TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Introducing Kirk

41 点作者 IanChiles超过 11 年前

16 条评论

blueblob超过 11 年前
I take quite a big issue with the statement &quot;The lack of information and documentation on developing OSes is, in my opinion, a pretty big problem&quot;<p>There are entire forums for this, ie osdev <a href="http://wiki.osdev.org/Expanded_Main_Page" rel="nofollow">http:&#x2F;&#x2F;wiki.osdev.org&#x2F;Expanded_Main_Page</a><p>There are entire books for this <a href="http://www.amazon.com/s/ref=sr_ex_n_1?rh=n%3A283155%2Cn%3A5%2Ck%3Aoperating+systems&amp;bbn=5&amp;keywords=operating+systems&amp;ie=UTF8&amp;qid=1379950869" rel="nofollow">http:&#x2F;&#x2F;www.amazon.com&#x2F;s&#x2F;ref=sr_ex_n_1?rh=n%3A283155%2Cn%3A5%...</a> amazon lists about 57,000 results for operating systems books<p>You will get out of LFS what you put into it. It tells you how to download the source, extract and install it. When you have the source, nothing is preventing you from looking at it.<p>I also take issue with &quot;how can we hope to continue developing and innovating on the OS level when it’s practically impossible to figure out what anything does or how it works.&quot;<p>There are general concepts that you will learn from getting a book on os development (including the kernel) and writing a minimal kernel yourself will give you insights into the work that goes into a full kernel such as the linux one. The concepts, not the specifics are what you need to learn, and you do this through theory, you pick up the rest from implementation (but starting with a kernel, half of the os is already developed fof you).
评论 #6431805 未加载
runjake超过 11 年前
<p><pre><code> For as long as I can remember, I’ve wanted to create my very own operating system, one which I have complete and total control over. Because I have neither the time, nor the skill to write my own kernel, this means that I’d eventually land up creating a OS based on the Linux Kernel. </code></pre> Non sequitur. If you want complete and total control, you&#x27;re going to have to learn the ins and outs of a kernel.<p>I applaud your decision to go forth with this, but is it really worthy of an HN self-post given that Kirk does not exist and is just a blog post at this point?
评论 #6431834 未加载
antocv超过 11 年前
Kirk does not exist, this is a blog post. Just a reminder to everyone like me who went looking for it.<p>And the blog post is about the same dust that was relevant 10 years ago and sparked the creation of LFS. Nothing new to see here.
fusiongyro超过 11 年前
&gt; an OS without limits, built to be open from the ground up - with as open of a development process as possible.<p>I see the benefit of documenting each step. The result will be a pretty great tutorial, or at least a very interesting series of blog posts. But what is closed about the development of the current distros?<p>I hope that Ian is hoping to build a very different kind of distro, breaking the mold. NEXTSTEP was fundamentally a Unix, but everything about it was synthesized to create a totally new organism. Same with OS X. This kind of experimentation hasn&#x27;t been attempted very much with Linux distros, which mostly amount to different installers and package managers. Etoile comes to mind, and apparently CoreOS is discarding much of the mold, but a completely new, holistically built system would be quite interesting to see. I suspect it hasn&#x27;t been attempted much because the sheer amount of work is overwhelming. Writing a new shell is a lot of work. Writing a different init, shell, filesystem layout, and window manager, all unified around some new concepts, would be a lot of work. If that&#x27;s what he&#x27;s aiming to do, the result will be really interesting to use and learn from.
评论 #6432184 未加载
评论 #6432219 未加载
examancer超过 11 年前
I like the concept, but wish this announcement would have come with at least one or two concrete steps taken towards the stated goal. Maybe your experience getting a basic shell working or something similar. At this point I have no reason to believe anything will come of this. Was maybe a bit too early for an &quot;announcement&quot;.
v0land超过 11 年前
Dreams: &quot;I’ve wanted to create my very own operating system, one which I have complete and total control over.&quot;<p>Reality: Another Debian clone out there.
评论 #6431973 未加载
olalonde超过 11 年前
&gt; the development process of Kirk be accessible enough for anyone interested to jump right in and contribute<p>Would be a good idea to host it on Github.
评论 #6431916 未加载
jff超过 11 年前
I want to echo the recommendations of others: learning about kernels and writing your own would be a lot more valuable.<p>I&#x27;m speaking from experience when I say that trying to provide a userland sucks. You&#x27;ve said elsewhere on this page that you want to write your own shell and such; the problem is that people don&#x27;t want IanSH, they want bash or zsh or whatever. If you write IanMACS, they&#x27;ll ask how to run emacs. And I hope the libraries you write can compile Firefox.<p>Or if you just want to write init replacements, modify how configuration is managed... remember that Ubuntu (and Gentoo, but nobody cares about Gentoo anymore) did a ton of this and everyone I&#x27;ve talked to hates it.<p>So here&#x27;s my concerns:<p>- If you intend to write an entirely new userland, you&#x27;re going to spend a lot of time replicating existing work, only to have people complain about it.<p>- If you intend to do a new init, improve configuration management, change how logging works, whatever, your life will be consumed with modifying every program to use your stuff.<p>If you want to learn about operating systems, I&#x27;ll echo the others and say write your own kernel and a minimal userland.
Skoofoo超过 11 年前
The open source desktop is kept alive mostly through donated spare time. This time is limited. We would do well to maximize the impact of time spent creating and maintaining open source desktop software.<p>Right now the standard language to build Linux software is C, which is great for performance, but not great for development speed. There are ports of things like GTK+ to other languages, but ports are less supported and usually don&#x27;t take full advantage of their language&#x27;s unique flavor.<p>To maximize impact per unit of time spent developing open source software, and also perhaps to entice more people into contributing to open source software, I believe it would be a good idea to build a desktop environment&#x2F;widget toolkit with an opinionated API in Ruby. Anyone who has experience writing Ruby on Rails code with an automated red-green-refactor cycle likely could testify that it is pleasant and fast. Why not bring that sort of experience to building desktop software?
评论 #6432207 未加载
clarkm超过 11 年前
If you&#x27;re interested in how a simple version of unix works, I highly recommend checking out Xv6, a teaching OS based on UNIX version 6:<p><a href="https://pdos.csail.mit.edu/6.828/2012/xv6/book-rev7.pdf" rel="nofollow">https:&#x2F;&#x2F;pdos.csail.mit.edu&#x2F;6.828&#x2F;2012&#x2F;xv6&#x2F;book-rev7.pdf</a>
ivanbrussik超过 11 年前
I applaud his efforts - but how is it so many people find main stream Linux distros that non-configurable that they <i>need</i> to create their own?<p>I mean, even distros like Ubuntu are highly configurable. You can kill off as many packages as you want and even compile your own kernel if you really want to.
评论 #6431656 未加载
评论 #6431782 未加载
评论 #6431933 未加载
diaz超过 11 年前
I completly understand the feeling and it&#x27;s something that I wanted to do for a long time. I thought of starting with LFS but never got much time to play with it.<p>I&#x27;ll definetly follow your posts about this.<p>I don&#x27;t get why so many negative comments here. There were many times that I really wanted to understand and configure all my machine just for the sake of it.<p>Understanding and learning, it&#x27;s simple. That&#x27;s the main reason I use Arch, but even with it I don&#x27;t usually go really under it to understand what program calls what and what is the function of program foobar, why it really is needed or how many switches and power can it give.<p>This is great :).
agumonkey超过 11 年前
I&#x27;m curious about the result, I share his idea about LFS being a little bit too &#x27;automatic&#x27; and the feeling of blindness that resulted. The joy of booting a system build from source was there, but didn&#x27;t last long, in fact I quickly hit very weird network issues such as elinks seeing the network but not wget, reinforcing the notion that I there was a need for a different kind of information about the system.
drill_sarge超过 11 年前
I don&#x27;t understand. You want to do a non-GNU&#x2F;Linux OS or what? Or &quot;just another GNU&#x2F;Linux distro&quot;?
评论 #6432283 未加载
donniezazen超过 11 年前
Arch Linux strikes perfect balance between usability and learning. I am looking forward to what you have to say.
internalnet超过 11 年前
How in the hell is this kind of vaporware garbage on the front page?<p>Shut up and code.