Hi all, I have little bit investigate about state of libattr and differences to libudev. Also programs that may or need depend on libudev. It seems to me questionable what to implement and if it is reasonable to do.
In debian package system (which I count as good reference) there are these packages that depend on libudev: connman - intel connection manager daemon dmsetup, libdevmapper - DM tools dvdstyler - DVD authoring system gvfs lvm2 ofono - mobile telephony stack system-config-printer udisks xdmx - distributed multihead x server libvirt (in dragonfly) pulseaudio (in dragonfly) vlc (in dragonfly) xorg (in dragonfly) >From the libudev vs libattr function difference, there are ~30 functions missing, some of them may be easy to implement with just makro or some easy function. Libdevattr seems to missing udev_queue at all. I still need to dive in to udev deeply to investigate for what is this use for. But Alex is right that a lot of missing functions are somehow relevant to sysfs interaction. It is questionable how much of them need to be implemented in dragonfly so far. And how much of them are really used. Robert. V Fri, 23 Mar 2012 08:26:35 +0100 Robert David <robert.david.pub...@gmail.com> napsáno: > Hi Alex, > > thanks for tips, I think I will concentrate on the libdevattr stuff, > even it seems to be a bigger size problem. I will look at the problem > more deeply. > > > Robert. > > > V Wed, 21 Mar 2012 16:50:13 +0000 > Alex Hornung <ahorn...@gmail.com> wrote: > > > Hi Robert, > > > > On 21/03/12 00:18, Robert David wrote: > > > I'm new to DragonFly, but I like the ideas you provide for > > > enhancements. There seems to be many tasks I'm interested, but I > > > dont know I can handle a lot of them, because I dont know > > > DragonFlyBSD (*BSD) kernel so much just now. > > > > if you want to learn about DragonFly, I suggest you just grab an > > ISO, install it (a VM will do) and just browse around. If you are > > really keen you could submit a patch or two for some of the bugs on > > the bugtracker. > > > > > > > As there are more ideas to improve in your list, maybe you as a > > > developers has some priority list what is more and what less > > > important. > > > > It is not so much about what we feel is important, but about your > > proposal. If your proposal is great, you got good chances to get a > > slot even for a project someone might consider less relevant. > > > > > > > From the things that seems reasonable for me and my experiences > > > are: > > > - Improve compatibility of libdevattr with Linux' libudev > > > > I'd certainly welcome that - but beware that there's more work to be > > done here than might appear at a first glance. In particular > > regarding sysfs - I'm not sure how much software benefits from > > libudev without explicit use of sysfs. If most also use sysfs, this > > project might evolve rapidly into a partial sysfs implementation > > (maybe even as a userland fs using puffs/fuse and an expanded > > libdevattr?) > > > > > > > - Port FreeBSD's USB stack to DragonFly > > > > While a lot of us feel that it would be great to have FreeBSD's USB > > stack, I'm not quite convinced it's a good gsoc project. > > > > > > > - Adapt pkgsrc to create a package system with dependency > > > independence. > > > > A proper package system for pkgsrc would certainly be great. If you > > are interested in this, I suggest you discuss this further on the > > mailing lists and dive into how pkgsrc works by simply using > > DragonFly for a while. > > > > > > HTH, > > Alex >
