corp186 | cjwatson: so the udebs listed in the pkg-lists directory? | 00:07 |
---|---|---|
corp186 | cjwatson: also, is there any way to "fork" the installation | 00:07 |
corp186 | instead of going through the normal install process, say I enable a certain item and then the installation needs to go through a different set of udebs | 00:08 |
corp186 | or maybe a different set, and then get back to the main path of udebs at a certain point | 00:08 |
cjwatson | corp186: those listed in pkg-lists are what're used to build the initrd | 00:12 |
cjwatson | corp186: there are various tricks that can be used to do that kind of thing, although maybe not exactly as stated - it very much depends on the situation | 00:12 |
cjwatson | Frans' paper explains a lot of it, IIRC | 00:13 |
corp186 | so I guess I'm missing something | 00:14 |
corp186 | where in the build process is the main udebs (not the ones in the initrd) defined? | 00:14 |
cjwatson | it isn't | 00:14 |
cjwatson | the build process is responsible for building the initrd | 00:14 |
cjwatson | the rest is determined at run-time | 00:14 |
corp186 | how is it determined? | 00:14 |
cjwatson | that *is* in Frans' paper - section "anna and retrievers" | 00:15 |
corp186 | I think I'm looking at a different paper: http://d-i.alioth.debian.org/doc/internals/ | 00:16 |
corp186 | I don't see that section | 00:16 |
corp186 | where is the paper you are referring to? | 00:16 |
cjwatson | hmm, I'm looking at something checked into d-i svn, no idea where it is on the web | 00:18 |
corp186 | cjwatson: do you have a repo and a path? | 00:18 |
cjwatson | anyway, any udeb with priority: standard, important, or required is automatically installed | 00:18 |
cjwatson | in expert mode you can select others | 00:18 |
cjwatson | dependencies of those udebs are selected as well | 00:18 |
cjwatson | so it's all driven by the Packages file | 00:19 |
corp186 | cjwatson: can you point me to the doc? | 00:20 |
corp186 | I probably have the repo checked out somewhere :) | 00:21 |
corp186 | I just need a path, or even a filename | 00:21 |
cjwatson | I've just explained the relevant bit anyway | 00:21 |
corp186 | yeah, but I'm interested in reading more | 00:21 |
cjwatson | installer/doc/talks/d-i_internals/2006/d-i_internals.tex in d-i | 00:21 |
cjwatson | I can't vouch for how helpful it is - I was misremembering it as the source of http://d-i.alioth.debian.org/doc/internals/ | 00:22 |
cjwatson | so actually I'd sort of rather not confuse you with potentially out-of-date information | 00:22 |
cjwatson | but up to you | 00:22 |
corp186 | well, I can read code, so even if it's wrong it'll probably start me on the right path | 00:22 |
corp186 | thanks | 00:22 |
cjwatson | if you can read code, packages/anna/ is more useful | 00:23 |
corp186 | I thought that was just the udeb installer? | 00:23 |
cjwatson | yes, but that's what you're asking about | 00:24 |
corp186 | sort of, but either way, I have some more pointers now | 00:24 |
corp186 | thanks again for all the info | 00:24 |
cjwatson | anna is the udeb analogue of apt; udpkg is the udeb analogue of dpkg. But they aren't exact analogues - anna has more policy in it than apt does | 00:25 |
corp186 | cjwatson: the amd64 packages list for debian-installer has both grub-installer and lilo-installer, though with grub coming first in installer item ordering | 04:21 |
corp186 | how is the decision made between the two? | 04:22 |
=== mpt_ is now known as mpt | ||
cjwatson | corp186: first in installer-menu-item order wins, unless the isinstallable file in the package's control area says otherwise (which, nowadays, it normally doesn't) | 12:14 |
kim0 | Hi folks ... Any idea why in early_command, is not expanding "/dev/sd[a-d]" ? Which shell does early_command run in | 13:02 |
cjwatson | sh (a.k.a. dash) but that is not your problem since it understands such globs | 13:03 |
cjwatson | you clearly have some kind of quoting problem | 13:03 |
kim0 | d-i preseed/early_command string for device in `ls /dev/sd[b-z]`; do dd if=/dev/zero of=$device bs=512 count=1 ; done | 13:04 |
kim0 | I'm getting | 13:05 |
kim0 | log-output: ls: /dev/sd[b-z]: No such file or directory | 13:05 |
kim0 | cjwatson: The exact command worked when I typed it directly into ALT+F2 shell | 13:06 |
cjwatson | well, your symptoms match what happens if there are simply no devices matching that pattern right now | 13:07 |
cjwatson | beyond that I don't know, and am not going to investigate fully today since I'm on holiday | 13:08 |
kim0 | cjwatson: I see .. | 13:08 |
kim0 | the device exists though | 13:08 |
kim0 | but thanks for the help | 13:09 |
kim0 | cjwatson: would something like bash -c "voodoo here" help :) | 13:13 |
cjwatson | doubt it, try 'set -x' at the start to see what it's doing, or an extra 'ls /dev; ' or something - the usual debugging techniques you might think of | 13:14 |
cjwatson | besides, you don't have bash | 13:14 |
kim0 | set -x in the early command itself | 13:14 |
cjwatson | yes | 13:14 |
kim0 | ls /dev would go into the logs right | 13:14 |
cjwatson | I guess. Be creative, I'm not going to walk you through it all today :) | 13:15 |
kim0 | hehe :D | 13:15 |
kim0 | thanks man | 13:15 |
corp186 | cjwatson: so if grub-installer "wins", how is lilo-installer skipped? | 13:15 |
=== robbiew_ is now known as robbiew | ||
corp186 | cjwatson: it appears to me in the main-menu.c satisfy_virtual function that for providers of dependencies, the highest prio is taken, or the lowest menu item if prios are equal | 15:26 |
corp186 | is this where the decision is made? | 15:26 |
rek2 | hi everyone I am having problem with parted seen my partitions,etc please refer to my problem for pictures here: http://ubuntuforums.org/showthread.php?p=8378529#post8378529 | 18:30 |
rek2 | I can see them with fdisk -l but the install will just not see it so I can't continue | 18:31 |
rek2 | :-( | 18:31 |
rek2 | brb | 18:40 |
=== robbiew is now known as robbiew_ |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!