SoC boot project weekly summary week #4/14

[Carlos Villegas]

Hi, this is the report for the 4th week of the debian boot project. That 
is, from last thursday until yesterday (wednesday).

1. According to what was mentioned during the last report: a daily blog 
has been written with the aim of informing about the daily progress and 
getting some feedback. Besides, the first deliverable was made available 
for revision last sunday evening.

2. The remaining distribution to test was installed (sid) and a 
bootchart publised.

3. The bootcharts from sarge and etch were repeated with a stopwatch to 
"validate" the results from bootchart. The results were published in the 
blog and showed that from 15 to 20 seconds were not accounted for.

4. The bootchart script (/sbin/bootchartd) was modified to stop logging 
after KDE finishes to load. The results from bootchart started to be 
very similar to the ones obtained with a stopwatch.

5. A section Resources was added to the webpage to share bootcharts, 
script modifications, installatation logs among other things. Besides, 
contact information was added to both, the webpage and the blog.

6. The first two hotspots more easy to implement were tested in sid: 
loading hwclock in the background and not loading depmod. With hwclock 
in the background the boot reduced by 2 seconds. The bootchart is 
presented in the Resources area of the webpage. Depmod was already taken 
off the boot in Sid by the script maintainer and to test the difference 
an old script was used to load depmod. The boot time increased by 2 
seconds when depmod was loaded. This results just comfirm some of the 
results from Margarita Manterola in DebConf6.

In the project's timetable (in webpage) it is said that this week, the 
Ubuntu and FreeBSD boot systems were going to be analysed. This was done 
to some extent in the first deliverable. Nevertheless, I think that a 
SUSE system should be further investigated instead. Another option could 
be Mandriva or both. Your comments upon this and if other systems should 
be tested would be greatly appreciated.

On the other hand, perhaps I should concentrate on implementing the 
hotspots and the lintian/LSB compliance checking.

For the next week I originally propose to:
+ substitute /bin/sh with dash.
+ work with a script to check scripts for lintian/LSB compliance.
+ test parallel execution (with startpar) with the current scripts.
+ test parallel execution (with startpar) and reoganizing scripts (with 
a program like insserv).
+ setup the network in the background

Cheers,

carlos



-- 

Carlos Villegas
----------------------------------------
Hamilton Institute
NUIM, Maynooth, Co. Kildare
Ireland
Tel: +353 1 708 6458
Fax: +353 1 708 6269
www.hamilton.ie/cemacs