Active Topics
-
Is there a section to talk about Java ME? Apps, etc. (6)
to General by Kalatti - 2 days, 12 hrs ago -
Full linux distros on Sailfish OS (253)
to SailfishOS by qoh - 5 days, 6 hrs ago - more...
 |
|
2008-03-27
, 12:08
|
|
Posts: 5,478 |
Thanked: 5,222 times |
Joined on Jan 2006
@ St. Petersburg, FL
|
#32
|
Originally Posted by BoxOfSnoo
Too much? That's kind of the point of Red Pill, open up all the options for those who know how to use them.
Yes, red pill does way too much all in one fell swoop.
Originally Posted by BoxOfSnoo
I illustrated nothing. It all goes back Nokia's user friendliness philosophy.
You illustrated it here. What other distro uses categories in this way? Why not have some kind of category of "support libraries"?
Originally Posted by BoxOfSnoo
No, it means relevant to the user. The end user shouldn't have to concern themselves with libraries and support files, that's the entire point of Linux package management. The problem, again, goes back to packagers (as you can see with the issues stemming from the removal of Python-runtime from User/), if an application needs something, it should depend on it and the software should resolve the dependencies without user intervention.
What does "User/" really mean? "Safe/" ?
| The Following User Says Thank You to GeneralAntilles For This Useful Post: | ||
 |
|
2008-03-27
, 12:55
|
|
Posts: 2,535 |
Thanked: 6,681 times |
Joined on Mar 2008
@ UK
|
#33
|
GeneralAntilles is making lots of good points here (saving me having to post). As he's already alluded to, X-Fade is working on maemo.org with the additional of an autobuilder etc. but a lot of the blame for problems can be laid at the feet of the developers' poor packaging strategies. Any time someone creates their own repository, doesn't add a description, puts it in a new or incorrect category, exposes a library in "User/" or FUBARs their dependencies; they cause pain for their users.
Some of this is tooling, and there are efforts afoot to improve that (my own suggestions have been thrown into the pot) and QA process.
The end-goal is obvious, however: all decent applications in extras; all beta applications in extras-devel; no packages visible which an end user won't want to use directly; well-packaged apps with no dependencies outside of standard repos; no conflicts between libraries used by app A and app B.
The UI issues in Application Manager (such as rebuilding the view after clicking 'Cancel') are trivial in comparison to these larger, deeper, issues.
Some of this is tooling, and there are efforts afoot to improve that (my own suggestions have been thrown into the pot) and QA process.
The end-goal is obvious, however: all decent applications in extras; all beta applications in extras-devel; no packages visible which an end user won't want to use directly; well-packaged apps with no dependencies outside of standard repos; no conflicts between libraries used by app A and app B.
The UI issues in Application Manager (such as rebuilding the view after clicking 'Cancel') are trivial in comparison to these larger, deeper, issues.
__________________
Andrew Flegg -- mailto:andrew@bleb.org | http://www.bleb.org
Andrew Flegg -- mailto:andrew@bleb.org | http://www.bleb.org
| The Following 5 Users Say Thank You to Jaffa For This Useful Post: | ||
|
|
2008-03-27
, 14:21
|
|
Posts: 259 |
Thanked: 72 times |
Joined on Dec 2007
@ Halifax, NS
|
#34
|
Originally Posted by GeneralAntilles
Linux package management is done on many platforms, without a "nerf-squishy" mode and a "everything else" mode. Why is this such a roadblock? Why is it all or nothing?
No, it means relevant to the user. The end user shouldn't have to concern themselves with libraries and support files, that's the entire point of Linux package management. The problem, again, goes back to packagers (as you can see with the issues stemming from the removal of Python-runtime from User/), if an application needs something, it should depend on it and the software should resolve the dependencies without user intervention.
Why not break it down to groupings that are logically relevant to the user. "Applications you probably only care about installing", "Oddball utilities", "Support libraries", "Base functionality" and so on. Run through aptitude or YaST or Anaconda. Filter out a few things if necessary, but that list should be quite limited (base functionality of course)
I am sad to see all the newcomers to maemo told that the solution to getting a simple PIM is gpe and erminig - neither of which can be installed from blue pill mode through application manager, IIRC.
And yes, this can theoretically all be solved with proper packaging and a good extras repository. And yes, you can install many of these packages straight from deb files off the web, bypassing any crazy pill modes.
 |
|
2008-03-27
, 14:48
|
|
Posts: 11,700 |
Thanked: 10,045 times |
Joined on Jun 2006
@ North Texas, USA
|
#35
|
I think some are talking past each other here. I don't know as much as you guys but I will distill the issue into some basic facts:
1) We're talking a mobile Linux device here. Not desktops. Not servers. This is a whole new ball game and requires at least a slightly different approach than what works for landlocked devices.
2) Nokia is trying to make Linux truly mainstream. This is monumental. Despite years of trying, that hasn't happened yet in the desktop/server world.
3) Average users should not be faced with any visible complexity when installing or upgrading applications. Period. Conversely, advanced users should have every opportunity to get "under the hood" if they like. Anything less than the latter violates the spirit of Linux as I understand it.
Now... how do we get there? I think many good ideas have been presented. Next is to come to a consensus and, as a community, present an improvement model and plan to Nokia.
1) We're talking a mobile Linux device here. Not desktops. Not servers. This is a whole new ball game and requires at least a slightly different approach than what works for landlocked devices.
2) Nokia is trying to make Linux truly mainstream. This is monumental. Despite years of trying, that hasn't happened yet in the desktop/server world.
3) Average users should not be faced with any visible complexity when installing or upgrading applications. Period. Conversely, advanced users should have every opportunity to get "under the hood" if they like. Anything less than the latter violates the spirit of Linux as I understand it.
Now... how do we get there? I think many good ideas have been presented. Next is to come to a consensus and, as a community, present an improvement model and plan to Nokia.
__________________
Nokia Developer Champion
Different <> Wrong | Listen - Judgment = Progress | People + Trust = Success
My personal site: http://texrat.net
Nokia Developer Champion
Different <> Wrong | Listen - Judgment = Progress | People + Trust = Success
My personal site: http://texrat.net
|
|
2008-03-27
, 14:48
|
|
Posts: 5,478 |
Thanked: 5,222 times |
Joined on Jan 2006
@ St. Petersburg, FL
|
#36
|
Originally Posted by BoxOfSnoo
Because (as Jaffa mentioned above) if the user has to care about support libraries and other assorted miscellanea that is not directly relevant to their interestes, then things are broken. The packagers are to blame here, not the Application manager and its pill modes.
Why not break it down to groupings that are logically relevant to the user.
Originally Posted by BoxOfSnoo
You recall incorrectly. I know for certain GPE can be installed fine without Red Pill and erminig shouldn't need it either (though I haven't bothered to test this directly myself).
I am sad to see all the newcomers to maemo told that the solution to getting a simple PIM is gpe and erminig - neither of which can be installed from blue pill mode through application manager, IIRC.
But that's beside the point, as, even if they did, it still comes back to the packagers.
Originally Posted by BoxOfSnoo
Theoretical nothing. Poor packaging practices and a difficult-to-get-into Extras are the root of the problem. As Jaffa mentioned above, everything else is superfluous.
And yes, this can theoretically all be solved with proper packaging and a good extras repository. And yes, you can install many of these packages straight from deb files off the web, bypassing any crazy pill modes.
Originally Posted by Texrat
There's an easy one. Take everything that Jaffa has put forward on the issue over the past few months and years—that's exactly what needs to be done.
Now... how do we get there? I think many good ideas have been presented. Next is to come to a consensus and, as a community, present an improvement model and plan to Nokia.
Last edited by GeneralAntilles; 2008-03-27 at 14:50.
|
|
2008-03-27
, 14:53
|
|
Posts: 11,700 |
Thanked: 10,045 times |
Joined on Jun 2006
@ North Texas, USA
|
#37
|
Originally Posted by GeneralAntilles
OK, bud, all facetiousness aside-- is there a presentable plan? Has it been presented?
There's an easy one. Take everything that Jaffa has put forward on the issue over the past few months and years—that's exactly what needs to be done.
As one of your voluntary liaison folks here I will gladly help where I can there, but I'm thinking such a plan needs to be presented to Ari Jaaksi and Quim Gil at the least... either via their blogs as feedback or a direct email (or both). I can help with the latter.
__________________
Nokia Developer Champion
Different <> Wrong | Listen - Judgment = Progress | People + Trust = Success
My personal site: http://texrat.net
Nokia Developer Champion
Different <> Wrong | Listen - Judgment = Progress | People + Trust = Success
My personal site: http://texrat.net
|
|
2008-03-27
, 14:58
|
|
Posts: 5,478 |
Thanked: 5,222 times |
Joined on Jan 2006
@ St. Petersburg, FL
|
#38
|
Originally Posted by Texrat
http://www.maemopeople.org/index.php..._muliple_repos
OK, bud, all facetiousness aside-- is there a presentable plan? Has it been presented?
|
|
2008-03-27
, 15:10
|
|
Posts: 11,700 |
Thanked: 10,045 times |
Joined on Jun 2006
@ North Texas, USA
|
#39
|
Cool, that's a start. I think I had seen mention of that proposal recently but had forgotten the details. Unfortunately, maemopeople does not appear to be heavily visited (one reason I quit adding to my blog there)... and was there feedback from maemo-developers (I'm not on the list anymore)? Andrew?
EDIT: I agree with Pupnik's response to Andrew's proposal, and wrote a comment of my own there.
Last edited by Texrat; 2008-03-27 at 15:23.
EDIT: I agree with Pupnik's response to Andrew's proposal, and wrote a comment of my own there.
__________________
Nokia Developer Champion
Different <> Wrong | Listen - Judgment = Progress | People + Trust = Success
My personal site: http://texrat.net
Nokia Developer Champion
Different <> Wrong | Listen - Judgment = Progress | People + Trust = Success
My personal site: http://texrat.net
Last edited by Texrat; 2008-03-27 at 15:23.
|
|
2008-03-27
, 16:15
|
|
Posts: 5,478 |
Thanked: 5,222 times |
Joined on Jan 2006
@ St. Petersburg, FL
|
#40
|
Originally Posted by Texrat
This is an important point. What I really should have said was "streamlined access", not "easier access". Access to Extras shouldn't really be "easier" (i.e. available to every Tom, Dick, and Harry with a recompiled text editor) rather, the application/submission process needs to be clarified and streamlined—developers need a clear, straightforward way to get their applications into the repository. That, while being a bit selective, doesn't get in the way of itself.
EDIT: I agree with Pupnik's response to Andrew's proposal, and wrote a comment of my own there.
I think Extras-devel is an important part of the solution here, too. Access to that should be easier.
Side note: Isn't it interesting how front page stories seem to elicit either no response or a very large one, but never anything in between?
Last edited by GeneralAntilles; 2008-03-27 at 16:27.
What we need is quality control..
Last edited by traveller604; 2008-03-27 at 11:49.