Re: Preparing for 6.3 Release
On 6/11/07, Alexander E. Patrakov <[EMAIL PROTECTED]> wrote: > Dan Nicholson wrote: > > > "If you're using the toram option, you must remount the CD to use the > > package tarballs in /lfs-sources." > > > > I don't know if that's all that's correct. I didn't investigate when I > > was using toram yesterday. > > Now we have an additional variable: the -nosrc CD, which doesn't contain > sources, but must contain the same /etc/issue file, because root.ext2 is the > same. If we don't find a wording that covers all three cases, I will have to > generate /etc/issue during boot. Anyway, the /lfs-sources symlink should be > removed during boot if it is dangling, so a script (or part of initramfs > /init) is needed anyway. I think that generated would be the best since I can't think of any words that aren't convoluted. What about something like this in intramfs/init.in (wording to be improved, of course): toram=0 [EMAIL PROTECTED]@ # This would be 0 or 1 as substituted by the Makefile ... case ${toram}${nosrc} in 00) cat >> /.root/etc/issue << "EOF" # or however you would access it from the initramfs Find the sources in /lfs-sources EOF ;; *1) ;; # No sources, so don't bother saying anything? 10) cat >> /.root/etc/issue << "EOF" Since you are using the toram option, you must remount the CD to find the sources for the LFS packages. EOF ;; esac > >>> 3) Undocumented initramfs boot options. Maybe we should document things > >>> like > >>> "noapic pci=noacpi", too. > > > > That'd be nice. Is it possible to add more documentation pages to > > isolinux? We could add options2.msg and add "F2 options2.msg" to > > isolinux.cfg, right? > > Yes. I don't know all the options, but I'll play with this a bit. > > The only package I saw that I thought was borderline was > > xf86-video-intel-2.0.0. You and I both read the xorg mailing list and > > know that it's not stable yet. If you're curious, the G965 is > > supported just fine in xf86-video-i180-1.7.4. I've been using it for > > months. It doesn't have all the fancy functionality of intel-2.0.0 > > (but neither does the livecd since it's using libXrandr-1.1.2), but > > it's working fine. What was the reason for updating to 2.0.0? > > Ability to set modes that the monitor prefers (e.g., native resolution on > LCD monitors). According to forums, 915resolution is not too reliable (e.g., > Xorg and the monitor disagree upon the synchronization frequencies). > However, now I am not sure whether xf86-video-intel-2.0.0 is really better. > Anyway, it supports more hardware (new: i965GM, found in laptops). I forgot about the 915resolution package. Things have always just worked for me. Well, until it breaks something, I say keep it in. I don't think anyone's going to try any xrandr'ing from a livecd, anyway. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Preparing for 6.3 Release
Dan Nicholson wrote: > "If you're using the toram option, you must remount the CD to use the > package tarballs in /lfs-sources." > > I don't know if that's all that's correct. I didn't investigate when I > was using toram yesterday. Now we have an additional variable: the -nosrc CD, which doesn't contain sources, but must contain the same /etc/issue file, because root.ext2 is the same. If we don't find a wording that covers all three cases, I will have to generate /etc/issue during boot. Anyway, the /lfs-sources symlink should be removed during boot if it is dangling, so a script (or part of initramfs /init) is needed anyway. >>> 3) Undocumented initramfs boot options. Maybe we should document things like >>> "noapic pci=noacpi", too. > > That'd be nice. Is it possible to add more documentation pages to > isolinux? We could add options2.msg and add "F2 options2.msg" to > isolinux.cfg, right? Yes. > The only package I saw that I thought was borderline was > xf86-video-intel-2.0.0. You and I both read the xorg mailing list and > know that it's not stable yet. If you're curious, the G965 is > supported just fine in xf86-video-i180-1.7.4. I've been using it for > months. It doesn't have all the fancy functionality of intel-2.0.0 > (but neither does the livecd since it's using libXrandr-1.1.2), but > it's working fine. What was the reason for updating to 2.0.0? Ability to set modes that the monitor prefers (e.g., native resolution on LCD monitors). According to forums, 915resolution is not too reliable (e.g., Xorg and the monitor disagree upon the synchronization frequencies). However, now I am not sure whether xf86-video-intel-2.0.0 is really better. Anyway, it supports more hardware (new: i965GM, found in laptops). >> 6) In Japanese locales, fontconfig chooses the wrong default font (AR PL New >> Sung, while it should use Kochi fonts) > > Is there any way to change this in fonts.conf? It looks like you're > adding configuration for firefly, but not for kochi. Ignore this. It was a panic from my side after seeing non-antialiased fonts, not a real bug. Kochi fonts are known to fontconfig out of the box, but AR PL New Sung isn't. -- Alexander E. Patrakov -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Preparing for 6.3 Release
I want to just say out front what my opinion of the LiveCD is before I respond to individual points. The most important task of the livecd is to provide a host with a known working kernel and toolchain to allow people to build *LFS. Next, it should provide tools that allow the *LFS support channels to be reached. Everything else is a bonus and should not determine whether a livecd release is made. IMO. On 6/11/07, Alexander E. Patrakov <[EMAIL PROTECTED]> wrote: > I wrote: > > 1) Unclear status of wireless network support. The CD doesn't contain > > wpa-supplicant and doesn't have firmware for most wireless network cards > > (even though we qualify as ISV and thus can redistribute ipw firmware). > > However, I can't test this because I don't have a wireless card. I'd say don't bother. We don't even have a working wireless setup in BLFS. I don't think it's asking people too much to plug in an ethernet cable if they want to use the livecd. This would be a great feature, but it shouldn't stop a release from being made. > > 2) Incorrect contents of /etc/issue with the "toram" boot option. The > > sources are intentionally not copied to RAM (so that one can use "toram" on > > a 512M box), but the message still says they are available in /lfs-sources. > > Since it is easy to add generation of the no-sources ISO, we may want to > > cover this case, too. "If you're using the toram option, you must remount the CD to use the package tarballs in /lfs-sources." I don't know if that's all that's correct. I didn't investigate when I was using toram yesterday. > > 3) Undocumented initramfs boot options. Maybe we should document things like > > "noapic pci=noacpi", too. That'd be nice. Is it possible to add more documentation pages to isolinux? We could add options2.msg and add "F2 options2.msg" to isolinux.cfg, right? > > 4) It is necessary to add back the reiser4 and loop-aes kernel patches once > > the kernel version in LFS stabilizes. I suppose. I don't consider that high priority at all. If you're depending on functionality that's not in the mainline kernel, then it's a huge bonus if anyone supports you out of the box. > > 5) I have updated the packages "blindly" too frequently. Many of them are > > simply untested. Someone should go through _all_ of them and attempt to make > > sure that the basic functionality is there. Well, LFS-6.3 will put a test to that. It's not a problem if bugs are found and a -2 release needs to be made. It seemed to work fine for me, although I didn't do too much. Like I said above, the main two tasks for me are the toolchain and the support tools. I hope the toolchain is fine or LFS is in trouble, too. I played with seamonkey and pidgin briefly. My network was setup fine out of the box, and I was able to jump right onto the internet and irc. So, from my end, it was working fine. The only package I saw that I thought was borderline was xf86-video-intel-2.0.0. You and I both read the xorg mailing list and know that it's not stable yet. If you're curious, the G965 is supported just fine in xf86-video-i180-1.7.4. I've been using it for months. It doesn't have all the fancy functionality of intel-2.0.0 (but neither does the livecd since it's using libXrandr-1.1.2), but it's working fine. What was the reason for updating to 2.0.0? > 6) In Japanese locales, fontconfig chooses the wrong default font (AR PL New > Sung, while it should use Kochi fonts) Is there any way to change this in fonts.conf? It looks like you're adding configuration for firefly, but not for kochi. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Preparing for 6.3 Release
I wrote: > 6) In Japanese locales, fontconfig chooses the wrong default font (AR PL New > Sung, while it should use Kochi fonts) Sorry for the wrong report. I confused fonts because I didn't know that Kochi fonts included embedded bitmaps, too. Since the font was not antialiased, I wrongly thought that it must be AR PL New Sung. The terminal does pass the U+6D77 test (see http://people.debian.org/~kubota/U6D77.png), so there is no bug. -- Alexander E. Patrakov -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Preparing for 6.3 Release
I wrote: > 1) Unclear status of wireless network support. The CD doesn't contain > wpa-supplicant and doesn't have firmware for most wireless network cards > (even though we qualify as ISV and thus can redistribute ipw firmware). > However, I can't test this because I don't have a wireless card. > > 2) Incorrect contents of /etc/issue with the "toram" boot option. The > sources are intentionally not copied to RAM (so that one can use "toram" on > a 512M box), but the message still says they are available in /lfs-sources. > Since it is easy to add generation of the no-sources ISO, we may want to > cover this case, too. > > 3) Undocumented initramfs boot options. Maybe we should document things like > "noapic pci=noacpi", too. > > 4) It is necessary to add back the reiser4 and loop-aes kernel patches once > the kernel version in LFS stabilizes. > > 5) I have updated the packages "blindly" too frequently. Many of them are > simply untested. Someone should go through _all_ of them and attempt to make > sure that the basic functionality is there. 6) In Japanese locales, fontconfig chooses the wrong default font (AR PL New Sung, while it should use Kochi fonts) -- Alexander E. Patrakov -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page