Active Topics

 


Reply
Thread Tools
Posts: 3,319 | Thanked: 5,610 times | Joined on Aug 2008 @ Finland
#101
Originally Posted by volt View Post
Read the 2nd quote again.
I have installed way more than 20 packages (some quite hefty at that) so the general approach does work (except for the FS accounting bug, but that's not /opt related). My guess is that some packages (or, maybe their dependencies !) were not as optified as the OP thought. A list of the packages (with versions) would be helpful to track down the culprit(s), fill bug reports, etc.
 

The Following 4 Users Say Thank You to attila77 For This Useful Post:
GunnerzMate's Avatar
Posts: 70 | Thanked: 11 times | Joined on Oct 2009
#102
This is my greatest concern about the N900. I have a feeling it will work fine the first 2 weeks and then little by little it will slow down just like symbian...
 
w00t's Avatar
Posts: 1,055 | Thanked: 4,107 times | Joined on Oct 2009 @ Norway
#103
Originally Posted by GunnerzMate View Post
This is my greatest concern about the N900. I have a feeling it will work fine the first 2 weeks and then little by little it will slow down just like symbian...
Symbian and Maemo are two totally different horses. There's no direct comparison that can be made between the two (except that everyone seems to hate the first.. hehe).

Nor is this thread in any way concerned with a slow down so much as a problem with development software on a pre-release device. Don't worry
__________________
i'm a Qt expert and former Jolla sailor (forever sailing, in spirit).
if you like, read more about me.
if you find me entertaining, or useful, thank me. if you don't, then tell me why.
 

The Following 4 Users Say Thank You to w00t For This Useful Post:
Posts: 1,224 | Thanked: 1,763 times | Joined on Jul 2007
#104
Originally Posted by Rushmore View Post
I actually changed my post and got rid of the "spin" and everything else, since spin was a poor word choice. Adding 256mb more flash would have added $50 to $75 more dollars to the device. I think that common yield for flash is 256mb, so that is what EVERYONE buys- Nokia, HTC, Motorolla, etc. for phone devices.
Your information is obviously incorrect.

The Pandora uses OMAP3430 with 512MB nand POP, and only costs 330$. A year ago they announced an upgrade from 128MB ram and 256MB nand to 256M+512M, claiming the upgrade costs 10$. For Nokia and for the nand only it is probably more like 1$ for the upgrade.

About nand chip capacity: 8MB SD cards are common and cheap. Open one and count the chips inside - there are no 32 nand chips inside.
 

The Following User Says Thank You to Matan For This Useful Post:
qgil's Avatar
Posts: 3,105 | Thanked: 11,088 times | Joined on Jul 2007 @ Mountain View (CA, USA)
#105
Originally Posted by attila77 View Post
A list of the packages (with versions) would be helpful to track down the culprit(s), fill bug reports, etc.
Indeed. Has anybody published a list of installed applications when hitting the rootfs limit? It feels like we have shared 101 discussion posts and yet nobody has been able to point to concrete examples af apps non optified that are causing trouble to testers.

fwiw yesterday I started going through http://maemo.org/packages/repository...xtras-testing/ and filing blockers to apps installing more than 500kb without optification, according to http://wiki.maemo.org/Extras-testing...root_partition

You are encouraged to help. I believe there is even a way to check this without a device, checking the source code? Instructions welcome if anybody knows how to.

Last edited by qgil; 2009-10-25 at 06:59.
 

The Following User Says Thank You to qgil For This Useful Post:
Posts: 279 | Thanked: 293 times | Joined on Oct 2009 @ Italy
#106
Originally Posted by qgil View Post
Indeed. Has anybody published a list of installed applications when hitting the rootfs limit? It feels like we have shared 101 discussion posts and yet nobody has been able to point to concrete examples af apps non optified that are causing trouble to testers.

fwiw yesterday I started going through http://maemo.org/packages/repository...xtras-testing/ and filing blockers to apps installing more than 500kb without optification, according to http://wiki.maemo.org/Extras-testing...root_partition

You are encouraged to help. I believe there is even a way to check this without a device, checking the source code? Instructions welcome if anybody knows how to.
Wait till i get my n900 shipped and i'll test the world :P
__________________
LPSMagic
Donate
 
Posts: 388 | Thanked: 1,340 times | Joined on Nov 2007 @ Finland
#107
Originally Posted by qgil View Post
fwiw yesterday I started going through http://maemo.org/packages/repository...xtras-testing/ and filing blockers to apps installing more than 500kb without optification, according to http://wiki.maemo.org/Extras-testing...root_partition

You are encouraged to help. I believe there is even a way to check this without a device, checking the source code? Instructions welcome if anybody knows how to.
I tried to check few packages randomly but they already are
  • checked and commented
  • are below 500KB (from the wiki) with their / usage

It would be nice to know which packages have already been checked so there isn't duplicate work to be done.

My method to check the package: Download DEB package. Open it with the GNOME's Archive Manager. Open data.tar.gz by clicking it. Go down by hierachy until you find out how much combined other directories than /opt are using (sum the size columns).
 
Posts: 2,802 | Thanked: 4,491 times | Joined on Nov 2007
#108
I posted a simple script to check (on-device) an installed package's rootfs usage here, feel free to add any other methods that come to mind.
 

The Following 4 Users Say Thank You to lma For This Useful Post:
Posts: 7 | Thanked: 34 times | Joined on Oct 2009 @ London, UK
#109
Originally Posted by lma View Post
I posted a simple script to check (on-device) an installed package's rootfs usage here, feel free to add any other methods that come to mind.
I got a bit irritated with this situation last night and came up with a temporary solution. Take a look at http://talk.maemo.org/showthread.php...407#post357407

There are a few big caveats, but if you're careful it will handle things safely until we have more optified packages.
__________________
--
for((P=10**8,Q=P/100,X=320*Q/(`tput cols`-1),Y=210*Q/`tput lines`,y=-105*Q,v=-2\
20*Q,x=v;y<105*Q;x=v,y+=Y));do for((;x<P;a=b=i=k=c=0,x+=X));do for((;a*a+b*b<4*\
P*P&&i++<99;a=((c=a)*a-b*b)/P+x,b=2*c*b/P+y));do :;done;(((j=(i<99?i%16:0)+30)>\
37?k=1,j-=8:0));echo -ne "\E[$k;$j"mE;done;echo -e \\E[0m;done # Charles Cooke
 

The Following 2 Users Say Thank You to ccooke For This Useful Post:
ewan's Avatar
Posts: 445 | Thanked: 572 times | Joined on Oct 2009 @ Oxford
#110
This is something of a work in progress, but the following bit of bash will take the name of a .deb package on the commandline, and total up the sizes of files in /usr, /opt, and elsewhere according to the index of the embedded data.tar.gz. It doesn't need the package to be installed, doesn't use any Debian specific packaging tools, and doesn't create any temporary files.

It shouldn't be too difficult to automatically run it on every .deb in a copy of the repository, but that's next on the todo list.

Code:
#!/bin/bash

deb=$1

OIFS=$IFS
IFS=$'\n'

opt_total=0
home_total=0
other_total=0


for line in $(ar p ${deb} data.tar.gz | tar -tzvf -) ; do {
        IFS=$OIFS
        array=(${line})

        case $(echo ${array[5]} | cut -b 1-5) in
        ./usr ) usr_total=$(( ${usr_total} + ${array[2]})) ;;
        ./opt ) opt_total=$(( ${opt_total} + ${array[2]})) ;;
        * ) other_total=$(( ${other_total} + ${array[2]})) ;;
        esac

        } ;
done

echo -e "Usr\tOpt\tOther"
echo -e "${usr_total}\t${opt_total}\t${other_total}"

IFS=$OIFS
 

The Following 4 Users Say Thank You to ewan For This Useful Post:
Reply


 
Forum Jump


All times are GMT. The time now is 20:53.