Hi, I missed the dev meeting (by about 30 minutes I guess), but I read the logs. (Thanks to Ikkebr for posting them at http://www.hashphp.com/144 -- perhaps they could be copied to somewhere more permanent; are the others in the wiki?). It was great to see how development is more focused, with the clear list of bugs, etc.
I didn't go through each entry in the bugtracker but here are some general comments: > 2.6.24.x kernel: This kernel opps while mounting squashfs > filesystems. Obviously, this is a problem for the livecd. Last I > talked with Lucas, he had a few patches to try but I don't know the > current status. > > OpenOffice (238): I've built 2.3.1_bin-r2 in ChrootCompile. This > solved the freedesktop entry problem but openoffice won't load. I > tracked this down to the lack of symlinks in bin (to ../programs) for > openoffice.org2.3.bin, soffice.bin, pagein. Oddly, the recipe makes > the soffice.bin symlink. Also, the soffice link is made twice. > > mplayer (260): need an updated skel (maybe enchanced?) for a > ~/.mplayer/subfont.ttf symlink. Recommendations for the font to use? In my machine here I have a link /Programs/MPlayer/Current/Shared/mplayer/subfont.ttf -> /Files/Fonts/TrueType/Vera.ttf Perhaps we could even copy the font there, to simplify things. > Manpages in share (255): Already talked to you. Need a new Scripts > package with updated MANPATH (in /S/S/bashrc and zshrc). This is temp > fix until we rebuild packages (015). > > Xorg (282): Newest revision fixes xcb problems. I have built and > tested this on livecd (using skype). My package isn't signed. I need > to add my key to Scripts package (above) or have someone else > build/sign package. > > Squashfs on AM2 (283): Unable to reproduce on my AM2 machine. Is the > bug a gross failure or subtle (read corruption)? I'm not sure, but I think it fails to boot. I have access to an AM2 machine, still need to test this. > Boot in VirtualBox (284): I don't have Virtualbox installed (and no > gcc 3.x to build). If someone has a Virtualbox package I'd appreciate > it. Would rather not introduce gcc3 into my system. > > DBus, no machince-id (247): This is a general problem with Installer > not handling PostInstall scripts. Lucas came up with a quick patch to > ProfileInstall that runs PostInstall but it hasn't been tested. There > are other issues wrt PostInstall but I'll cover those below. > > Install not copying /F/Fonts (246): Tracked down to incomplete kernel > args (needed maxloop=32) when using AM2 workaround. Can this be > closed? Yes; I updated http://www.gobolinux.org/?page=known_issues_014 accordingly and closed the bug. > Some 014-rc bugs that might need attention. These should be closed or > moved to 014. > > Broken Links after install (199): RemoveBroken needs to be run in Installer? > > Sudo overwrites files (204): This is a longstanding problems. There > is a suggested change to the sudo recipe in the bug tracker. Has the > recipe been updated? > > InstallPackage error msgs (205): Notes imply it is fixed. Should > this be closed? Like I wrote in the bug, I think so. Jonas, what do you say? > In a nutshell, we need new Scripts, EnchancedSkel, Installer releases > (changes may not be in svn). Updated Sudo, OpenOffice > recipes/packages. Updated Xorg package. I'll take ownership of the > OpenOffice changes. They seem to be LiveCD specific. > > Now, to talk about PostInstall. I'm going to be a bit verbose because > I won't be there tomorrow to respond to question regarding what I > mean. The PostInstall script is run to create an EFFECT (I'll use > this a lot) on the system (create id/keys, set ownership based on > created user/group, etc). This effect should be distinct for each > install. The LiveCD is a type of install. It should be affected by > the PostInstall scripts. Whether this happens during ISO generation > or boot is debateable but probably minor. > > So the LiveCD needs PostInstall effects. These effects should not be > mirrored to the HD during install. During install, PostInstall needs > to be run to generate the distinct effects for that system. > > I'm worried that Installer is not sophisticated enough to prevent > transferring the LiveCD effects to the HD. There are only 4 > PostInstall scripts on the LiveCD, so they can be special cased for > 014.b but a plan for general use would be great for beyond. Similar > concerns apply for R/Requirements as well. I believe the effects applied to the LiveCD environment all go into Settings directories and the user's home directory, while the Installer does not use these directories, using each package's Defaults/Settings instead. IIRC, André was the one who worked on this, so he may be able to confirm. In this case, we shouldn't have a problem running PostInstalls (for example) on LiveCD boot for the LiveCD environment, and then running them again for the installed system. > On IRC, Jonas and I discussed upping KDE to 3.5.9. At that point, I > started to think that 014.b could be more than a bugfix release. > Keeping the stage 1 core, updating the stage 2 desktop and stage 3 > apps. With the timeline, I'm sure that is not feasible. I'd like more > details wrt pressing CD. Is this for a magazine? Is it worth the > rush? Especially since 014 is a little dated. Given that KDE 3.5.9 itself is also a bugfix release, I'm kinda favorable to that idea as well. > For the ISO, I'd prefer to receive binary packages. Hopefully those > providing packages would have more detailed knowledge of the > application and could test the package thoroughly. This should catch > bugs before they get to an ISO rc. To implement this, I could set up > a webdav dropbox on calica.com. Maybe ftp as well. Another option is > SVN. To build packages, you'd just need ChrootCompile on 014. For > 015, we could have an image that runs under lguest. Please discuss > this on IRC and provide feedback. As mentioned in the IRC metting, simpler is better. I'd prefer using ftp. > SVN migration. This requires changes in BuildLiveCD tools. Hopefully > these changes won't take too long to implement or cause other > regressions. Yes. I can look at that later in the week unless someone beats me to it. -- Hisham _______________________________________________ gobolinux-devel mailing list gobolinux-devel@lists.gobolinux.org http://lists.gobolinux.org/mailman/listinfo/gobolinux-devel
