[systemd-devel] setting up to allow separate udev and systemd builds

Kok, Auke-jan H auke-jan.h.kok at intel.com
Thu Jun 14 15:35:09 PDT 2012


On Thu, Jun 14, 2012 at 3:15 PM, William Hubbs <w.d.hubbs at gmail.com> wrote:
> On Thu, Jun 14, 2012 at 02:05:01PM -0700, Kok, Auke-jan H wrote:
>> On Thu, Jun 14, 2012 at 1:40 PM, William Hubbs <w.d.hubbs at gmail.com> wrote:
>> > That is exactly my point. There are people in these distros who see
>> > systemd as a kitchen sink they don't want. However, udev is more like
>> > coreutils, which they need.
>>
>> or pick mdev
>
> Would there be regressions if we do that though? I'm under the
> impression that mdev is a very stripped down version of udev.
>
>> >> Just because you're a source distro, doesn't mean that you should be bleeding
>> >> edge, and force upstream to potentially break the build for everyone already.
>> >
>> > Again you make my point for me. There are folks who see systemd as
>> > bleeding edge and are not comfortable running it, but they still need
>> > udev.
>>
>> and can stay on udev-182.
>
> unless they need something in 185, like a bug fix or something that has
> changed since then.
>
>> >  Define how this forces upstream to break the build. The default state
>> >  of my patches is to leave things as they are unless you mess with the
>> >  two added configure switches. I do not see how this breaks anything.
>> >
>> >> I'm no fan of these patches, and I'm less of a fan of them appearing *now*. Keep
>> >> them in the refrigerator for 3 months, then post them.
>> >
>> > Why do you hate these patches so much? Like I said, I made sure not to
>> > change the default state, so if you don't mess with the configure
>> > switches you don't change anything.
>>
>> I don't hate them. I think the timing is inappropriate. I don't like
>> multiple Makefile.am's though - that's bound to be problematic further
>> on.
>
> What's so inappropriate about the timing, or to put it another way, why
> does timing matter so much?
>
> There was a patch proposed on the hotplug list which put the fixes in
> Makefile.am directly. I spoke with Kay on irc and he stated that he didn't
> like that patch.
>
>> Nothing prevents you from staying on udev-182 for a few more
>> weeks/months. You're not missing that much.
>
> That depends. are there fixes that any users need in the newer versions,
> or are the only differences new features?
>
> Either way, All I'm advocating for is  supporting *building* udev
> standalone as well as *running* udev standalone.
>
> I've done the work. If there is a problem with the work I have done, I
> would like to know what it is so I can fix it.
>
> If there isn't a problem with it, why can't it be supported?

It's not up to me - I'm not a maintainer.

My concerns are:

1) Timing. systemd+udevd just got merged.  We had a huge build system
change. Let's not change it drastically again entirely now.

2) Splitting up of Makefile.am. Sounds like a disaster waiting to
happen and a reversal of the whole idea of merging the code in the
first place.

3) Urgency: none to low. Source distro's would be perfectly fine
building and maintaining the few security and bug fixes needed. AFAIK
the merge only removed features in udevd.

4) Can be maintained out of tree for now: Nothing prevents gentoo from
keeping this patch out of tree for their purposes.

5) Not in the best interest of systemd as a whole, this patch only
increases fragmentation. I'd rather see someone maintain the
non-merged udev source with fixes only.

Cheers,


Auke


More information about the systemd-devel mailing list