Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
dynamically integrate GRP with Portage? -easy on the flames
View unanswered posts
View posts from last 24 hours

Goto page 1, 2  Next  
Reply to topic    Gentoo Forums Forum Index Gentoo Chat
View previous topic :: View next topic  
Author Message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Thu Feb 12, 2004 3:50 am    Post subject: dynamically integrate GRP with Portage? -easy on the flames Reply with quote

:wink:

So, I’ve got an idea for portage - or Gentoo in general actually - and I’d like to try and put it out here for general consumption if I could. Please keep an open mind, and the flames to a minimum. If you can think of a good reason why this wouldn't work, or would be a bad idea, that’s cool, please feel free to say something - but as I said, please try to keep an open mind and bear with me while I try to put this idea forth in a manner that makes sense.

It’s not a new idea I don't think. But the idea is to integrate the GRP into portage, actively.

Why couldn't prebuilt, generic binaries be included in the portage tree?

For example

Code:
#emerge Blah

- would emerge Blah just like normal - dl the source compile and install with benefits and all that noise

Code:
#emerge -k Blah

- like before, this would emerge prebuilt binaries if they exist locally on the system, and if they don't, they would default to your basic instance of emerge and do source code, etc...- see above example.

Code:
#emerge -K Blah

- in this idea I have, this is a little different than before - this would emerge prebuilt binaries if they existed locally on the system, and if they didn't exist locally emerge -K would then DL whatever the latest prebuilt GRP binary on the portage tree and install that. Even if it weren’t THE latest version, it would merely use the latest GRP version, whatever it may be.

For example, user punches in
Code:
#emerge -K blah

Portage looks for the GRP package in his system. Nothing. So it looks online. blah-7.3 is the latest version but the latest version portage has compiled as GRP is blah-7.1. So portage downloads and installs the GRP blah-7.1.

You could easily introduce the -u argument to supplement this. If 7.1 was newer than what the user had (in this instance the user has 7.0) then badda bing, off it goes and installs the GRP version of 7.1 - If the latest GRP version is 7.1 but the user has 7.2 installed already, then nothing happens. dig?

There are arguments that say that they'd have to have dozens upon dozens of different versions, but that appears to be because we Gentoo users have difficulty wrapping our heads around the concept that some people don't want to compile anything, have never compiled anything, and if they have any choice in it, won't compile anything in the future. If you don't believe me, I will merely point to the raging success of RPM's. We are all aware that RPM's are total crap, but there is a massive market for RPM's and their sister systems - they have their place in the market whether we like it or not.

Point of fact - Gentoo has a GRP install that has prebuilt generic packages that work fine on almost anyone’s system - yes we are aware that these are not built with any optimizations - yes we are aware that these will not perform quite as well as our custom compiled binaries. This is not the point. the point is there is a market for them, everyone can use them, and they're quick. Again, see RPM's if you have doubts.

There’s rumors that Gentoo is trying to put live CDs out quarterly, which means that there will be quarterly updates to the GRP in one fashion or another. If these GRP packages were to be put into the portage tree, it would merely be a matter of emerging the updated GRP packages from the tree. Easy cheesy.

I believe that by building in flexability like this, Gentoo would see increased usability (and thus popularity) with such a move. and using this model you wouldn't even have to sacrifice any of the functionality or efficiency of the system we use now. The way I have it layed out in my head, this system would operate transparently within it. It would merely make Gentoo more useable for a wider range of people I think by supporting prebuilt binaries (read: GRP).

Now this could stop there, or continue on. I’m going to continue on, just for the sake of argument.

Additional arguments against this idea are how to integrate #emerge sync with such an idea? My thought is you don't. You build in functionality transparently. For example, create something like

Code:
#emerge snapshot.


#emerge snapshot would be a toned down version of #emerge sync. #emerge snapshot would only look at the portage tree in quarters, where emerge sync looks at it daily or whatever it does. for example.
Code:
#emerge -K blah-package

would emerge the latest quarterly GRP version. You could even do
Code:
#emerge -uK blah-package

to only emerge GRP packages as long as they were newer versions of what you had installed. Whatever. You get the idea.

#emerge snapshot would be used by those users who just wanted to keep their system up to date via the GRP. They could log in, punch in
Code:
#emerge snapshot

voila, they'd have the last quarter portage tree.
Code:
#emerge -K gaim

would emerge the latest quarterly GRP version of gaim and install it.

If users decided to compile gaim from source like the rest of us, they'd merely have to punch it in normally as
Code:
#emerge gaim

at which point portage would look at their portage tree, download the appropriate source, and compile/install it like normal. No, their portage tree won't be as up to date as someone doing #emerge sync because they're only looking at it quarterly, but it will still be current as of that quarter- which is pretty okay for most people. And if they really wanted, they could just #emerge sync to update their portage tree and thus get the latest sources.

To continue with the idea,
Code:
#emerge snapshot

would ideally be used with something like
Code:
#emerge nation (or perhaps #emerge -K world).


#emerge nation would behave much like #emerge world in that it would update the whole system but as you'd expect, #emerge nation would only do it with the most recent GRP packages instead of the most recent source code like #emerge world.

(perhaps the K argument would be substituted for G ((for Gentoo)) by this time... get it? #emerge -k blah = #emerge -g blah... #emerge -K world = #emerge -G world).

So to sum up, the idea is to integrate GRP binaries into the portage tree and build flexibility in transparently. Users who wanted to keep their system up to date via the GRP would sit down, punch in
Code:
#emerge snapshot

to get their quarterly portage tree snapshot, and
Code:
#emerge nation

(or emerge -K world) to update their system with the newest GRP packages – (or at least as recent as last quarter). Users would benefit from being able to use the portage system without necessarily having to compile everything on earth to do it. They would benefit from using Gentoo/portage even if they didn't use it at full strength. Fact is, MOST users don't seem to use Linux at full strength. I believe that introducing a system like this and giving users the option to use a slightly toned down version of portage is a seamless method to introduce flexability that will make Gentoo more usable for a wider range of people.

What do you think?

/hides under desk :wink:[/code]
Back to top
View user's profile Send private message
ckdake
l33t
l33t


Joined: 10 Apr 2003
Posts: 889
Location: Atlanta, GA

PostPosted: Thu Feb 12, 2004 4:35 am    Post subject: Reply with quote

Well, I like it. It would be a great way to get more people I know using gentoo that don't like what they hear from me about 3 day installs and day long upgrades when a new xfree/kde needs to be compiled. I would stick with compiling everything, but there are definitely people out there that would benifit from your idea and it wouldn't take away anything from the way things are now for everyone else.
_________________
http://ckdake.com/
Back to top
View user's profile Send private message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Thu Feb 12, 2004 4:48 am    Post subject: Reply with quote

i first posted this in the Gentoo Community on Livejournal - someone there came up with the idea of possibly even having a cron script run in the background during down times to compile things while you weren't using your computer. imagine that. you have a GRP install and as time goes on, your system gets faster8O !!!

Last edited by endtransmission on Fri Feb 13, 2004 8:40 am; edited 1 time in total
Back to top
View user's profile Send private message
steel300
Veteran
Veteran


Joined: 10 Jul 2003
Posts: 1155

PostPosted: Thu Feb 12, 2004 5:16 am    Post subject: Reply with quote

It's not a bad idea at all. The problem is having the storage and machines to build for every specific architecture. A binary portage would be a huge increase in space requirements on the already full Gentoo servers. Plus building every app in binary form would take lots of processing power on various architectures. I don't think that the resources are there yet to accomodate such requirements.
_________________
Rationality is the recognition of the fact that nothing can alter the truth and nothing can take precedence over that act of perceiving it.
Back to top
View user's profile Send private message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Thu Feb 12, 2004 5:33 am    Post subject: Reply with quote

steel300 wrote:
It's not a bad idea at all. The problem is having the storage and machines to build for every specific architecture. A binary portage would be a huge increase in space requirements on the already full Gentoo servers. Plus building every app in binary form would take lots of processing power on various architectures. I don't think that the resources are there yet to accomodate such requirements.


I think if they can put out a Live CD quarterly with complete with a GRP install (which they can and will shortly here), then some of the binaries are already made. It's just a matter or distributing them to the servers. piece of toast.

You're right of course, server storage might be a problem initially. Possible solutions would be to start out small - for instance start out doing only for x86 architectures to begin with. That would mean one build per architecture and that would cover the vast majority of users - that would cut down on the space required. Limiting it to only the GRP or a few of the initial big packages (like open office and kde) initially may also be a way to approach this. to do it incrimentally in the beginningbut you're right, it would be a concern.

Do you think if it proved successful they could expand out into other architectures and even beyond the GRP if they wanted? i think its one of those things - if theres a demand and it works, they'll find the space.

/devils advocate

('preciate the debate - hope you don't mind)


Last edited by endtransmission on Thu Feb 12, 2004 11:09 pm; edited 2 times in total
Back to top
View user's profile Send private message
steel300
Veteran
Veteran


Joined: 10 Jul 2003
Posts: 1155

PostPosted: Thu Feb 12, 2004 5:51 am    Post subject: Reply with quote

The question is, which app's make it to binary form? Why should we limit ourselves to the largest applications. The GRP CD's only have a limited number of applications prebuilt. If you just want to use the GRP package, then I think that it is too limited to gain popularity.
_________________
Rationality is the recognition of the fact that nothing can alter the truth and nothing can take precedence over that act of perceiving it.
Back to top
View user's profile Send private message
mceachrw
n00b
n00b


Joined: 12 Feb 2004
Posts: 3

PostPosted: Thu Feb 12, 2004 5:59 am    Post subject: isn't that what emerge -G and PORTAGE_BIN are for? Reply with quote

isn't that what emerge -G and PORTAGE_BIN are for? i found these current binaries, but can't figure out how to use them.

http://ftp.fredan.org/gentoo/binaries/amd-athlonxp/All/

i added PORTAGE_BIN="http://ftp.fredan.org/gentoo/binaries/amd-athlonxp/All/" to /etc/make.conf, emerge -G kde and got an error i forgot, sorry.

how do i make use of these binaries with portage?
Back to top
View user's profile Send private message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Thu Feb 12, 2004 6:15 am    Post subject: Reply with quote

steel300 wrote:
The question is, which app's make it to binary form? Why should we limit ourselves to the largest applications. The GRP CD's only have a limited number of applications prebuilt. If you just want to use the GRP package, then I think that it is too limited to gain popularity.


Ah... thats right, I forgot that GRP packages are only a portion of the possible available binaries.

Again, i still think it could be accomplished but you're right it would probably take a little more reorganization on the server level than i had anticipated. As far as building the binaries, a combination of distcc and the buildpackages option should make short work of it.

hrm.


Last edited by endtransmission on Fri Feb 13, 2004 1:31 am; edited 1 time in total
Back to top
View user's profile Send private message
Lews_Therin
l33t
l33t


Joined: 03 Oct 2003
Posts: 657
Location: Banned

PostPosted: Thu Feb 12, 2004 7:06 am    Post subject: Reply with quote

It would probably take a lot of space and cpu cycles to build all those packages, plus they would not have been really built for your arch (probably i386 for Pentiums and Athlons). But if we had specific ones for the larger packages (openoffice, KDE, Gnome, etc) this would be great.
Back to top
View user's profile Send private message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Fri Feb 13, 2004 3:37 am    Post subject: A REVISED PROPOSAL Reply with quote

So I forgot when making my proposal that the GRP only represents a portion of the available packages on the portage tree. When i said GRP, i meant all prebuilt binaries. I think everyone who read it seemed to understand what I was talking about, but I thought I'd say something about it.

Anyhow, the main two arguments to this idea appear to be.

1. Server space

2. Binary building

But are either of these show stoppers? It seems both these issues be addressed by implementing a process like this incrimentally, using small steps.

For instance, to start out, perhaps only do x86 architecture. Or only start out doing only GRP packages. Or both. If it works and is popular, the number of packages available on the snapshot system could be increased, and different architecutres included.

Using this model, package building only has to happen quarterly, and the rsync servers only hit quarterly (the only effective difference between emerge sync and emerge snapshot is the frequency in which it updates the portage tree. They wouldn't actually be two different trees. Emerge -K would simply be designed get the most recent version it could, even if it's not the most recent version thats available via source). A combination of distcc and 3 months I think should be more than enough time to find the resources to get things built. I mean if every other distro can build RPM's (or Apts), then we can build GRP binaries once every three months, right?

Hows that sound?

PS How does one present ideas to the maintainers of Gentoo typically? I think I would eventually like to present this to them formally if possible.


Last edited by endtransmission on Fri Feb 13, 2004 8:43 am; edited 1 time in total
Back to top
View user's profile Send private message
tageiru
n00b
n00b


Joined: 26 Oct 2002
Posts: 46

PostPosted: Fri Feb 13, 2004 8:25 am    Post subject: Reply with quote

You will have a huge problem with use flags. There are loads of diffirent options that cant be incorporated into one package. Debian solves this by having many small packages instead of one large. If the same idea is incorporated into gentoo you will have two different types of portage.
Back to top
View user's profile Send private message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Fri Feb 13, 2004 8:36 am    Post subject: Reply with quote

tageiru wrote:
You will have a huge problem with use flags. There are loads of diffirent options that cant be incorporated into one package. Debian solves this by having many small packages instead of one large. If the same idea is incorporated into gentoo you will have two different types of portage.


Not true.

emerge -k already ignores the USE flags. Use flags are only for compiling. As stated, this propsal is dealing with prebuilt binaries. Yes they can and have been made (read: GRP) and yes, they will be made again.

The model I've tried to outline here would not require two versions of portage. In this model, it is seamlessly integrated as an extension of portage, not a replacement or alternate.
Back to top
View user's profile Send private message
bopman
n00b
n00b


Joined: 15 Jan 2003
Posts: 17

PostPosted: Fri Feb 13, 2004 12:16 pm    Post subject: Let's do things, the Gentoo Way... Reply with quote

He has a point, and a very good one. Contrary to what everyone will say, Gentoo is not about compiling source, so, quoting from the Philosophy section in www.gentoo.org:
Quote:
People liked the Portage concept, and Gentoo Linux grew rapidly. We have become known as a "from source" distribution, but the heart of the Gentoo concept is not "from source." "From source" is an important and key aspect of Gentoo, and something that was and will continue to be necessary for Gentoo, but it is not the only issue or most fundamental issue. The most fundamental issue is designing a technology that allows us and others to do what they want to do, without restriction.

Therefore, I think it's not bad at all to look at our neighbours and "learn" something from them. Who are these neighbours you say? As I come from the debian world, let me just put an example: www.apt-get.org
All these packages for Debian are "unofficial", I believe everyone understands perfectly this. But the whole matter is not that, the matter is that Gentoo Linux has the capacity, the technology it takes, to make a similar approach to the end-user (and we as end-users, who consider ourselves "power-users" should not be scared of the scalability of an arquitecture...). We have Portage Overlay, and we normally create our custom ebuilds, so why can't we also share our binary packages?
That way, I'll just have to add an ebuild from one site to my portageoverlay path and emerge it, a binary package.
So, the two problems you all say we face:
1. Server Space.
2. Binary building.
These will have been solved by us the users' with enough disk-space / bandwitdth to share packages, (let me call that, the community instead just users...)
I compile packages everytime I need to, so if my packages are perfectly good for me, why can't they be good for another person? Most of you will be thinking already about your USE flags being different from mine, and your Optimisation, be it 686, athlonxp, or pentium4... but I don't see a problem there. I compile for athlonxp, but maybe you compile for 686, as well as power-joe user has his own packages for pentium4... etc. Now, the USE flags... maybe this is where the rest will say it won't fit, but well, take a look at www.apt-get.org: different optimisations, and each package compiled with one thing or the other, be it xinerama support, be it with gcc3.whatever_version_you_like, be it with artsd support, alsa, esd, gtk+, hell!... there's almost every type of package for your choice!
And if you don't find what suits your needs, then you can just compile from source, as we have been doing all our lives, the official way right?
So, to end this boring post of mine, let me quote another paragraph from the wonderful Gentoo Philosophy page, which I recommend everyone to read:
Quote:
The Gentoo philosophy, in a paragraph, is this. Every user has work they need to do. The goal of Gentoo is to design tools and systems that allow a user to do their work pleasantly and efficiently as possible, as they see fit. Our tools should be a joy to use, and should help the user to appreciate the richness of the Linux and free software community, and the flexibility of free software. This is only possible when the tool is designed to reflect and transmit the will of the user, and leave the possibilities open as to the final form of the raw materials (the source code.) If the tool forces the user to do things a particular way, then the tool is working against, rather than for, the user. We have all experienced situations where tools seem to be imposing their respective wills on us. This is backwards, and contrary to the Gentoo philosophy.
Back to top
View user's profile Send private message
guy
Apprentice
Apprentice


Joined: 31 Mar 2003
Posts: 286
Location: USA

PostPosted: Fri Feb 13, 2004 2:49 pm    Post subject: Reply with quote

Damn I'm getting sick of these talks. Not that I don't like the idea-- I love it. But it never goes anywhere. The people in charge of things like portage don't seem to care for it much.

But I'll through in my thoughts anyways: make.conf could have 2 more fields called "STRICT_USE" and "STRICT_CFLAGS" Both are boolean and mean what they sound like: if STRICT_USE is true, than ONLY use packages that match the use flags exactly. Otherwise, get whatever binary is available. Of course this means binary packages need to include some meta data if they don't (I asked about this a while back, but never got an answer)

If STRICT_CFLAGS is false, then portage has to make sure the architectures are at least compatible. (athlon-xp -> i386, but not the other way around)

The tricky part is how to store multiple binaries, WHERE to store them, and how to get them compiled in a trusted way.

Maybe for how to compile them, we could look to the BOINC project of SETI@home.
_________________
www.manatee.servemp3.com
www.seas.upenn.edu/~bdodson/gina
www.goatse.cx [it will rise again!]
Back to top
View user's profile Send private message
nbrown
Tux's lil' helper
Tux's lil' helper


Joined: 26 Oct 2002
Posts: 140
Location: Somewhere in the NE

PostPosted: Fri Feb 13, 2004 3:39 pm    Post subject: Reply with quote

You might find this interesting.

-nbrown
Back to top
View user's profile Send private message
bopman
n00b
n00b


Joined: 15 Jan 2003
Posts: 17

PostPosted: Fri Feb 13, 2004 4:02 pm    Post subject: Reply with quote

Project Chistrap sounds good, but it remains to see how it will continue. The only problem I see is I would have to stick to their USE flags, their optimisations... etc. I'd prefer to have more choice, more freedom... but well, it's better than nothing, I'm gonna try it out. :twisted:
Back to top
View user's profile Send private message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Sat Feb 14, 2004 1:12 am    Post subject: Re: Let's do things, the Gentoo Way... Reply with quote

bopman wrote:
He has a point, and a very good one. Contrary to what everyone will say, Gentoo is not about compiling source, so, quoting from the Philosophy section in www.gentoo.org:
Quote:
People liked the Portage concept, and Gentoo Linux grew rapidly. We have become known as a "from source" distribution, but the heart of the Gentoo concept is not "from source." "From source" is an important and key aspect of Gentoo, and something that was and will continue to be necessary for Gentoo, but it is not the only issue or most fundamental issue. The most fundamental issue is designing a technology that allows us and others to do what they want to do, without restriction.

Therefore, I think it's not bad at all to look at our neighbours and "learn" something from them. Who are these neighbours you say? As I come from the debian world, let me just put an example: www.apt-get.org
All these packages for Debian are "unofficial", I believe everyone understands perfectly this. But the whole matter is not that, the matter is that Gentoo Linux has the capacity, the technology it takes, to make a similar approach to the end-user (and we as end-users, who consider ourselves "power-users" should not be scared of the scalability of an arquitecture...). We have Portage Overlay, and we normally create our custom ebuilds, so why can't we also share our binary packages?
That way, I'll just have to add an ebuild from one site to my portageoverlay path and emerge it, a binary package.
So, the two problems you all say we face:
1. Server Space.
2. Binary building.
These will have been solved by us the users' with enough disk-space / bandwitdth to share packages, (let me call that, the community instead just users...)
I compile packages everytime I need to, so if my packages are perfectly good for me, why can't they be good for another person? Most of you will be thinking already about your USE flags being different from mine, and your Optimisation, be it 686, athlonxp, or pentium4... but I don't see a problem there. I compile for athlonxp, but maybe you compile for 686, as well as power-joe user has his own packages for pentium4... etc. Now, the USE flags... maybe this is where the rest will say it won't fit, but well, take a look at www.apt-get.org: different optimisations, and each package compiled with one thing or the other, be it xinerama support, be it with gcc3.whatever_version_you_like, be it with artsd support, alsa, esd, gtk+, hell!... there's almost every type of package for your choice!
And if you don't find what suits your needs, then you can just compile from source, as we have been doing all our lives, the official way right?
So, to end this boring post of mine, let me quote another paragraph from the wonderful Gentoo Philosophy page, which I recommend everyone to read:
Quote:
The Gentoo philosophy, in a paragraph, is this. Every user has work they need to do. The goal of Gentoo is to design tools and systems that allow a user to do their work pleasantly and efficiently as possible, as they see fit. Our tools should be a joy to use, and should help the user to appreciate the richness of the Linux and free software community, and the flexibility of free software. This is only possible when the tool is designed to reflect and transmit the will of the user, and leave the possibilities open as to the final form of the raw materials (the source code.) If the tool forces the user to do things a particular way, then the tool is working against, rather than for, the user. We have all experienced situations where tools seem to be imposing their respective wills on us. This is backwards, and contrary to the Gentoo philosophy.


Well I'm glad that this is an idea that you like. Curiously though, if gentoo users are sharing optimized compiled binaries, wouldn't the logical extension of that be to integrate a kind of P2P functionality to look for optimized packages built by other people with the same USE flags/architectures, and then DL from them directly? The search and DL could happen automatically. Does that make sense? A portage with added P2P functionality dedicated to finding and locating binaries and then community members downloading from each other - the more people that have it the faster the updates spread... what a thought! And if nobody had it yet, you'd compile it from source and then it'd be available for others to DL.

Boy that would shake some stuff up wouldn't it. I mean in a purist sense of the term, thats the kind of application that P2P was originally developed for - not for the trade in proprietary work, but to share/trade real data, open source stuff. Wanna set the Gentoo community apart? Integrate something like that into portage and we'll find out what true community really is.

If it could be done securely with confidence, i think it would be revolutionary.



crazy. :D


Last edited by endtransmission on Wed Feb 18, 2004 2:17 am; edited 1 time in total
Back to top
View user's profile Send private message
bopman
n00b
n00b


Joined: 15 Jan 2003
Posts: 17

PostPosted: Sat Feb 14, 2004 4:26 pm    Post subject: Dude. great idea Reply with quote

Yeah it's a great idea ^^
We're all having ideas though, but we're going nowhere if we don't put action into our words... starting from me, I'm the number 1 lazy bastrd around :roll:
guy has said it, these talks never get anywhere... I think I'm gonna put on some nerve to read through the Developer's section and see if I can do some action 8)
Back to top
View user's profile Send private message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Tue Feb 17, 2004 1:12 am    Post subject: Re: Dude. great idea Reply with quote

bopman wrote:
Yeah it's a great idea ^^
We're all having ideas though, but we're going nowhere if we don't put action into our words... starting from me, I'm the number 1 lazy bastrd around :roll:
guy has said it, these talks never get anywhere... I think I'm gonna put on some nerve to read through the Developer's section and see if I can do some action 8)


I think thats a great idea. As for me, after discussing this with everyone here I think I will look into the possibility of developing a standalone P2P client specifically for gentoo users that will search for compiled binaries on other community members computers, using qualifiers like USE flags to narrow searches. Without adding to the bulk of Portage, I think this might be the best way to approach and develop the concept of distributing precompiled binaries, and still keep it simple.

Also doing it this way, using USE flags as search qualifiers allows for the possible distribution of optimized binaries as well. The downside to this is that by using this method one might have to track down all the individual dependencies - so it would probably only have practical application with some of the bigger files like gcc, glibc, XFree, etc... Once Downloaded to your portage tree, it's merely a matter of compiling the dependencies...

I wonder if theres already metadata in binaries that supports this - or could be altered to support this - that would be a good way of failsafe error checking... if every compiled binary had a listing of what use flags were used to compile it...

I don't know. Seems like theres merit in the idea somewhere though.
/the gerbles kick into high gear...
Back to top
View user's profile Send private message
spartacus
n00b
n00b


Joined: 26 May 2003
Posts: 4
Location: Germany

PostPosted: Tue Feb 17, 2004 7:43 am    Post subject: Reply with quote

Ohh man!!! You should be careful with the security aspects. The binary packages will be shared and installed. There are certainly people out there which want to destroy some good going things.
The install servers of gentoo are read-only and in that way I am sure that the packages I download are not harmful but how can you download packages from others' computers.
A security and reliability concept should also be implemented if you think of such kind of a job.. Perhaps md5 generation for each package..
Back to top
View user's profile Send private message
endtransmission
n00b
n00b


Joined: 16 Nov 2003
Posts: 48

PostPosted: Tue Feb 17, 2004 7:55 am    Post subject: Reply with quote

spartacus wrote:
Ohh man!!! You should be careful with the security aspects. The binary packages will be shared and installed. There are certainly people out there which want to destroy some good going things.
The install servers of gentoo are read-only and in that way I am sure that the packages I download are not harmful but how can you download packages from others' computers.
A security and reliability concept should also be implemented if you think of such kind of a job.. Perhaps md5 generation for each package..


Well it wouldn't arbitrarily search anyones computer. Like other P2P clients, other members of the community would have to be running a client too, so presumably you'd only be searching computers that had this P2P functionality included by the administrator - in order to DL a file the P2P client/server/service would have to be started by someone or set to start as an RC option QED the administrator would already be aware that it was running - and in fact would have to set it (and presumably leave it) running - like something similar to a distcc daemon - something small and in the background.

I don't know at this stage of things. If I were to go ahead with the idea obviously community participation would be a critical component.

Thanks for the feedback.
Back to top
View user's profile Send private message
slonocode
Apprentice
Apprentice


Joined: 03 Jun 2002
Posts: 273

PostPosted: Thu Feb 19, 2004 9:24 am    Post subject: Reply with quote

Quote:
Well it wouldn't arbitrarily search anyones computer. Like other P2P clients, other members of the community would have to be running a client too, so presumably you'd only be searching computers that had this P2P functionality included by the administrator - in order to DL a file the P2P client/server/service would have to be started by someone or set to start as an RC option QED the administrator would already be aware that it was running - and in fact would have to set it (and presumably leave it) running - like something similar to a distcc daemon - something small and in the background.


I think the fear is that someone could share a "tainted" binary and how would you make sure that can't happen.


Otherwise I like your idea.
Back to top
View user's profile Send private message
toMeloos
Tux's lil' helper
Tux's lil' helper


Joined: 24 Jul 2003
Posts: 101
Location: The Netherlands

PostPosted: Thu Feb 19, 2004 11:46 am    Post subject: Reply with quote

Great plan, I am all for it.

Just to look at the impact this move might have:
Gentoo is a distribution that has shown that their package management works. Combined with the level of control Gentoo offers it is a fairly good combination. One weakness it has is the lack of a binary version of the package management. Many users, like myself, don't use Gentoo on a number of machines because of that reason. If a binary 'tree' can be created within portage this will make gentoo available to many more computers. The most important group will be the servers. Most Gentoo users do not put it on servers because of the compiling load. This will most definately help Gentoo grow and attract even more users. In other words: great! let's do this!

I read some people were looking into how to start this project. I hope they will provide more info here when it becomes available (project website, mailing list, etc.) so more people can help to design the best way for this binary thing to work.
Back to top
View user's profile Send private message
nils_a
n00b
n00b


Joined: 20 Apr 2004
Posts: 16

PostPosted: Tue Apr 20, 2004 10:54 pm    Post subject: Reply with quote

Great Idea. Using p2p for what it was designed...

Personally I do not (jet) have any need for a "binay" Portage -- I'm away from RPM for four days now :-)

But I'd love to share my Bandwidth doing some good for the community!

Quote:
I think the fear is that someone could share a "tainted" binary...

Indeed... security is needed here.

Cheers!
Back to top
View user's profile Send private message
Genone
Retired Dev
Retired Dev


Joined: 14 Mar 2003
Posts: 9237
Location: beyond the rim

PostPosted: Wed Apr 21, 2004 3:23 pm    Post subject: Reply with quote

For the p2p idea:
https://bugs.gentoo.org/show_bug.cgi?id=23747
https://bugs.gentoo.org/show_bug.cgi?id=8468
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Gentoo Chat All times are GMT
Goto page 1, 2  Next
Page 1 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum