| mason | But agreed, it might be preferable. | 00:01 |
|---|---|---|
| fsmithred | mason, what branch did you use for the build? | 00:12 |
| fsmithred | I only have master here now | 00:12 |
| mason | just master | 00:17 |
| mason | isobuild@amd64-isobld:~/autoiso/installer-iso$ git branch | 00:17 |
| mason | * master | 00:17 |
| fsmithred | ok, I now have master and your branch | 00:18 |
| fsmithred | I can only find one place where debootstrap runs | 00:19 |
| mason | My branch was just a README update, already committed to master IIRC. | 00:19 |
| fsmithred | ah, ok | 00:19 |
| mason | Yeah. So it's implicit in something being pulled in, since it definitely didn't show the added flag. | 00:19 |
| fsmithred | I see --no-merged-usr in build.sh line 85 | 00:20 |
| mason | It should be invoking build-sudo.sh. But if I've misread it then I'll change that as well. | 00:20 |
| fsmithred | oh, I see that now. | 00:21 |
| golinux | mason: fsmithred: OT . . . had to send the images via my dyne account. | 00:22 |
| fsmithred | ok | 00:22 |
| mason | kk | 00:22 |
| mason | I see them now. | 00:23 |
| fsmithred | my usual method is to kick it around until it does what I want, so yeah, I would change the other instance even if it's not supposed to be doing anything. | 00:27 |
| mason | fair enough, and yeah - I'll look for the argument string you posted | 00:29 |
| rrq | note that the build scripts are for building the installer, and they play no role in running the installer | 01:24 |
| rrq | afair the installer installs base-installer, and its postinst runs a debootstrap | 01:24 |
| rrq | (the base-installer udeb is forked and despite actual reality I'm tagged as its maintaner; ... brings memories of times past :) | 01:27 |
| * onefang waits for rrq to say "TAG, your IT!" to one of us. | 01:28 | |
| rrq | ok.. please confer with LeePen about its maintenance :) | 01:29 |
| * rrq afk | 01:29 | |
| mason | rrq: thank you | 01:34 |
| golinux | We need to clone LeePen . . . | 01:45 |
| onefang | lol | 01:47 |
| Xenguy | git clone LeePen | 01:47 |
| onefang | Don't fork him, sounds painful. | 01:48 |
| Xenguy | Only the Devil loves a good fork | 01:48 |
| Xenguy | Oh wait | 01:49 |
| rrq | mason: as you know the installer has a preamble that is supposed to load up modules to the kernel so that the ISO is mountable... apparently your test machine requires some module that doesn't get loaded... | 02:07 |
| rrq | it's not the nvme (which is handled) but something with USB stack. | 02:09 |
| rrq | if you have a live ISO that works with that, then you can check which modules it has, and later rerun the failing boot to see which modules are loaded and which are available | 02:09 |
| rrq | there is a udev-udeb to possibly include into the preamble, but even that needs initial triggering by loading bus handling modules | 02:10 |
| rrq | also, note that the boot partition is limited to 32M so the preamble cannot be too large | 02:13 |
| rrq | (the 32M is an El Torrito limit enforced by xorriso when mastering the hybrid iso) | 02:14 |
| onefang | 32M, might make a speaking initrd hard. | 03:09 |
| rrq | that's why there's a preamble initrd, to find the ISO, to load the actual installer initrd and pivot to it | 03:11 |
| onefang | Ah, layers within layers. | 03:11 |
| onefang | The whole "speaking as early in the boot process as possible" goal has me wondering where to draw the line in the sand. | 03:13 |
| onefang | So "leave the 32M for the job of finding the ISO, put fancy stuff in the OTHER initrd" seems like a viable line in the sand. | 03:15 |
| rrq | or, one could scrap the notion of using a cd/dvd drive and limit to disk image booting... but the UEFI partition still has to be FAT | 03:18 |
| onefang | Yep, I realise this limitation is just for ISOs. | 03:19 |
| rrq | and with UEFI generally being 32-bit, the UEFI FAT needs to stay within the 4G addressing barrier | 03:20 |
| onefang | I still have two optical drives, one can even write DVDs. Last time I touched them was in February, last time I moved house. The disks will still be in one of the crates. | 03:20 |
| onefang | 4G is a lot easier to cram speach into than 32M. B-) | 03:21 |
| rrq | yeah; especially Scottish voices ;) | 03:21 |
| onefang | I'm half Scottish. | 03:22 |
| rrq | I'm less than that | 03:22 |
| onefang | Well technically it's 45% Scottish, 5% Irish, 50% German, all Aussie. OY! OY! OY! | 03:22 |
| onefang | espeak can handle a variety of voices, I've only listened to the default, no idea if it comes with more or I'd have to install them. | 03:23 |
| onefang | espeak-ng-data looks to be a lot of languages and such. | 03:26 |
| onefang | Ah and voices, lots of voices. | 03:26 |
| onefang | I guess AnxiousAndy had stage fright? | 03:27 |
| mason | rrq: I'll get a list of modules. Good call. | 03:37 |
| fsmithred | third item on the boot menu of my live-speech iso is Italian. | 03:38 |
| fsmithred | it can be changed by editing the boot line on the fly. | 03:38 |
| golinux | See y'all tomorrow . . . | 03:46 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!