| View previous topic :: View next topic |
| Author |
Message |
jdmulloy
Tux's lil' helper
Joined: 24 Dec 2004
Posts: 139
Location: Massachusetts, USA
|
 Posted: Thu May 07, 2009 8:17 pm Post subject: |
 |
|
| open_coder wrote: | You know what I really wish you could do. Set restraints on packages. Like maybe I only want to update the kernel on my terms. Restrict its version number to the current one, then only update when I decide to. If a package requires a newer version of the software, prompt the user and have him make an exception, then the new version becomes the restriction.
Also, it would be nice to be able to set up something similar to click n run, where each website has an "Install" link for their application. The Install link doesn't just install the application, but adds repository information to sources.list. This would scale better than the current bureaucratic mindset for distribution repositories, where a package maintainer makes all the decisions instead of the user. It would also help, because each repository could have the libraries needed for the application.
I also think it would help to make the system distinguish between applications and libraries. Sure we have things like Adept_Installer, but it really just uses apt and apt doesn't know the difference. I should probably research how apt determines when a package is no longer required. Because I never understood the algorithm for that. How does it know not to uninstall applications simply because another application lists it as a prerequisite?
At this point, I'm just rambling. Really need to organize my thoughts. Sigh.
Alex |
Aptitude does a better job of cleaning up unused packages when you uninstall stuff.
_________________
Joe Mulloy | http://twitter.com/jdmulloy | Ron Paul in 2012! | 5-1-07 | Unban Playfool | Fire your "Too big to fail" bank http://moveyourmoney.info |
|
| Back to top |
|
 |
aidanjt
Veteran
Joined: 20 Feb 2005
Posts: 1096
Location: Rep. of Ireland
|
| Posted: Thu May 07, 2009 8:19 pm Post subject: |
|
|
| open_coder wrote: | | I should probably research how apt determines when a package is no longer required. Because I never understood the algorithm for that. |
That part is rather easy. Reverse dependency tree lookups.
_________________
| drizek wrote: | | Here in America, we are like a bunch of shit-slinging monkeys. |
|
|
| Back to top |
|
|
rrbrussell
n00b
Joined: 21 Jun 2004
Posts: 64
|
| Posted: Thu May 07, 2009 11:57 pm Post subject: |
|
|
| open_coder wrote: | The real problem with packaging is that if you are a developer and you are making software - good software - you want to distribute it to as many people as possible. Sure, Linux started out as a hobby operating system. But, for many of us, it's transcended that purpose into our full fledged desktop of choice. We are computer programmers and IT experts. The truth is, that as a computer science student with a strong interest in user interfaces and software engineering, I don't want to have to recompile my video card driver every time Ubuntu decides to do a kernel update. And I don't want to have to make packages for every distribution under the sun just to get my code out there.
Sure, I can distribute a tar.gz and have people
| Code: | $ ./configure
$ make
$ sudo make install |
but thats not good for some users. And I have never really liked compiling from source outside the package manager. Right now, If I were to distribute a new project, I'd make an Ubuntu repository and be done with it. And I would add a link on my website to the source tarball. The major problem with packaging is that Linux sucks for deployment.
As for the other problems, the truth is that solving them can only open up new possibilities for the Linux platform. The truth is that there are so many other interesting problems that we could be working on, and we are still floundering trying to fix problems that Windows and Mac have had fixed for years. It shouldn't be hard for users to get their equipment working on Linux. It shouldn't be hard to update drivers, and Xorg shouldn't fail whenever a kernel update is issued, simply because the user is using a proprietary driver. We shouldn't still be tackling these problems today, when they were noticed two or three years ago. But we are. And all this does is slow down developers from working on new and interesting problems and applications that could put Linux/FOSS light-years ahead of the competition in terms of commercial viability. It shouldn't be easier for a user to install an AIR application on Linux than it is to install a Linux application on Linux. But it is. And anybody who doesn't see this as a fault, is either fooling themselves or too elitist to want the Linux platform to become mainstream. The truth is that these problems need attention.
A while ago, I started what was supposed to be a series of posts on my blog about what Linux should be. Maybe I should go back and finish that. Or at the very least post about this topic, because I seem to have gotten a little too riled up. But I guess I just have conviction and want to see Linux achieve more that a meager one percent market share. More users means more contribution and a better system overall. And that's what I want.
Alex |
The video card driver interface is not a problem on Windows and Macs because there is only one kernel api to deal with until a new version of the entire os is released. And, yes most drivers will not work correctly until they are ported to the new api. The driver writers just have a few months or so to write the updated driver before the new os version comes out. Unlike Linux which release a new kernel api every few versions or so. |
|
| Back to top |
|
|
aidanjt
Veteran
Joined: 20 Feb 2005
Posts: 1096
Location: Rep. of Ireland
|
| Posted: Fri May 08, 2009 12:56 am Post subject: |
|
|
| rrbrussell wrote: | | The video card driver interface is not a problem on Windows and Macs because there is only one kernel api to deal with until a new version of the entire os is released. And, yes most drivers will not work correctly until they are ported to the new api. The driver writers just have a few months or so to write the updated driver before the new os version comes out. Unlike Linux which release a new kernel api every few versions or so. |
And semi-intentional ABI breakage practically every kernel release doesn't help either.
_________________
| drizek wrote: | | Here in America, we are like a bunch of shit-slinging monkeys. |
|
|
| 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
|
|
Off the Wall
