 1) when krix added the files to our tree, he fogot to change the default copyright holder 2) he sent here the patch 3) Judd corrected the copyright line 4) we never noticed he modified the patch we sent him
conclusion: we never changed "Judd Vinet" to "Yoyodyne, Inc."
This kind of stuff doesn't bother me. I'm pretty sure none of you guys work for Yoyodyne, Inc. :)
yes, that's strange. i could say don't talk about this at all, because it's Judd and my business, definitely not your one
maybe you don't see, but i still would like to be constructive. let's try to turn this endless flame thread to some positive discussion and let's try to find out a solution
Thank you, VMiklos. Constructive is good. Here are my thoughts -- all in all, I would like to retain the copyright to pacman. I worked hard on it for a number of years and I think I deserve it. With that said, people like you, Aaron, Christian, and Aurelien certainly deserve credit for large portions of the code and what it can do today. This is what I want the AUTHORS file to say, and prominently. We can also reference it on the pacman website and in the README file. It is not my intention to hide the fact that I had help with the project, because I did have a lot of help. But if the day comes where a big decisions needs to be made that only the copyright holder can make, I wouldn't want to be bogged down by a game of who-wrote-what -- I think that could tear the project apart.
3) if you think the copyright lines in the cvs are valid, then please have a look at the kernel's source, there is zero "copyright 1991-2006 Linus Torvalds" line
Again, I think the lack of Linus' name on some files is due to the crazy modularity of the kernel. The core kernel itself is still Linus' baby, and as benevolent dictator, he retains the right to make decisions governing it. If he decides to sell the code to Microsoft, then it may be that the ipw2200 module cannot go with it. But the majority of the code is under his control.
4) in other words, i've sent here a patch to modify those lines, you said the patch is not ok, then feel free to improve it. and of course you have the right to reject the patch, but please consider the ideas mentioned above before doing so
Let me take this opportunity to thank you, Christian, and the other developers at Frugalware for your hard work on pacman. While we are often too busy to process your changes, we do appreciate them. I think pacman will grow to be much more powerful now that two strong distributions are using it. Look out rpm and deb/apt-get. :) - J
VMiklos wrote:
On Wed, Oct 04, 2006 at 01:31:53AM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote: that's the situation with Aurelien, not with Judd. i don't know if Judd is the author of alpm.c or not. as you can see i haven't removed his name even originally the whole file may be writted by Aurelien
Just to make things clearer, and share the library history a little bit more : I am the author of all new files (I mean files that can't be found in pacman 2x source archives) at the time the CVS repository was created. I kept Judd's copyright in every existing file, and added it in every new file I created. You won't find my name anywhere in pacman files because this is the way I intended it to be. pacman is one project and I prefer seeing it hold by one copyright, its father's one. Seeing my name in the AUTHOR files is far enough for me. My 2 cents. -- Aurelien
2006/10/4, Judd Vinet <jvinet@zeroflux.org>:
When I see a copyright in a source file that is part of a the entire source code for an autonomous/whole project, I see it as a copyright representing the WHOLE project, not that individual file. Following that logic, if I write a new source file (eg, db.c or whatever) in 2004, I still add the 2002-2004 (C) to it, since the pacman project itself began in 2002.
Approaching this idea from another angle... imagine if all of pacman was in a single .c file. Then it would make perfect sense to keep the copyright from 2002, since that's when the file began? Files are just a way of organizing the code -- the project itself is what I'm considering when I write down those dates.
The AUTHORS (or better yet, CREDITS) file should be used to list all major contributors to a project. That's there so we don't have to amend the copyright headers of all the source files everytime someone makes a contribution. If we amended the copyright headers, then pacman.c would probably have about 30 names at the top of it.
And besides, you talk about "invalid copyrights".... says who? It is completely within legal limits for an author to not be a copyright holder.
Definitely. Some projects require that any patches be submitted with a little email stub that basically says "I give up any rights to this work to the original copyright holder" just to avoid issues like this. The author of the patch is still known, but he/she does not retain copyright rights to it any longer.
Here are my thoughts -- all in all, I would like to retain the copyright to pacman. I worked hard on it for a number of years and I think I deserve it. With that said, people like you, Aaron, Christian, and Aurelien certainly deserve credit for large portions of the code and what it can do today. This is what I want the AUTHORS file to say, and prominently. We can also reference it on the pacman website and in the README file. It is not my intention to hide the fact that I had help with the project, because I did have a lot of help.
But if the day comes where a big decisions needs to be made that only the copyright holder can make, I wouldn't want to be bogged down by a game of who-wrote-what -- I think that could tear the project apart.
Well said, Judd. Authorship and copyright holding are different things. I think most (at least >50%) OSS projects use this aproach to solve copyright issues. -- Roman Kyrylych (Роман Кирилич)
2006/10/4, Aurelien Foret <aurelien@archlinux.org>: Offtopic (?) here: Wow! Aurelien is back! Where have you been so much time? -- Roman Kyrylych (Роман Кирилич)
On Wed, Oct 04, 2006 at 11:12:27AM -0400, Robert Howard <howard3@clemson.edu> wrote:
Even if you forked the code into another project
which is not the situation, as you all know udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On Wed, Oct 04, 2006 at 07:50:36PM +0200, Aurelien Foret <aurelien@archlinux.org> wrote:
new file I created. You won't find my name anywhere in pacman files because this is the way I intended it to be.
here comes the difference. CMIIW, but every author has the right to own the copyright of his code, and to be listed in the headers udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On Wed, Oct 04, 2006 at 10:22:34AM -0700, Judd Vinet <jvinet@zeroflux.org> wrote:
Here are my thoughts -- all in all, I would like to retain the copyright to pacman. I worked hard on it for a number of years and I think I deserve it. With that said, people like you, Aaron, Christian, and Aurelien certainly deserve credit for large portions of the code and what it can do today. This is what I want the AUTHORS file to say, and prominently. We can also reference it on the pacman website and in the README file. It is not my intention to hide the fact that I had help with the project, because I did have a lot of help.
okay, so you would like to have the same header for each file and your copyright from 2002 to 2006. what about the follwong: /* foo.c - This file is part of Pacman, see the COPYRIGHT file for more * information */ then all the names could be listed there
But if the day comes where a big decisions needs to be made that only the copyright holder can make, I wouldn't want to be bogged down by a game of who-wrote-what -- I think that could tear the project apart.
if there are several authors, then the license can be changed to an other one in case all the copyright holders agree about the change, which is almost impossible. currently (yes, of course i really hope this will never happen :) ) if you go crazy and change the license, then you can do it, so the license of my own code is not guarantated. i hope you can see the problem
Again, I think the lack of Linus' name on some files is due to the crazy modularity of the kernel.
hopefully we will have backends other than 'files' and then we'll have crazy modules, too :)
Let me take this opportunity to thank you, Christian, and the other developers at Frugalware for your hard work on pacman.
i'm happy to contribute :) so, i don't plan to revert the copyright updates in our tree, but in case something is changed in the cvs (somehow we're listed as copyright holders) then of course i'll pull that change and drop our copyright fixes. and again, this whole "copyright thingy" from my part is about to be able to control under what license is my code distributed, not about hurting you in any way udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
2006/10/5, VMiklos <vmiklos@frugalware.org>:
On Wed, Oct 04, 2006 at 10:22:34AM -0700, Judd Vinet <jvinet@zeroflux.org> wrote: okay, so you would like to have the same header for each file and your copyright from 2002 to 2006. what about the follwong:
/* foo.c - This file is part of Pacman, see the COPYRIGHT file for more * information */
then all the names could be listed there
I don't think this will be better than one copyright holder. But anyway I don't have the right nor power to impose my mind.
But if the day comes where a big decisions needs to be made that only the copyright holder can make, I wouldn't want to be bogged down by a game of who-wrote-what -- I think that could tear the project apart.
if there are several authors, then the license can be changed to an other one in case all the copyright holders agree about the change, which is almost impossible. currently (yes, of course i really hope this will never happen :) ) if you go crazy and change the license, then you can do it, so the license of my own code is not guarantated. i hope you can see the problem
I don't see the problem. You can always take latest GPL version and make a fork. Or you just don't want that somebody can use your code under different license, not approved by you?
Again, I think the lack of Linus' name on some files is due to the crazy modularity of the kernel.
hopefully we will have backends other than 'files' and then we'll have crazy modules, too :)
Modules with different copyright holder are not a problem because they are not vital part of Pacman. One module can be replaced by another or just rewritten from scratch. I don't think that there will be many modules ever, or they will be too complex to rewrite if needed.
so, i don't plan to revert the copyright updates in our tree, but in case something is changed in the cvs (somehow we're listed as copyright holders) then of course i'll pull that change and drop our copyright fixes. and again, this whole "copyright thingy" from my part is about to be able to control under what license is my code distributed, not about hurting you in any way
Wouldn't it be strange that there will exist two very similar _branches_ of one software but with different copyright notices? ;-) I don't remember such case in OSS history yet. :-D -- Roman Kyrylych (Роман Кирилич)
On Thu, Oct 05, 2006 at 01:50:49AM +0300, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
Or you just don't want that somebody can use your code under different license, not approved by you?
exactly. udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On 10/4/06, VMiklos <vmiklos@frugalware.org> wrote:
On Thu, Oct 05, 2006 at 01:50:49AM +0300, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
Or you just don't want that somebody can use your code under different license, not approved by you?
exactly.
To rephrase this: I, VMiklos, have full power to copy pacman code verbatim and relicense it as I see fit, and claim the original copyrights were incorrect, but no one can use my code without my approval. Sounds a bit bass-ackwards, don't ya think?
On 10/4/06, VMiklos <vmiklos@frugalware.org> wrote:
so, i don't plan to revert the copyright updates in our tree, but in case something is changed in the cvs (somehow we're listed as copyright holders) then of course i'll pull that change and drop our copyright fixes. and again, this whole "copyright thingy" from my part is about to be able to control under what license is my code distributed, not about hurting you in any way
This is illegal. This is not about licenses. This is about copyright, Everything here is under the GPL. Let's review the "human readable" GPL: http://creativecommons.org/licenses/GPL/2.0/ I'll quote a sentence: You must conspicuously and appropriately publish on each copy distributed an appropriate copyright notice and disclaimer of warranty and keep intact all the notices that refer to this License You have not done this, and have now, on a public mailing list, refused to comply with the GPL. Is this an action you sure you want to take?
On Wed, Oct 04, 2006 at 08:29:24PM -0400, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
I, VMiklos, have full power to copy pacman code verbatim and relicense it as I see fit, and claim the original copyrights were incorrect, but no one can use my code without my approval.
Sounds a bit bass-ackwards, don't ya think?
sorry, i think you misunderstood me. of course nobody has the right to relicense a code in case he does not hold the copyrights. i never stated i have such "powers" udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On Wed, Oct 04, 2006 at 07:47:17PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 10/4/06, VMiklos <vmiklos@frugalware.org> wrote:
so, i don't plan to revert the copyright updates in our tree, but in case something is changed in the cvs (somehow we're listed as copyright holders) then of course i'll pull that change and drop our copyright fixes. and again, this whole "copyright thingy" from my part is about to be able to control under what license is my code distributed, not about hurting you in any way
This is illegal. This is not about licenses.
sure, this is not about licenses. but then why do you quote the gpl?
You have not done this, and have now, on a public mailing list, refused to comply with the GPL. Is this an action you sure you want to take?
of course not. i would like to see the copyrights fixed in the cvs, as i stated several times previously. at the moment when you contribute to pacman you don't have to say "the copyright holder of my code will be Judd" and i never said so. possible solutions: 1) let every contributor sign such a document 2) respect other's copyrights and _mention_ them oh, and our pacman tree was always public from the begining (pacman2's cvs tree is still not public), so the copyright update was never a secret (also the darcs logs are on a public mailing list) udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On 10/4/06, VMiklos <vmiklos@frugalware.org> wrote:
On Wed, Oct 04, 2006 at 07:47:17PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
This is illegal. This is not about licenses.
sure, this is not about licenses. but then why do you quote the gpl?
Um, because the GPL guarantees copyright freedoms? It's the GPL that actually protects your copyright.
You have not done this, and have now, on a public mailing list, refused to comply with the GPL. Is this an action you sure you want to take?
of course not. i would like to see the copyrights fixed in the cvs, as i stated several times previously. at the moment when you contribute to pacman you don't have to say "the copyright holder of my code will be Judd" and i never said so.
Ok, there is a misunderstanding here. Your darcs repository is a historical reference of GPL violation. You ARE in violation of the GPL. There's no way around it. Regardless of who's was public first, or other pissing contests, Judd holds the pacman copyright. Copying or even forking the code, under the GPL, requires you to maintain this copyright exactly as is. You have failed this. You are in violation. This is not a light issue. You are, as we speak, violating the GPL. You copied code, and changed the copyright attribution. I can look at the patch you provided, and in under 10 seconds, count 4 files which have 0 changes, but contain your copyright changes - this is ALSO not allowed under the GPL. I will go through and determine exactly which files are patched by you, and apply the authors name to the copyright in CVS. This will waste my time, sure. If you guys would like, you can feel free to waive your rights on existing code in CVS, giving copyright ownership to Judd, though I doubt you'd do that.
From now on, all patches must explicitly waive the author's copyright claim in order to be accepted into pacman. We cannot allow this rampant GPL violation to continue.
On Wed, Oct 04, 2006 at 08:44:26PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
I will go through and determine exactly which files are patched by you, and apply the authors name to the copyright in CVS.
thanks! as soon as you have done this, i'll pull all the headers from cvs and drop our own
This will waste my time, sure. If you guys would like, you can feel free to waive your rights on existing code in CVS, giving copyright ownership to Judd, though I doubt you'd do that.
is it possible to share the copyright ownership? when i first contributed to OpenOffice.org, i had to sign the jca: http://www.openoffice.org/licenses/jca.pdf quoting http://www.openoffice.org/FAQs/faq-licensing.html#jcaa: "Under the Joint Copyright Assignment, both Sun and the contributor retain full rights to use, modify, and redistribute the copyrighted work." this is a bit better: Judd has still the full copyright (as he wished) but if i would like to re-use my own code in some other projects, then i can do using the license i wish. ie.: if i want to contribute our regex-search function to a bsd-licensed project, it won't be a problem
From now on, all patches must explicitly waive the author's copyright claim in order to be accepted into pacman.
okay, thanks for clarifying this. if you think this is agood idea, then mention this in a COPYRIGHT file (or something like that) in the cvs udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On 10/5/06, VMiklos <vmiklos@frugalware.org> wrote:
is it possible to share the copyright ownership? when i first contributed to OpenOffice.org, i had to sign the jca:
http://www.openoffice.org/licenses/jca.pdf
quoting http://www.openoffice.org/FAQs/faq-licensing.html#jcaa:
"Under the Joint Copyright Assignment, both Sun and the contributor retain full rights to use, modify, and redistribute the copyrighted work."
This is interesting. This gives both parties full rights to the source in question, if I'm not mistaken. That is, Judd still has full rights as if he were sole copyright holder, as does a contributor on the code he contributed. Is this a correct interpretation?
Am I correct to assume that contributors only retain rights to their contributed code, not the code of the entire project under which the copyright applies? So, if someone wanted they could roll their code into another product, but not the whole code of pacman? On 10/5/06, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 10/5/06, VMiklos <vmiklos@frugalware.org> wrote:
is it possible to share the copyright ownership? when i first contributed to OpenOffice.org, i had to sign the jca:
http://www.openoffice.org/licenses/jca.pdf
quoting http://www.openoffice.org/FAQs/faq-licensing.html#jcaa:
"Under the Joint Copyright Assignment, both Sun and the contributor retain full rights to use, modify, and redistribute the copyrighted work."
This is interesting. This gives both parties full rights to the source in question, if I'm not mistaken. That is, Judd still has full rights as if he were sole copyright holder, as does a contributor on the code he contributed. Is this a correct interpretation?
_______________________________________________ pacman-dev mailing list pacman-dev@archlinux.org http://www.archlinux.org/mailman/listinfo/pacman-dev
On Thu, Oct 05, 2006 at 09:51:39AM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
This is interesting. This gives both parties full rights to the source in question, if I'm not mistaken. That is, Judd still has full rights as if he were sole copyright holder, as does a contributor on the code he contributed. Is this a correct interpretation?
exactly. so both Judd and myself could be happy :) udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
On Thu, Oct 05, 2006 at 02:47:57PM -0400, Robert Howard <howard3@clemson.edu> wrote:
Am I correct to assume that contributors only retain rights to their contributed code, not the code of the entire project under which the copyright applies? So, if someone wanted they could roll their code into another product, but not the whole code of pacman?
i think you are correct udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
I will go through and determine exactly which files are patched by you, and apply the authors name to the copyright in CVS. This will waste my time, sure. If you guys would like, you can feel free to waive your rights on existing code in CVS, giving copyright ownership to Judd, though I doubt you'd do that.
Hey there. Oops. Please if not so hard, then please apply me too if you see my name at patches :) I'm not a live-for-a-copyright-line-which-contain-my-name man, but if this will reviewed in patches then will good to see my name too in CVS. Thx for that. btw i dont want to talk in this copyright discussion, because i'm not a lawyer and not want to talk anything in this theme. :) Regards Christian Hamar alias krix Hungary
On Thu, 28 Sep 2006 12:09:06 -0700 Judd Vinet <jvinet@zeroflux.org> wrote:
I'll get your back when I can, Aaron. I've merged most of the libarchive stuff in, but it's not committed yet. For some reason, the static linking fails. Dynamic one works fine. I tested this with FW's pacman and I had the same error, so it's likely something on my system, not in the pacman build setup.
If you haven't found the problem here yet it's libtool-slay. Use ABS to rebuild libarchive without removing libarchive.la and pacman3 will build fine :-) I've started to play around with pacman myself and while I'm no expert with C - in fact I'm still learning, if I make any changes that I feel others can benefit from I'll post them here. Nice to see development picking up again! - cs-cam
On 10/7/06, Cameron Daniel <me@camdaniel.com> wrote:
On Thu, 28 Sep 2006 12:09:06 -0700 Judd Vinet <jvinet@zeroflux.org> wrote:
I'll get your back when I can, Aaron. I've merged most of the libarchive stuff in, but it's not committed yet. For some reason, the static linking fails. Dynamic one works fine. I tested this with FW's pacman and I had the same error, so it's likely something on my system, not in the pacman build setup.
If you haven't found the problem here yet it's libtool-slay. Use ABS to rebuild libarchive without removing libarchive.la and pacman3 will build fine :-)
Yeah, I noticed that too - I hate libtool and autotools and all that messy complexity... but what the hell? Maybe later I'll remake the Makefiles by hand (far easier that way IMO), but not now
On Sat, Oct 07, 2006 at 05:13:28PM -0500, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
Yeah, I noticed that too - I hate libtool and autotools and all that messy complexity... but what the hell?
Maybe later I'll remake the Makefiles by hand (far easier that way IMO), but not now
autoconf + pure Makefiles are good :) using autoconf for checking for libarchive, gettext, etc etc is handy. _imho_ :) udv / greetings, VMiklos -- Developer of Frugalware Linux, to make things frugal - http://frugalware.org
participants (12)
-
Aaron Griffin
-
Aurelien Foret
-
Cameron Daniel
-
Christian Hamar
-
Douglas Andrade
-
Essien Ita Essien
-
Jason Chu
-
Judd Vinet
-
kfs1@online.no
-
Robert Howard
-
Roman Kyrylych
-
VMiklos