I've been playing with remastersys to make an iso, and hacking at corenominal's cb-welcome and lib-cb-welcome scripts The idea was to try and make what so many here have been asking for - a more extensive choice of apps to install, while keeping the iso to a reasonable size. Guess what - it's getting very unwieldy, and is a very inelegant process imo.
So I'm not going to bother any more!
man up && man apt-get
I second the "clean slate" opinion. I like crunchbang exactly because of that approach.
Could I humbly ask if there is a test ISO or a pre-release of sorts for testing on #! 12?
(I figured there would be a thread about this, but I couldn't find one. )
Last edited by pachanka (2014-03-06 21:08:31)
Here is what I'm imagining:
I basic open box setup that allows you a simple/clean desktop, a few themes (which are probably just variations of the basic theme: green, light, dark, blue, etc) and a web browser. Then on top of this we have what are called 'build packs', which might just be puppet/chef/ansible scripts that do the heavy lifting. These build packs then come in a couple different flavors:
One flavor is obviously GUI based. If you want your conky to have sunshines when it is sunny and raindrops when it is raining there is a build pack for that. If you want a more modern, industrial feel there is a build pack for that.
The next flavor is what you might call philosophy driven. These are where most of this thread is based. It focuses on the packages. There would probably be a default build pack that ships and is installed via the welcome screen if one isn't selected. I can imagine one that installs uzbl, dmenu, etc for those minimalist purist. But, also one that installs chrome, libre office, shot well, etc for the complete desktop experience.
I can also imagine 'ruby-developer' build back that installs vim, tmux, rvm, etc. You could also see one that installs a bunch of security packages that look like Kali linux, etc.
This seems like a positive way to go and it would get more community involvement and also decrease the CD install size to under 700mb. It's possible that you could also offer #! + default build pack for DVD install if someone didn't have a network connection available.
Any feedback on this idea?
I think its a cool idea
The clean slate sounds really good.
One of the reasons I choose #! is because of the fast setup. Less applications to remove after installation.
In fact, my "dream version" of #! would be the nice Openbox configuration it comes with, Iceweasel and a terminal, nothing more, besides drivers and the usual behind the scenes stuff.
Children of the night... What music they make.
Ozitraveller, Thanks for that. It looks like they are using chef. I have a slight preference for ansible or puppet over chef, but it's all probably the same. There's no reason we couldn't use all of them, but given the community size, it probably makes sense to pick one and focus our community efforts on just that technology as far as resources go.
Unia (or others), do you have any thoughts on why it failed as a community effort? Was there not a central repo? No one created the first basic ones? Etc.
CrunchBang used to have a "lite" edition, and it was beloved by at least a few users (I don't know the download statistics of how it compared in popularity to the full edition). So it is not unreasonable to speculate that such a thing might exist again in the future. LinuxBBQ people for example have shown there is a userbase for this type of project.
That said, in my opinion, a "distribution" must meet the basic needs of users, or else it is not a "distribution" at all. I personally consider a distro "half baked" if it does not allow the user out-of-the-box to surf the web, listen to music, watch a video, edit images, create office documents for work/school, etc.
This is 2014 and a gigabyte of storage costs mere pennies. It's time as a community to lose the "oh no, it has an image editor, it is so bloated!!" mentality.
Like antisoup mentioned at the end, we could have a dvd install that includes the basic #! and a default build pack. This would satisfy users needing a traditional full-fat distro and the barebones edition can be used by others. Just like the old #!, we would have two editions. It's not like we can only have one.
Last edited by tamikan (2014-03-07 15:37:13)
Personally I don`t like the idea. Because let`s say a package contains chrome, libre-office and shotwell etc..... I probably just want some apps from this package and other apps from other packages. There will be fights about what should be in each of those packages, and what should be excluded. If at all doing something like this, I think there should be only two options:
1. The default Crunchbang install
2. A minimal openbox-base install, possibly with tint2, Thunar, Geany and Terminator, and everything else up to the user post-install.... With the following philosofy, Take it from here yourself....
- apt-mark hold account
I see your point, but I don't think your approach is incompatible with mine. A couple points. I'm imagining a tool such as ansible, chef, or puppet that would allow you to customize the build packs themselves. So there shouldn't be any fighting over what goes in what. There might be a couple 'official' ones that are highly recommended, but that shouldn't stop anyone from rolling their own.
As far as your first option, I think that is fine as I've mentioned above. however, regarding #2, I don't see why one person can't install the openbox-base install and then pick from any number of build packs and/or roll their own and take it from there themselves.
Last edited by antisoup (2014-03-07 21:37:07)
Last edited by anonymous (2014-03-08 03:18:27)
Last edited by KrunchTime (2014-03-08 10:50:29)
To Address how this might be different from a simple apt-get install:
I understand this is a major cultural shift for CrunchBang, but it is easy to imagine these being extended beyond just apt-get foo baz bar. If this extension happened, we should be able to decrease resources dedicated to support and also make crunch bang very powerful and friendly. Let me explain…
Imagine we have a tool that can fetch these scripts and a new user is having issues with wifi. In order to trouble shoot wifi, you might want any number of outputs from commands ran at the command line. You could ask the user to run the commands your self after you consulted the previous forum post that asked another user to run. Or you could ask them to run: app run wifi-troubleshooting and copy the url of the the paste bin to the forum thread. The application would automatically run those commands, save the output and load it up to web resource that all could access.
Now we've identified that this user is using some realtek wifi driver that needs to be jiggered with to make it work. Fortunately, we've already got a solution to that. Instead of linking to some forum post about it, which is 6 months old and people report it helped or it didn't, and asking have you tried this? We point them to app run realtek8xxx which is the latest and best way the community has decided to fix the driver issue.
Let's now turn to the way we shape the way the GUI works. There are people who have absolutely beautiful desktops. Imagine a world where those desktops also sit in a repository. You find the one you want and you should be able to run the script that would then automatically make that happen on your desktop.
There will be some community work to do to establish how somethings are handled because of the issue of idempotence and not wanting to overwrite existing config. However, I see this as a huge win if we can figure out how to get it off the ground.
I'm willing to write the client to fetch and run these, as well as build the server to store these from. I'm also willing to get my hands dirty with some build scripts, but I have much less experience there.
You obviously haven`t fixed many wifi-issues. The best way the community has decided to fix the driver-issue? If I should listen to how the majority in this forums thinks that I should make my wifi work, it wouldn`t work, because I have a special revision of that chip, that doesn`t work with the "best way". It isn't possible to make decitions about the best way to solve anything.
It's a bad idea no matter how you look at it. An automated support-system sounds quite horrific. People wouldn't learn from it, and it would generate more issues than it solved. Then people might just as well use Windows or Mac who has such solutions. I get really annoyed when Windows thinks they have solutions to my issues, most of the times they run trough a lot of steps that hasn`t got anything to do with the actual issue, and you always end up with a unfixed issue. The whole script is just an annoyance.
Beside, Linux is for the people that want the freedom to control every aspect of the install themselves. Free source code is what it's all about, and it's even worse to leave decitions about your system to a community, than it is to leave it to Microsoft or Apple, because let's face it, a linux-community consists of only of a few skilled people, and any kind of democracy would generate the worst solutions possible.
Last edited by ew (2014-03-08 17:56:27)
- apt-mark hold account