Hacker News new | past | comments | ask | show | jobs | submit login

Well, I want to make a few things clear here. First of all, I was being a bit snarky obviously. I didn't know about "aptitude purge", but it's a bit ironic that when googling for that, the first result is a forum post about someone with orphaned config files that aptitude purge wouldn't remove... sounds reliable.

I think it's time for desktop OS makers (and that includes Apple, because OS X isn't any different from Linux or Windows in this regard at all) rethink the way software is installed and organized. In my opinion, the way you can install and try applications without consequences on iOS devices has a lot going for it. Users actually install stuff on their iPads, which they don't on their Windows computers because they're scared of unintended consequences.




Edited to remove unhelpful rage

It sounds like you don't know Debian very well. That's fine. No reason you should. I know Debian reasonably well. I can tell you that in four years of using aptitude, I have found a very small number of edge cases where it does not work sanely, efficiently and helpfully. (To clarify it's really APT + aptitude. aptitude is one possible tool to use APT.) It's not perfect; no software is. But it's quite remarkable. I believe it to be a far better packaging system than anything available for OSX, including Homebrew - which I like a lot and which gets better all the time.

To respond to your follow-up, I deeply disagree about iOS representing a good future for software. The kernel already implements one filesystem - and that's a big job. Why would we want every application to have to reinvent that wheel? Worse yet, how we will we allow and help users to share data between all our little sandboxes if we go that route? I think the iOS sandboxing is a disaster and huge step backwards for software.

A final question: in your original post you say if you remove an iPhone app, no trace of it is left. How would you know? My point being that the system is so locked down, there's really no basis for us to know that. For all we know, iApps leave the drive cluttered with their configuration files and preferences. Maybe you develop for Apple and can show me I'm wrong (or someone else can), but as a user there's no real way for me to know this, is there?


(Edit: This response was written at a point when the parent post was quite a bit more unfriendly than it is now. I'll leave this here regardless, even though it's out of context now.)

Was I in any way rude to you?

1. That's not what I said.

2. I admit that I don't use Debian, but internally, package managers don't differ that much. I was criticizing the model where software has to install files all over the place. I think the proper response should be "make a simpler model", not "write a complex tool that can keep track of this mess".

3. So I was wrong and Debian packages come with cleanup scripts. That's good. Am I right in assuming that those cleanup scripts are only as good as the maintainer of the package, i.e. there's nothing inherent in the design of Debian that makes this cleanup function work reliably. (Note that some config files might have been created after installation, i.e. are not part of the original package.)

4. What?

5. Have you talked to a casual computer user recently? They are afraid of installing unknown software, yes. Almost everyone I know has had malware on their PC at one point or another.


You were not rude. My initial response was unhelpful and aggressive, and I quickly edited it to respond to your arguments. I apologize for the initial response.

For the record, regarding 3, there is something in the design of Debian that makes the cleanup function work very reliably. (Again, not perfectly, but very reliably.) There are strict, strict rules for Debian maintainers in terms of the pre-install and post-removal scripts that go with packages. These rules help to maintain a very uniform system - even though its actively developed by thousands of people all over the world who are often only loosely connected to each other.


Apology accepted. Much of my dislike for package managers is on theoretical grounds (I use Macports, but only casually), so it's great to argue with someone who actually knows something about them.


well, as a user, there's no real way for you to know that aptitude purge completely removes everything, you have to trust that it works. Sure, as a power-user or developer, you can go look around /etc or /usr/share or wherever and remove all those configuration and documentation files, but will you?

I used debian for about a year, I trusted that aptitude purge removed as much as it could. I knew that trust was misplaced, but for my purposes as a user, it was good enough.

Similarly, for iOS devices, I don't know how well it removes config files and whatnot. I simply trust that it removes everything. I have slightly more trust in iApp config files being removed because these apps are sandboxed, and it's a relatively simple matter of removing the directory the app fills.


I hear what you're saying, but I think you misunderstood me.

I wasn't trying to distinguish regular users from power users. I was trying to distinguish all users (however knowledgeable) from specially licensed developers. I was trying to make about about the openness and freedom of Debian as a platform versus the locked-down nature of iOS.

Also, I disagree about your trust: it wasn't misplaced.


> I deeply disagree about iOS representing a good future for software.

Now that's a point worth discussing. iOS is locked down to the point where I would almost call it unusable -- we don't disagree here. However, I'm also convinced that some steps in that direction are needed to keep the "general-purpose computer" viable as a mass-market device. A computing environment that's open enough so that software can be useful, yet closed enough so that it can't be harmful. (Harmful defined in the broad sense that it can't have any unwanted or permanent effect on the system, including: changing the homepage, installing toolbars, running persistently in the background and so on.) All this has to work in the absence of a central authority.


Stallman has a nice fable about locking computers down - http://www.gnu.org/philosophy/right-to-read.html


He did say that it has to work in the absence of a central authority. It's hard for a central authority to control your device when it doesn't rely on a central authority.

Maybe we just need to re-imagine the operating-system level security model in a way that tries to tackle the security issues that we have today.




Consider applying for YC's first-ever Fall batch! Applications are open till Aug 27.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: