Bug#925489: unblock: elogind/241.1-1+debian1

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

Bug#925489: unblock: elogind/241.1-1+debian1

Mark Hindley
Package: release.debian.org
Severity: normal
User: [hidden email]
Usertags: unblock

Please unblock package elogind.

Version 241.1-1+debian1 of src:elogind has been in unstable for 11 days with no
regressions reported.

I am aware that the debdiff against testing (239.3+20190131-1+debian1) is
significantly larger than you would normally want at this stage in the release
cycle. However, I believe it is warranted and ask for your consideration.

The benefits of migration centre around this version of elogind providing ABI
compatibility between libelogind0 and libsystemd0 (see #923244). This means that
packages compiled against libsystemd-dev headers work correctly with either
libsystemd0 or libelogind0 runtime libraries. This is particularly important for
policykit-1 authorization of many desktop functions (particularly restart, halt,
suspend and for pkexec which is essential for synaptic) on non-systemd init
installations.

It is also worth pointing out that as elogind is derived from systemd which is
alrelady at version 241-2 in testing. Therefore, much of the contained upstream
code is already in included in buster.

In addition, at the moment this is rather a niche area into which few 'regular'
users will dip their toes; systemd is not currently uninstallable by APT once it
is PID 1 and so elogind systems can only be installed from scratch and do not
happen without considerable knowledge and care.

Many thanks. I look forward to your thoughts.

Mark

-- System Information:
Debian Release: 8.11
Architecture: amd64 (x86_64)

Kernel: Linux 3.16.0-7-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8) (ignored: LC_ALL set to en_GB.UTF-8)
Shell: /bin/sh linked to /bin/dash
Init: sysvinit (via /sbin/init)

Reply | Threaded
Open this post in threaded view
|

Bug#925489: unblock: elogind/241.1-1+debian1

Adam Borowski-3
It's not a "niche" area.  Without this, any modern GUI desktop environments
are not installable with any pid 1 other than systemd.  That'd be a massive
regression that's certainly not acceptable (and it's caused by removal of a
systemd component with a hard dependency).

This regression had a plan, with coded and tested patches by January 2018
(with a refresh + retesting in June, then November, December).  In that
plan, policykit packages had alternatives built against elogind.  Yet
patches did not get applied.  Plan 2 was to dlopen() relevant libraries.

Then, once a required one-line patch was finally applied to src:systemd
(already in testing), policykit maintainers requested plan 3: to change
libelogind0 to implement 100% of libsystemd0's ABI, so the alternative
works at a different level:

Plan 1:
 ├─ policykit-systemd ─── libsystemd0 ─── systemd
 │
 └─ policykit-elogind ─── libelogind0 ─── elogind

Plan 2:
 └─ policykit ─── libsystemd0 ─┬─ systemd
                               │
                               └─ elogind

Plan 3:
 └─ policykit ─┬─ libsystemd0 ─── systemd
               │
               └─ libelogind0 ─── elogind

(Consumers of logind API other than policykit go mostly via libpam-*d).

With help of elogind's upstream, this request has been implemented.  Of
course, such a change has a pretty large debdiff.  Yet, with no reports of
regression within 12 days of testing in unstable, I believe it should be
relatively safe.

I'm not really happy with asking for an unblock for such a debdiff -- but
if this can't go in, we'd need to use one of the other plans.  Please say
if you consider that to be better.


Meow!
--
⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Did ya know that typing "test -j8" instead of "ctest -j8"
⢿⡄⠘⠷⠚⠋⠀ will make your testsuite pass much faster, and fix bugs?
⠈⠳⣄⠀⠀⠀⠀

Reply | Threaded
Open this post in threaded view
|

Bug#925489: unblock: elogind/241.1-1+debian1

Michael Biebl-3
Am 26.03.19 um 14:54 schrieb Adam Borowski:

> It's not a "niche" area.  Without this, any modern GUI desktop environments
> are not installable with any pid 1 other than systemd.  That'd be a massive
> regression that's certainly not acceptable (and it's caused by removal of a
> systemd component with a hard dependency).
>
> This regression had a plan, with coded and tested patches by January 2018
> (with a refresh + retesting in June, then November, December).  In that
> plan, policykit packages had alternatives built against elogind.  Yet
> patches did not get applied.  Plan 2 was to dlopen() relevant libraries.
>
Just to set the record straight here:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=923244

This bug report is from Mon, 25 Feb 2019 11:49:14 +0000

That this all is getting rushed on the last minute is not the fault of
the policykit-1 maintainers and I'm not amused that Adam tries to paint
it like that.



signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Bug#925489: unblock: elogind/241.1-1+debian1

Adam Borowski-3
On Tue, Mar 26, 2019 at 06:52:11PM +0100, Michael Biebl wrote:
> Just to set the record straight here:
>
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=923244
>
> This bug report is from Mon, 25 Feb 2019 11:49:14 +0000

That's the "plan 3" bug.  We had plan 1 over a year ago.

> That this all is getting rushed on the last minute is not the fault of
> the policykit-1 maintainers and I'm not amused that Adam tries to paint
> it like that.

I'm not amused either how long it takes to get any response to even a
single-line patch that had been discussed before.  But, the blame game is
counterproductive.  Could we please discuss how to fix present state?

It had been requested that the point of alternative gets moved.  That
request is now fulfilled, the code is uploaded, and has seen 12 days of
testing.  At this point, I kindly request your review.  Is the current
version of elogind, as packaged by Mark Hindley, good enough for you?


Meow!
--
⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Did ya know that typing "test -j8" instead of "ctest -j8"
⢿⡄⠘⠷⠚⠋⠀ will make your testsuite pass much faster, and fix bugs?
⠈⠳⣄⠀⠀⠀⠀

Reply | Threaded
Open this post in threaded view
|

Bug#925489: unblock: elogind/241.1-1+debian1

Michael Biebl-3
Am 26.03.19 um 19:45 schrieb Adam Borowski:
> On Tue, Mar 26, 2019 at 06:52:11PM +0100, Michael Biebl wrote:
>> Just to set the record straight here:
>>
>> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=923244
>>
>> This bug report is from Mon, 25 Feb 2019 11:49:14 +0000
>
> That's the "plan 3" bug.  We had plan 1 over a year ago.

I'm not aware of such a bug report. References please.

>> That this all is getting rushed on the last minute is not the fault of
>> the policykit-1 maintainers and I'm not amused that Adam tries to paint
>> it like that.
>
> I'm not amused either how long it takes to get any response to even a
> single-line patch that had been discussed before.  But, the blame game is
> counterproductive.

Why did you start it then?

> It had been requested that the point of alternative gets moved.  That
> request is now fulfilled, the code is uploaded, and has seen 12 days of
> testing.  At this point, I kindly request your review.  Is the current
> version of elogind, as packaged by Mark Hindley, good enough for you?

You honestly think with a behaviour like yours I'm motivated to review
your package and spend my time on it? The motivation/time I had dropped
basically to zero reading what you wrote.

If you had a carefully layed out plan, why do we have chaotic and rushed
bug reports like [1]. That doesn't look like a well thought out plan to me.

Anyway, I don't have any interest anymore to spend more time on this, so
don't expect any responses from me from now on.

Michael


[1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=922160#31
--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?


signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Bug#925489: marked as done (unblock: elogind/241.1-1+debian1)

Adam Borowski-3
In reply to this post by Mark Hindley
On Mon, Apr 08, 2019 at 02:57:03PM +0000, Debian Bug Tracking System wrote:

> > I am aware that the debdiff against testing (239.3+20190131-1+debian1) is
> > significantly larger than you would normally want at this stage in the release
> > cycle. However, I believe it is warranted and ask for your consideration.
>
> > The benefits of migration centre around this version of elogind providing ABI
> > compatibility between libelogind0 and libsystemd0 (see #923244). This means that
> > packages compiled against libsystemd-dev headers work correctly with either
> > libsystemd0 or libelogind0 runtime libraries. This is particularly important for
> > policykit-1 authorization of many desktop functions (particularly restart, halt,
> > suspend and for pkexec which is essential for synaptic) on non-systemd init
> > installations.

> This change is way to big to be unblocked.

I agree -- but this approach was what was requested for policykit-1.

What would you suggest instead?


--
⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Did ya know that typing "test -j8" instead of "ctest -j8"
⢿⡄⠘⠷⠚⠋⠀ will make your testsuite pass much faster, and fix bugs?
⠈⠳⣄⠀⠀⠀⠀