| View previous topic :: View next topic |
| Author |
Message |
twstd3bc
Apprentice
Joined: 07 Feb 2003
Posts: 289
Location: Los Angeles, USA
|
 Posted: Sun Aug 27, 2006 9:20 pm Post subject: |
 |
|
| legine wrote: | twstd3bc
From my Experience you always have time when you want to.
|
I don't mean to sound rude, but your are soooo missing the point. I am not less capable than anyone on this forum in keeping my system error-free and up-to-date. I am not ignorant of the useful methods needed to do so. Your reccomendations have already been suggested in this thread. They are not relevant to my claims. Apparently I haven't made the problem clear, for there are lots of people on this forum who just don't get it. Claiming "but I don't have any problems" doesn't do much for anyone. I'm not trying to dis Gentoo, dude. I have a server that's been running non stop since January 1 with no problems. You're missing the point.
People who write portage files don't have the same vision, which causes problems. For example, if I have a package A (version 1.3) which depends on packages B version 2.45) and C (version 1.87), and I write an ebuild for a new version of package A version 1.4, I don't necessarily need to require updated versions of packages B and C. Some people do, some people don't. With the complexity of portage, this problem compounded with others (I'm not mentioning), causes unnecessary headaches. This is hard for some people to believe, and it is a waste of time to diagnose a whimsical design philosophy. Now with the size of portage, the problem is out of hand. That is why the original poster thinks Gentoo will die-- because the problem keeps growing, and nobody even realizes it. |
|
| Back to top |
|
 |
iKiddo
Guru
Joined: 27 Jun 2002
Posts: 341
Location: Europe?
|
| Posted: Sun Aug 27, 2006 10:44 pm Post subject: |
|
|
| twstd3bc wrote: | | People who write portage files don't have the same vision, which causes problems. For example, if I have a package A (version 1.3) which depends on packages B version 2.45) and C (version 1.87), and I write an ebuild for a new version of package A version 1.4, I don't necessarily need to require updated versions of packages B and C. Some people do, some people don't. With the complexity of portage, this problem compounded with others (I'm not mentioning), causes unnecessary headaches. This is hard for some people to believe, and it is a waste of time to diagnose a whimsical design philosophy. Now with the size of portage, the problem is out of hand. That is why the original poster thinks Gentoo will die-- because the problem keeps growing, and nobody even realizes it. |
I, too, am one of the older bunch. I'm from back in the days stage1 was the only supported installation, and RPM-hell was not fixed yet. (Yes, it is now!) And I too have the feeling Gentoo has become unstable and bloated.
Over time the standards for Linux distributions have raisen steeply. Previously the emerge package management system was the top packagemanager available, but has been overtaken (as a Dutchman once proverbially said: being ahead will slow you down, "de wet van de remmende voorsprong" - Jan Romein). And as has been noted in this thread, it needs an overhaul to be competetive again. Some problem areas:
- When I started using Gentoo it was a relative pain to mix ~x86 and x86, now that it has become easier (with a.o. the introduction of the /etc/portage/ config files) it has also become required to mix ~x86, although I have the feeling my needs haven't changed. This has to be cleaned up. (The above quote is a more technical explanation of my point.)
- The same goes for USE flags.
- The forums have to be read to keep stuff from breaking. We need a mechanism to inform the user at the CLI of upgrade guides in wikis or at gentoo.org.
- This wave of people that started using Gentoo a while ago, can be due to the effect of humans generally being more capable of remembering the good aspects, than the bad. (The "good ol' days"-effect.) But my computer (hardware) hasn't been updated since I bought it (as it is a laptop), and I'm definetely sure that portage has become a lot slower! |
|
| Back to top |
|
|
Dralnu
Veteran
Joined: 24 May 2006
Posts: 1919
|
| Posted: Sun Aug 27, 2006 11:30 pm Post subject: |
|
|
| iKiddo wrote: | | twstd3bc wrote: | | People who write portage files don't have the same vision, which causes problems. For example, if I have a package A (version 1.3) which depends on packages B version 2.45) and C (version 1.87), and I write an ebuild for a new version of package A version 1.4, I don't necessarily need to require updated versions of packages B and C. Some people do, some people don't. With the complexity of portage, this problem compounded with others (I'm not mentioning), causes unnecessary headaches. This is hard for some people to believe, and it is a waste of time to diagnose a whimsical design philosophy. Now with the size of portage, the problem is out of hand. That is why the original poster thinks Gentoo will die-- because the problem keeps growing, and nobody even realizes it. |
I, too, am one of the older bunch. I'm from back in the days stage1 was the only supported installation, and RPM-hell was not fixed yet. (Yes, it is now!) And I too have the feeling Gentoo has become unstable and bloated.
Over time the standards for Linux distributions have raisen steeply. Previously the emerge package management system was the top packagemanager available, but has been overtaken (as a Dutchman once proverbially said: being ahead will slow you down, "de wet van de remmende voorsprong" - Jan Romein). And as has been noted in this thread, it needs an overhaul to be competetive again. Some problem areas:
- When I started using Gentoo it was a relative pain to mix ~x86 and x86, now that it has become easier (with a.o. the introduction of the /etc/portage/ config files) it has also become required to mix ~x86, although I have the feeling my needs haven't changed. This has to be cleaned up. (The above quote is a more technical explanation of my point.)
- The same goes for USE flags.
- [b]The forums have to be read to keep stuff from breaking. We need a mechanism to inform the user at the CLI of upgrade guides in wikis or at gentoo.org.[/b[
- This wave of people that started using Gentoo a while ago, can be due to the effect of humans generally being more capable of remembering the good aspects, than the bad. (The "good ol' days"-effect.) But my computer (hardware) hasn't been updated since I bought it (as it is a laptop), and I'm definetely sure that portage has become a lot slower! |
I've relativly new to Gentoo, and from my limited experiance, I do agree with some of these, although I fail to see where the bloat is. If you speak of un-needed config files, and non-modular packages like Gnome and OOo, then I can see it.
There is a seperate thread dealing with the USE flag issue, but I'm not sure I see your complaint. You talking about it being needlessly complex, or are you speaking of the number of them?
The emboldened comment I'm all for. That would be a very handy tidbit of info to have at times, although I think the use of links/lynx/elinks can partly fix that
The rewrite/overhaul of Portage has been covered before. I think one of the major issues with that is that currently, Portage doesn't use a database by default (I think you can covert it to use one, but I'm unsure of that), and another is the fact you're dealing with a relativly large scripted program. While granted they are just as fast as compiled, converting Portage to C/C++ might oculd be of help, if alot of the second-hand programs (eix, equery, ect.) where added in by default instead of the generally slow/crappy tools that are in Portage by default, and waste space.
Also, doing that you could also remove the python reliance from Portage, and reduce system size some (if you don't script/don't script in python).
Any of this make sense to anyone else?
_________________
The day Microsoft makes a product that doesn't suck, is the day they make a vacuum cleaner. |
|
| Back to top |
|
|
legine
Guru
Joined: 27 May 2004
Posts: 555
Location: Germany
|
| Posted: Mon Aug 28, 2006 7:34 am Post subject: |
|
|
| Quote: | | I don't mean to sound rude, but your are soooo missing the point. |
You are not. 
| Quote: | | People who write portage files don't have the same vision, which causes problems. For example, if I have a package A (version 1.3) which depends on packages B version 2.45) and C (version 1.87), and I write an ebuild for a new version of package A version 1.4, I don't necessarily need to require updated versions of packages B and C. Some people do, some people don't. With the complexity of portage, this problem compounded with others (I'm not mentioning), causes unnecessary headaches. |
So what you are saying is:
We should build ebuilds in a way that the lowest Possible Version should be able to merge?
We should upkeep a reverse bug support just because someone does not believe that an update is necessary?
Uhh I think I really do not get your Point. I see no sense to this.
| Quote: | | - When I started using Gentoo it was a relative pain to mix ~x86 and x86, now that it has become easier (with a.o. the introduction of the /etc/portage/ config files) it has also become required to mix ~x86, although I have the feeling my needs haven't changed. This has to be cleaned up. (The above quote is a more technical explanation of my point.) |
Hmm yea there are some issues with Stable and unstable tree. But I think the need of mixing and not mixing is unchanged.
_________________
quote from Spaceballs:
Dark Helmet:[...] we were told to comb the desert, so we're combing it! [puts down bullhorn] Find anything yet?!
Soldier: Nothing yet, sir. |
|
| Back to top |
|
|
jballou
Tux's lil' helper
Joined: 10 Mar 2005
Posts: 128
Location: Baghdad, Iraq
|
| Posted: Mon Aug 28, 2006 8:38 am Post subject: Re: why Gentoo sucks, and why it will ultimately die |
|
|
| Zork the Almighty wrote: | [/url]Two words: maintenance and bloat. When I started using Gentoo a number of years ago, it was a small, lightweight distribution with a reputation for speed. Despite the fact that most software had to be compiled, you could build a very lean system simply by installing a handful of packages and tweaking the USE flags to leave other stuff out. It took 5 seconds to update the portage cache. Packages were frequently updated, but for the most part you could just install the new version overtop of the old one. My standard system consists of X, KDE, Mplayer, and a few other packages like ssh and gcc which every system should have.
|
So you're saying you want less packages available? You could always use the rsync exclude option to prune your tree and voila. The distro is made for choice, and if we limit the software available, that's less choices. You may choose to prune your tree or just not install the packages. I rarely have problems with updates breaking anything, but that might just be that I've been running Gentoo as my primary OS for about 2 years now.
| Quote: | | The bottom line is that I go out of town for one month, and when I return I spend one or two full days updating and repairing my system. It's not just compile time - I may as well reinstall. In fact, if you don't update you will be forced to reinstall because your configuration will quickly become unsupported, or updating it will break it beyond all repair. |
This I agree with, but I've had systems that I have left for 6 months and went back, synced, updated, and rebooted a fresh kernel and no problems. There are a lot of updates, but Gentoo has a much faster release tempo than other distros. I partly agree that they should support older 'stable' versions a little longer, but with how many devs we have compared to how many packages, I don't see it happening. It's not like we're paying these guys or anything
| Quote: |
There are so many packages now that it becomes impossible to strip out bloat. When one piece of software requires 100 packages, someone in that list will always want to install Gnome or Ruby or GTK+ or some other unnecessary dependency. Bloat is also added whenever the underlying system is changed, because new dependencies are added and old ones are not removed. I know there are tools for detecting this, but I also know that they don't often work. I'm pretty sure that if I reinstalled Gentoo right now I could build a lean and modular system, but a year later it would be spaghetti again.
|
With USE flags I have avoided big dependencies I didn't want. I've never had an issue with something big coming up in the list that I couldnt track down the tree and a quick package.use entry couldn't fix.
| Quote: | The bottom line is that Gentoo reflects the state of Linux as a whole. It's like a house where every part, from the foundation to the roof, is constantly being worked on or replaced. This is good for house design, but it's not a place where anybody can live. Most distributions cherry pick the best versions of software for a "release", and then actually test the software (in a "beta" version) to see if it works. This way the users get the benefits of a fairly modern house, with only periodic interuptions instead of continuous construction. I don't think anything in Gentoo ever even gets to beta level anymore. The distribution as a whole certainly doesn't. The only people who can seem to maintain the system are its developers, and trying to "use" Gentoo is the craziest thing I have ever seen a large group of people do.
I miss the simplicity and speed of the old Gentoo. The distribution is now 10 times more complex, but it doesn't seem to deliver any benefits to its users. |
Yes, it's in a constant state of upkeep. That's sort of the price of admission. If you wanted a stable (and by stable I mean infrequent updates mainly for bugfixes) you could always run Debian, or any of the other binary distros which are nice too. Gentoo requires more work on the user's part, but I enjoy it. Portage takes most of the heavy lifting out of keeping the packages updated and installed, but Gentoo is arguably one of the most educational distros, even if it's just to learn you don't want to run it. I've had Debian, Fedora Core, Slack, and Suse systems, and I always come back to Gentoo. For me, the benefits far outweigh the pitfalls. If it's not for you though, good luck with whatever you end up doing. Just don't go formatting and installing Vista or anything crazy like that.
_________________
-Shuttle SN25P, Opteron 185, 74Gb WD Raptor, 2x 300Gb Barracuda 7200.9, 2x 1Gb Corsair dual channel, BFG GeForce 7800GT
-Asus Z7100 laptop, P-M 2.13, 2Gb DDR, 100Gb 7200RPM HDD, 128Mb GeForce 6600 Go. |
|
| Back to top |
|
|
krigav
Tux's lil' helper
Joined: 10 Nov 2005
Posts: 121
|
| Posted: Mon Aug 28, 2006 9:23 am Post subject: Re: why Gentoo sucks, and why it will ultimately die |
|
|
| Zork the Almighty wrote: | | The bottom line is that I go out of town for one month, and when I return I spend one or two full days updating and repairing my system. It's not just compile time - I may as well reinstall. In fact, if you don't update you will be forced to reinstall because your configuration will quickly become unsupported, or updating it will break it beyond all repair. |
WTF? Could you please explain me HOW a well running system could be broken when you leave it in a functional state. And when you return it's broken???
Maybe you should keep you children away from your gentoo box while you away?
_________________
There are no dumb questions, just dumb answers. So please help users that are new to linux/gentoo by answering unanswered questions. |
|
| Back to top |
|
|
legine
Guru
Joined: 27 May 2004
Posts: 555
Location: Germany
|
| Posted: Mon Aug 28, 2006 9:31 am Post subject: |
|
|
I think he refers to updateing after a month is pure pain since the update is breaking his well working system.
Which is from my point of view just a question of workflow.
_________________
quote from Spaceballs:
Dark Helmet:[...] we were told to comb the desert, so we're combing it! [puts down bullhorn] Find anything yet?!
Soldier: Nothing yet, sir. |
|
| Back to top |
|
|
ArturHawkwing
n00b
Joined: 20 Oct 2004
Posts: 22
Location: MN, USA
|
| Posted: Mon Aug 28, 2006 10:37 am Post subject: |
|
|
[quote="Dralnu"]
Also, doing that you could also remove the python reliance from Portage, and reduce system size some (if you don't script/don't script in python).
Any of this make sense to anyone else?/quote]
yup it makes perfect sense. Especially since if I remember correctly that the ebuilds are written in bash not python.
| Quote: |
So what you are saying is:
We should build ebuilds in a way that the lowest Possible Version should be able to merge?
We should upkeep a reverse bug support just because someone does not believe that an update is necessary?
Uhh I think I really do not get your Point. I see no sense to this.
|
One could keep a copy of the things they don't want upgraded in a local portage directory like /usr/local/portage am I correct? Of course this wouldn't garentee the compatibility with the newer ebuilds of course. |
|
| Back to top |
|
|
legine
Guru
Joined: 27 May 2004
Posts: 555
Location: Germany
|
| Posted: Mon Aug 28, 2006 12:52 pm Post subject: |
|
|
| Quote: | yup it makes perfect sense. Especially since if I remember correctly that the ebuilds are written in bash not python.
|
Portage is more then ebuids. And I remebering it right, it is a mix between python and bash.
The question is is it time for a rewrite? Maybe avoiding some design failures.
There was a Portage based on C++. We also had a portage system that was featuring databases (I think it could store data within a mysqlserver.) But it never got accepted...
(search for lost apps or tools or something. If I find time I search for it...)
The stuff with equery and thing is not with portage since they have different development cycles. So it is wise to keep them seperated.
_________________
quote from Spaceballs:
Dark Helmet:[...] we were told to comb the desert, so we're combing it! [puts down bullhorn] Find anything yet?!
Soldier: Nothing yet, sir. |
|
| Back to top |
|
|
iKiddo
Guru
Joined: 27 Jun 2002
Posts: 341
Location: Europe?
|
| Posted: Mon Aug 28, 2006 1:44 pm Post subject: |
|
|
| Dralnu wrote: | I've relativly new to Gentoo, and from my limited experiance, I do agree with some of these, although I fail to see where the bloat is. If you speak of un-needed config files, and non-modular packages like Gnome and OOo, then I can see it.
There is a seperate thread dealing with the USE flag issue, but I'm not sure I see your complaint. You talking about it being needlessly complex, or are you speaking of the number of them? |
Gentoo installations now need a lot more configuring than they used to need. This may be perceived as a form of bloat. I'm going to illustrate the point using USE flags, but the following bit concerns Gentoo as a whole:
Quite a bit of the added configuring is because nearly all compile-time options have been transformed into (package specific) USE flags. Back when I installed Gentoo for the first time, there were some 20 USE flags. And that was it. At the time of writing there are 357(!) USE flags in use.desc alone!!
This change can be interpreted two ways. At the moment the majority of developers seem to see it as a positive change, as it enables more choice. ("And isn't choice what Gentoo is all about?") But an increasing number of users who have experienced the older days notice that this has introduced an increased amount of effort necessary to use Gentoo, and see it as a negative change.
| Dralnu wrote: | | The emboldened comment I'm all for. That would be a very handy tidbit of info to have at times, although I think the use of links/lynx/elinks can partly fix that |
The whole point of adding it directly to the emerge interface is that users who do not frequently browse the forums, the mailing lists, or gentoo.org in general (and the GWN in particular), such as I am, are not made aware of things they ought to know to prepare them for certain events like for instance the change to udev. This is illustrated by one of the earlier posts in this thread, stating it wasn't possible to see the udev upgrade was a special one, from the output emerge gave.
Adding this notification feature is absolutely necessary in my opinion as it drastically improves the "Gentoo experience" (this hereby directly improves Gentoo), and also decreases the number of unnecessary help-threads in the Gentoo forums. I think both this to-be-added feature and the glsa check should be either somewhere in the emerge --help or integrated into the emerge command completely.
| Dralnu wrote: | The rewrite/overhaul of Portage has been covered before. I think one of the major issues with that is that currently, Portage doesn't use a database by default (I think you can covert it to use one, but I'm unsure of that), and another is the fact you're dealing with a relativly large scripted program. While granted they are just as fast as compiled, converting Portage to C/C++ might oculd be of help, if alot of the second-hand programs (eix, equery, ect.) where added in by default instead of the generally slow/crappy tools that are in Portage by default, and waste space.
Also, doing that you could also remove the python reliance from Portage, and reduce system size some (if you don't script/don't script in python). |
With rewriting/overhauling portage I explicitly do not mean using another language than Python. I believe a refactoring session would both improve the state portage is in sufficiently and save time. Removing the python dependancy is totally irrelevant in size. I can not image a situation where someone would want to use portage, but have such extreme requirements that they can not afford the installation of Python.
| legine wrote: | | Hmm yea there are some issues with Stable and unstable tree. But I think the need of mixing and not mixing is unchanged. |
Well, I'm not sure for how long you've been using Gentoo, but I think there seems to be a trend for people who used to run Gentoo from way back to stop using it, because it has been taking too much time, lately, to keep in good health. And I think this is because of "issues with Stable and unstable tree" combined with a slight change in the need to keep configuration files. See my bit at the top of this post.
| krigav wrote: | WTF? Could you please explain me HOW a well running system could be broken when you leave it in a functional state. And when you return it's broken???
Maybe you should keep you children away from your gentoo box while you away? |
This might be because of the difficulties portage is having with uninstalling packages. The way portage handles dependancies is absolutely great for installing packages. But it sucks for removing packages. This has to be fixed! |
|
| Back to top |
|
|
Dralnu
Veteran
Joined: 24 May 2006
Posts: 1919
|
| Posted: Mon Aug 28, 2006 1:48 pm Post subject: |
|
|
| legine wrote: | | Quote: | yup it makes perfect sense. Especially since if I remember correctly that the ebuilds are written in bash not python.
|
Portage is more then ebuids. And I remebering it right, it is a mix between python and bash.
The question is is it time for a rewrite? Maybe avoiding some design failures.
There was a Portage based on C++. We also had a portage system that was featuring databases (I think it could store data within a mysqlserver.) But it never got accepted...
(search for lost apps or tools or something. If I find time I search for it...)
The stuff with equery and thing is not with portage since they have different development cycles. So it is wise to keep them seperated. |
Two words: Modular System
_________________
The day Microsoft makes a product that doesn't suck, is the day they make a vacuum cleaner. |
|
| Back to top |
|
|
ArturHawkwing
n00b
Joined: 20 Oct 2004
Posts: 22
Location: MN, USA
|
| Posted: Mon Aug 28, 2006 2:08 pm Post subject: |
|
|
| legine wrote: | | Quote: | yup it makes perfect sense. Especially since if I remember correctly that the ebuilds are written in bash not python.
|
Portage is more then ebuids. And I remebering it right, it is a mix between python and bash.
The question is is it time for a rewrite? Maybe avoiding some design failures.
There was a Portage based on C++. We also had a portage system that was featuring databases (I think it could store data within a mysqlserver.) But it never got accepted...
(search for lost apps or tools or something. If I find time I search for it...)
The stuff with equery and thing is not with portage since they have different development cycles. So it is wise to keep them seperated. |
I know that it's more than just ebuilds. I"ve had a look at portage before. The referance to the ebuilds was just to specify that they don't rely on python. The official portage utility is probably about 85% python and 14 percent bash I was just agreeing with dralnu when he said it would remove the relience of python if it was written in c so it made sense.
Was the C++ version of portage put out by the gentoo devs? I remember a thread from a while back where it was meantioned as a project not picked up by the developers because it would be too much work.
I haven't heard about using mysql for a portage database but in the wiki there's instructions on how to use python-cdb which sped it up a bit. but i still think 5 minutes is still way too long. It didn't become standard because there were some issues with it not being 100% stable on all systems.
I'm not really sure what the referance of equery was for since I was talking about portage.
It's been time for a re-write for at least 2 years.
I was going by memory because I haven't been a gentoo user since december. I tried it again on a spare partition about a month and a half ago though. |
|
| Back to top |
|
|
ArturHawkwing
n00b
Joined: 20 Oct 2004
Posts: 22
Location: MN, USA
|
| Posted: Mon Aug 28, 2006 2:28 pm Post subject: |
|
|
| Quote: |
Gentoo installations now need a lot more configuring than they used to need. This may be perceived as a form of bloat. I'm going to illustrate the point using USE flags, but the following bit concerns Gentoo as a whole:
Quite a bit of the added configuring is because nearly all compile-time options have been transformed into (package specific) USE flags. Back when I installed Gentoo for the first time, there were some 20 USE flags. And that was it. At the time of writing there are 357(!) USE flags in use.desc alone!!
This change can be interpreted two ways. At the moment the majority of developers seem to see it as a positive change, as it enables more choice. ("And isn't choice what Gentoo is all about?") But an increasing number of users who have experienced the older days notice that this has introduced an increased amount of effort necessary to use Gentoo, and see it as a negative change.
|
| Quote: |
Well, I'm not sure for how long you've been using Gentoo, but I think there seems to be a trend for people who used to run Gentoo from way back to stop using it, because it has been taking too much time, lately, to keep in good health. And I think this is because of "issues with Stable and unstable tree" combined with a slight change in the need to keep configuration files. See my bit at the top of this post.
|
You don't know how much I agree with this. I wasn't using gentoo back in th beginning, but I can say I saw a big differance between spring of 2004 and last december
| Quote: |
This might be because of the difficulties portage is having with uninstalling packages. The way portage handles dependancies is absolutely great for installing packages. But it sucks for removing packages. This has to be fixed!
|
The original sorcerer distro has a way to keep track of what is installed directly and as a dependancy. And when you uninstall the directly installed program it would uninstall the no longer needed dependancies. I've also heard that BSD uses something like this as well. |
|
| Back to top |
|
|
Dralnu
Veteran
Joined: 24 May 2006
Posts: 1919
|
| Posted: Mon Aug 28, 2006 2:59 pm Post subject: |
|
|
Removal of deps have been there for a long time. Debian does it automatically I think, but I could be wrong.
As for there being 357 USE flags, that is quite a list, but if you were to, say, drop it back to ~100, then you would be cutting into the configurability. Did you figure about how many of those flags aren't used/used rarely? I think my USE flags list in make.conf is ~60 entries, and i havn't had to mess with them very often.
You might want to check out ELOG in make.config.example. It isn't a perfect fix, but its a step in the right direction. Its in the works from what I understand, just moving slow.
| iKiddo wrote: | | With rewriting/overhauling portage I explicitly do not mean using another language than Python. I believe a refactoring session would both improve the state portage is in sufficiently and save time. Removing the python dependancy is totally irrelevant in size. I can not image a situation where someone would want to use portage, but have such extreme requirements that they can not afford the installation of Python. |
Ok, so removing the interpreter wouldn't change system size. I wish to know how you store things that doesn't require hard drive/some form of space on a medium. The issue was removing bloat, the python dep could be considered bloat since it can be removed. It would also fix the issue with python updates killing Portage. That takes care of two issues.
Also, from my knowledge, you have the option to mix stable/unstable, or just run unstable. If a package doesn't work:
| Code: | | echo ">=foo.bar-x.y" >> package.masked |
If you want to maintain a bleeding-edge system, thats what you have to do. Gentoo is still the most advanced distro avalible from my knowledge, which also doesn't pull in unneeded deps for various packages which contributes to "bloat", which at this point is seeming more like a refrence to the size of a system as opposed to the size of the same basic system from, say, 4 or 5 years ago. Code changes, and things are added. It won't stay tiny forever (unless its part of the 10000 project, which tries to keep code to <10,000 lines of code).
From my understanding, if you want a system that is self-configuring/doesn't require much configuring, use something else. I am sure you can find something you like out there, but honestly I don't see the configuration a problem (I've been using Linux for about a year and a half, and was using SuSe before-hand, and I have only minor problems usually with issues with configuration outside of programs that the Gentoo Devs don't make).
I don't mean to sound, hmm, annoying or arrogant, but if thats the case then I may be failing to see the issue here. I by no means am saying that Gentoo is perfect, but from what I understand of it, its kind of like Burger King and having it done your way, and not how the devs want it done (which is an issue at times, but thats another case altogether)
_________________
The day Microsoft makes a product that doesn't suck, is the day they make a vacuum cleaner. |
|
| Back to top |
|
|
Dralnu
Veteran
Joined: 24 May 2006
Posts: 1919
|
| Posted: Mon Aug 28, 2006 3:04 pm Post subject: |
|
|
| ArturHawkwing wrote: | | legine wrote: | | Quote: | yup it makes perfect sense. Especially since if I remember correctly that the ebuilds are written in bash not python.
|
Portage is more then ebuids. And I remebering it right, it is a mix between python and bash.
The question is is it time for a rewrite? Maybe avoiding some design failures.
There was a Portage based on C++. We also had a portage system that was featuring databases (I think it could store data within a mysqlserver.) But it never got accepted...
(search for lost apps or tools or something. If I find time I search for it...)
The stuff with equery and thing is not with portage since they have different development cycles. So it is wise to keep them seperated. |
I know that it's more than just ebuilds. I"ve had a look at portage before. The referance to the ebuilds was just to specify that they don't rely on python. The official portage utility is probably about 85% python and 14 percent bash I was just agreeing with dralnu when he said it would remove the relience of python if it was written in c so it made sense.
Was the C++ version of portage put out by the gentoo devs? I remember a thread from a while back where it was meantioned as a project not picked up by the developers because it would be too much work.
I haven't heard about using mysql for a portage database but in the wiki there's instructions on how to use python-cdb which sped it up a bit. but i still think 5 minutes is still way too long. It didn't become standard because there were some issues with it not being 100% stable on all systems.
I'm not really sure what the referance of equery was for since I was talking about portage.
It's been time for a re-write for at least 2 years.
I was going by memory because I haven't been a gentoo user since december. I tried it again on a spare partition about a month and a half ago though. |
Making a point again, removing the reliance on python would also cure some issues people have with python updates breaking portage.
Plus doing a re-write, you might as well fix other issues with it as well.
To go into another point, the database system also, in some cases, could be an issue for some of us (me, for instance). Some of us with little or no previous experiance working on databases, plaintext makes sense for general work (and it works well). It would be nice, though, to have it as an option (if you run MySQL, there is no real sense in NOT using it for Portage, is there? If you don't run a database, then the addition could add to bloat is its sitting around doing only minor work from time to time).
EDIT:
While modifying Portage to be able to fixup the database itself via commands (some of which are already in in other forms) in some cases would be more of a pain then doing it longhand with an editor and .txt files. I would hate to try and handle USE flags this way, since chances are you would either have to hand-type in dozens of flags, or you would have to go through and turn them on and off one at a time. Only solution to this I could see would be a good, pre-formated/well-read file like make.conf which could be read into the database to fix this, but that is kind of redundant to have multiple instances of a single config in the system.
_________________
The day Microsoft makes a product that doesn't suck, is the day they make a vacuum cleaner. |
|
| Back to top |
|
|
ArturHawkwing
n00b
Joined: 20 Oct 2004
Posts: 22
Location: MN, USA
|
| Posted: Mon Aug 28, 2006 4:04 pm Post subject: |
|
|
| Dralnu wrote: |
Making a point again, removing the reliance on python would also cure some issues people have with python updates breaking portage.
Plus doing a re-write, you might as well fix other issues with it as well.
To go into another point, the database system also, in some cases, could be an issue for some of us (me, for instance). Some of us with little or no previous experiance working on databases, plaintext makes sense for general work (and it works well). It would be nice, though, to have it as an option (if you run MySQL, there is no real sense in NOT using it for Portage, is there? If you don't run a database, then the addition could add to bloat is its sitting around doing only minor work from time to time).
EDIT:
While modifying Portage to be able to fixup the database itself via commands (some of which are already in in other forms) in some cases would be more of a pain then doing it longhand with an editor and .txt files. I would hate to try and handle USE flags this way, since chances are you would either have to hand-type in dozens of flags, or you would have to go through and turn them on and off one at a time. Only solution to this I could see would be a good, pre-formated/well-read file like make.conf which could be read into the database to fix this, but that is kind of redundant to have multiple instances of a single config in the system. |
oh shoot I was editing my last post before I posted it to make it flow better and I guess I deleted some of it I didn't mean to. When I meantioned python-db I was meaning it's use to make searches faster. Especially searchdesc. It shortened the time down to about 5 minutes but I thought that was still too slow. I wasn't meaning using a database for configuration. |
|
| Back to top |
|
|
Dralnu
Veteran
Joined: 24 May 2006
Posts: 1919
|
| Posted: Mon Aug 28, 2006 4:21 pm Post subject: |
|
|
| ArturHawkwing wrote: | | Dralnu wrote: |
Making a point again, removing the reliance on python would also cure some issues people have with python updates breaking portage.
Plus doing a re-write, you might as well fix other issues with it as well.
To go into another point, the database system also, in some cases, could be an issue for some of us (me, for instance). Some of us with little or no previous experiance working on databases, plaintext makes sense for general work (and it works well). It would be nice, though, to have it as an option (if you run MySQL, there is no real sense in NOT using it for Portage, is there? If you don't run a database, then the addition could add to bloat is its sitting around doing only minor work from time to time).
EDIT:
While modifying Portage to be able to fixup the database itself via commands (some of which are already in in other forms) in some cases would be more of a pain then doing it longhand with an editor and .txt files. I would hate to try and handle USE flags this way, since chances are you would either have to hand-type in dozens of flags, or you would have to go through and turn them on and off one at a time. Only solution to this I could see would be a good, pre-formated/well-read file like make.conf which could be read into the database to fix this, but that is kind of redundant to have multiple instances of a single config in the system. |
oh shoot I was editing my last post before I posted it to make it flow better and I guess I deleted some of it I didn't mean to. When I meantioned python-db I was meaning it's use to make searches faster. Especially searchdesc. It shortened the time down to about 5 minutes but I thought that was still too slow. I wasn't meaning using a database for configuration. |
There is a good search utility called eix. Its pretty snappy 
_________________
The day Microsoft makes a product that doesn't suck, is the day they make a vacuum cleaner. |
|
| Back to top |
|
|
legine
Guru
Joined: 27 May 2004
Posts: 555
Location: Germany
|
| Posted: Mon Aug 28, 2006 4:21 pm Post subject: |
|
|
| iKiddo wrote: | Well, I'm not sure for how long you've been using Gentoo, but I think there seems to be a trend for people who used to run Gentoo from way back to stop using it, because it has been taking too much time, lately, to keep in good health. And I think this is because of "issues with Stable and unstable tree" combined with a slight change in the need to keep configuration files. See my bit at the top of this post.
|
Well I am here since 27.05.2004. Well I joined around that time then.
But I think most problems are due:
| iKiddo wrote: | | This might be because of the difficulties portage is having with uninstalling packages. The way portage handles dependancies is absolutely great for installing packages. But it sucks for removing packages. This has to be fixed! |
| Quote: | | there were some 20 USE flags. And that was it. At the time of writing there are 357(!) USE flags in use.desc alone!! |
Well I have about, uhh, let me lie 30(?) useflags in make .conf and the others I need come into package.use.
I think the Idea of useflags is changeing. Back when I started I think most useflag were very cryptic and for inpackage configuration. Today most useflags are describing some feature. (Yea modular Wizardry) I know most features. I know most things I to be included. And I can stop searching the right packages for it.
If I am not wrong the answer to how am I able to look at wincodecs?
Was back 2004 emerge wincodec
And today is activate the wincodec useflag.
Both are good ways for the same thing. But the latter has the advantage I can check which app supports what feature. I think most of the frightening 375 Useflags are of this type. And this is pretty intuitive to me. Of course for longtime Gentoo lovers it is a change. Maybe too big for some? (I can understand that)
| Dralnu wrote: | | Two words: Modular System |
Dont think that would do the trick. The difference with the portage package is that portage stops emergeing at the time itself has been emerged. So if there is a small update in Gentotools should it stop the hole update process? - Not a thing I like to have. A Metapackage would be okey For lets say picking vital admin tools. But Lets see how the modular stuff is evolving before I get fantasies
| Quote: | | The whole point of adding it directly to the emerge interface is that users who do not frequently browse the forums, the mailing lists, or gentoo.org in general (and the GWN in particular) |
Well GWN is the most basic method to be warened about major things. I am sure you can subscribe to some list to get it at home with your usual mail. I needed about 2 minutes to be aware on whats going on and to recognize that there is a migration guide in the case of X. Thats not to much of work beeing expecting to someone running gentoo right?
_________________
quote from Spaceballs:
Dark Helmet:[...] we were told to comb the desert, so we're combing it! [puts down bullhorn] Find anything yet?!
Soldier: Nothing yet, sir. |
|
| Back to top |
|
|
iKiddo
Guru
Joined: 27 Jun 2002
Posts: 341
Location: Europe?
|
| Posted: Mon Aug 28, 2006 4:30 pm Post subject: |
|
|
| Dralnu wrote: | | Removal of deps have been there for a long time. Debian does it automatically I think, but I could be wrong. |
"Removal of deps" isn't going to cut the cake. We need a solid method to keep track of dependancies, so as to be able to automatically get rid of unneeded dependancies in good fashion.
| Dralnu wrote: | | You might want to check out ELOG in make.config.example. It isn't a perfect fix, but its a step in the right direction. Its in the works from what I understand, just moving slow. |
I have followed the progress of the logging funcionality since the beginning, it has been productive for a while, but was part of 2.1, and therefore couldn't be released. It's not moving slow, it's finished. It has nothing to do with the rest of this thread, though.
| Dralnu wrote: | | iKiddo wrote: | | With rewriting/overhauling portage I explicitly do not mean using another language than Python. I believe a refactoring session would both improve the state portage is in sufficiently and save time. Removing the python dependancy is totally irrelevant in size. I can not image a situation where someone would want to use portage, but have such extreme requirements that they can not afford the installation of Python. |
Ok, so removing the interpreter wouldn't change system size. I wish to know how you store things that doesn't require hard drive/some form of space on a medium. The issue was removing bloat, the python dep could be considered bloat since it can be removed. |
Note I didn't say it wouldn't save disk space. The source files are only 7MB big. The form of bloat we are dealing with does not concern disk space of this order, therefore I consider the size of python irrelevant.
| Dralnu wrote: | | It would also fix the issue with python updates killing Portage. |
This might be something to look at, but I think the problem is inferior to the amount of human resources spared by staying with Python.
| Dralnu wrote: | Also, from my knowledge, you have the option to mix stable/unstable, or just run unstable. If a package doesn't work:
| Code: | | echo ">=foo.bar-x.y" >> package.masked |
|
You are completely missing the point here. After you've found out a package doesn't work, it's too late.
| Dralnu wrote: | | If you want to maintain a bleeding-edge system, thats what you have to do. Gentoo is still the most advanced distro avalible from my knowledge, which also doesn't pull in unneeded deps for various packages which contributes to "bloat", which at this point is seeming more like a refrence to the size of a system as opposed to the size of the same basic system from, say, 4 or 5 years ago. Code changes, and things are added. It won't stay tiny forever (unless its part of the 10000 project, which tries to keep code to <10,000 lines of code). |
Gentoo is loosing it's advancedness, because of all the options available nowadays. Apart from not pulling in unneeded deps, it is at the moment also not getting rid of unneeded deps properly. I do not consider bloat as the size of the system in bytes, or lines of code. Bloat is about not being efficient. We're not talking KDE, here. 
| Dralnu wrote: | | From my understanding, if you want a system that is self-configuring/doesn't require much configuring, use something else. I am sure you can find something you like out there, but honestly I don't see the configuration a problem (I've been using Linux for about a year and a half, and was using SuSe before-hand, and I have only minor problems usually with issues with configuration outside of programs that the Gentoo Devs don't make). |
Here you're being ignorant. As I've stated, I have the impression Gentoo needs more configuration nowadays. Bringing this back will stop all of the people who have been using Gentoo for years from leaving. (And keep the forums a tad more clean.)
See ArturHawkwing's post above.
| Dralnu wrote: | | I don't mean to sound, hmm, annoying or arrogant, but if thats the case then I may be failing to see the issue here. I by no means am saying that Gentoo is perfect, but from what I understand of it, its kind of like Burger King and having it done your way, and not how the devs want it done (which is an issue at times, but thats another case altogether) |
I agree you might be missing the point.
/me dreams of having drobbins back at gentoo...   |
|
| Back to top |
|
|
ArturHawkwing
n00b
Joined: 20 Oct 2004
Posts: 22
Location: MN, USA
|
| Posted: Mon Aug 28, 2006 5:50 pm Post subject: |
|
|
| Dralnu wrote: |
There is a good search utility called eix. Its pretty snappy
|
ya eix is allot snappier than searchdesc there are at least a couple of other search tools for gentoo. I've tried many of the non official portage tools. I liked some of them. Even some from the script depository. |
|
| Back to top |
|
|
Lore
Apprentice
Joined: 16 Feb 2004
Posts: 152
Location: Karlsruhe
|
| Posted: Wed Aug 30, 2006 11:52 am Post subject: |
|
|
Just my two cents...
In 2003, I bought my Notebook. Formated the preinstalled Windows and installed Gentoo. Since the, it's running without even one reinstall. I've never seen any OS on any of my computers running so long before.
It was exciting to configure and compile your own kernel. It was exciting to tune and break your system testing new technologies. It was exciting to compile your own software. I spent a lot of time doing this and I enjoyed it. I've learnt many things about linux und technology.
Nowdays, I think, I've seen nearly everything. There's no thrill anymore. I merely want to USE my system. So, what exhilarated me in the past, is now annoying.
That's not Gentoo's fault. I've only emerged to a new level of experience.
On this notebook, I will run gentoo until the hardware breaks (never change a running system...). But on my next piece of hardware, I probably will intall an other OS.
Gentoo is a distro for learning and having fun, but it's not a distro for life.
That's the view of a me as a user of gentoo on its single, private desktop machine. |
|
| Back to top |
|
|
legine
Guru
Joined: 27 May 2004
Posts: 555
Location: Germany
|
| Posted: Wed Aug 30, 2006 12:14 pm Post subject: |
|
|
Have you made up your mind what will be next?
Maybe going for Gentoo/BSD?
_________________
quote from Spaceballs:
Dark Helmet:[...] we were told to comb the desert, so we're combing it! [puts down bullhorn] Find anything yet?!
Soldier: Nothing yet, sir. |
|
| Back to top |
|
|
Shadow Skill
Veteran
Joined: 04 Dec 2004
Posts: 1023
|
| Posted: Thu Aug 31, 2006 12:52 am Post subject: |
|
|
I find it somewhat miraculous that my boxes run as well as they do consider the complete lack of reverse dependency resolution. I guess since I have all but given up on DE's I don't experience the horror that is trying to uninstall them.
_________________
Ware wa mutekinari.
Wa ga kage waza ni kanau mono nashi.
Wa ga ichigeki wa mutekinari.
"First there was nothing, so the lord gave us light. There was still nothing, but at least you could see it." |
|
| Back to top |
|
|
GWilliam
Guru
Joined: 29 Dec 2005
Posts: 350
|
| Posted: Sun Sep 03, 2006 11:19 pm Post subject: |
|
|
#NULL
Last edited by GWilliam on Sun Jul 25, 2010 3:18 pm; edited 1 time in total |
|
| Back to top |
|
|
mimosinnet
l33t
Joined: 10 Aug 2006
Posts: 717
Location: Barcelona, Spain
|
| Posted: Sun Sep 10, 2006 12:26 pm Post subject: When you come to Gentoo... |
|
|
I just arrived to Gentoo about a month ago and I started my trip to Linux two years ago with Mandriva. I was a windows user without much knowledge of computers.
I have been amazed how EASY has been to use gentoo (I have just followed the documentation and search the forums). It is true that most of the time I do not understand much of what I am doing, but I learn while I do things. I have just uptated from gcc 3.4.6 to gcc 4.1.1., and learned about a mysterius creator of programs that sometimes has to create a superior being (upgrade to a newer version) and re-recreate the world (emerge world).
It has been really shocking to find a post saying "why Gentoo sucks, and why it will ultimately die". Yes, it is provocative, but I do not see really the point. Looks like negative criticism: things are really bad ("gentoo sucks") and somebody should do something to get things better ("my only real gripe with Gentoo is that I'm getting really tired of things breaking"). Because of the surprise, I have search the posts by Zork the Almighty, and last year the oppinion was:
| Quote: | Wed Aug 10, 2005
Gentoo is not so much a Linux distribution at this point - it is more like a way to install Linux software. You don't get any sort of consistency beyond what you provide yourself, and if you wait too long to upgrade things you end up screwed because the ebuilds are removed from portage. I like Gentoo, I've used it for 3 years, but it really is a stupid system. My next install is going to be Slackware, for exactly these reasons. |
Surprisingly, the discussion in the forum has move from Negative Criticism to a discussion at the diferent aspects that could improve gentoo. From the contributions, I feel there are many people in the community that won't let Gentoo die.
As a recent Gentoo fan: Please, Zork the Almighty: do not let Gentoo die. |
|
| Back to top |
|
|
|
|
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
|
|
Gentoo Chat
