Hello,
On Mon, Jan 1, 2018 at 7:19 PM, Oliver Smith
<ollieparanoid@bitmessage.ch> wrote:
> Sorry for not reporting it yet.>> In our case the bug was triggered by a "dbus-nox11" package we created - probably because it ended up providing the same "so:..." dependency as the official "dbus" package. It worked fine, and at some point (maybe after an apk upgrade?) the package didn't work anymore - after we removed it, it worked again.>> If it helps with debugging/fixing this issue, I could try to reproduce the way the bug affected us, and provide apk's installed DB.
I think I can create a reproducer. I will do it after I finish cooking dinner.
William
---
Unsubscribe: alpine-devel+unsubscribe@lists.alpinelinux.org
Help: alpine-devel+help@lists.alpinelinux.org
---
Hello,
On Mon, Jan 1, 2018 at 8:26 PM, William Pitcock
<nenolod@dereferenced.org> wrote:
> Hello,>> On Mon, Jan 1, 2018 at 7:19 PM, Oliver Smith> <ollieparanoid@bitmessage.ch> wrote:>> Sorry for not reporting it yet.>>>> In our case the bug was triggered by a "dbus-nox11" package we created - probably because it ended up providing the same "so:..." dependency as the official "dbus" package. It worked fine, and at some point (maybe after an apk upgrade?) the package didn't work anymore - after we removed it, it worked again.>>>> If it helps with debugging/fixing this issue, I could try to reproduce the way the bug affected us, and provide apk's installed DB.>> I think I can create a reproducer. I will do it after I finish cooking dinner.
Unfortunately based on this data, I was unable to create a testcase.
Adding two packages resulted in an unsatisfiable constraints problem,
but it explained what the problem was (that it wanted the user to
choose a provider).
I did add those testcases to apk-tools git if you would like to look at them...
Could somebody reproduce the failure with apk-tools git compiled with
debug printing enabled?
Thanks in advance!
William
---
Unsubscribe: alpine-devel+unsubscribe@lists.alpinelinux.org
Help: alpine-devel+help@lists.alpinelinux.org
---
Sorry for not reporting it yet.
In our case the bug was triggered by a "dbus-nox11" package we created - probably because it ended up providing the same "so:..." dependency as the official "dbus" package. It worked fine, and at some point (maybe after an apk upgrade?) the package didn't work anymore - after we removed it, it worked again.
If it helps with debugging/fixing this issue, I could try to reproduce the way the bug affected us, and provide apk's installed DB.
Thanks,
Oliver
Tmp File:
> You can use this script to find out what's the package that gives the error.> > > > #!/bin/sh> for i in $(cat /etc/apk/world)> do> printf "%s: " "$i" >> file 2>&1> sudo apk add $i --no-progress >> file 2>&1> done> grep "ERROR" file> > > > This error was noticed before by postmarketOS.> https://github.com/postmarketOS/pmbootstrap/issues/952> They didn't find solutions yet.> > > >> Sent: Sunday, December 31, 2017 at 12:00 PM>> From: "Natanael Copa" <ncopa@alpinelinux.org>>> To: "Alpine Development" <alpine-devel@lists.alpinelinux.org>>> Subject: [alpine-devel] Huh? Error reporter did not find the broken constraints.>>>> Hi,>>>> after I upgraded apk to 2.9.0 on my work desktop I get the following error:>>>> ncopa-desktop:~$ sudo apk upgrade -U -a>> fetch http://dl-cdn.alpinelinux.org/alpine/edge/main/x86_64/APKINDEX.tar.gz>> fetch http://dl-cdn.alpinelinux.org/alpine/edge/community/x86_64/APKINDEX.tar.gz>> fetch http://dl-cdn.alpinelinux.org/alpine/edge/testing/x86_64/APKINDEX.tar.gz>> ERROR: unsatisfiable constraints:>> Huh? Error reporter did not find the broken constraints.>>>>>> I have tried to delete some of the --virtual package I had installed but it didn't solve anything.>>>> I idea how to troubleshoot this?>>>> I have attached my /etc/apk/world>>>> -nc> > > ---> Unsubscribe: alpine-devel+unsubscribe@lists.alpinelinux.org> Help: alpine-devel+help@lists.alpinelinux.org> ---> >
---
Unsubscribe: alpine-devel+unsubscribe@lists.alpinelinux.org
Help: alpine-devel+help@lists.alpinelinux.org
---
On Mon, 1 Jan 2018 22:49:07 -0600
William Pitcock <nenolod@dereferenced.org> wrote:
> Hello,> > On Mon, Jan 1, 2018 at 8:26 PM, William Pitcock> <nenolod@dereferenced.org> wrote:> > Hello,> >> > On Mon, Jan 1, 2018 at 7:19 PM, Oliver Smith> > <ollieparanoid@bitmessage.ch> wrote: > >> Sorry for not reporting it yet.> >>> >> In our case the bug was triggered by a "dbus-nox11" package we created - probably because it ended up providing the same "so:..." dependency as the official "dbus" package. It worked fine, and at some point (maybe after an apk upgrade?) the package didn't work anymore - after we removed it, it worked again.> >>> >> If it helps with debugging/fixing this issue, I could try to reproduce the way the bug affected us, and provide apk's installed DB. > >> > I think I can create a reproducer. I will do it after I finish cooking dinner. > > Unfortunately based on this data, I was unable to create a testcase.> > Adding two packages resulted in an unsatisfiable constraints problem,> but it explained what the problem was (that it wanted the user to> choose a provider).> I did add those testcases to apk-tools git if you would like to look at them...> > Could somebody reproduce the failure with apk-tools git compiled with> debug printing enabled?> > Thanks in advance!
I have attached output from debug print.
Thanks!
-nc
On Tue, 2 Jan 2018 10:55:40 +0100
Natanael Copa <ncopa@alpinelinux.org> wrote:
> On Mon, 1 Jan 2018 22:49:07 -0600> William Pitcock <nenolod@dereferenced.org> wrote:> > > Hello,> > > > On Mon, Jan 1, 2018 at 8:26 PM, William Pitcock> > <nenolod@dereferenced.org> wrote: > > > Hello,> > >> > > On Mon, Jan 1, 2018 at 7:19 PM, Oliver Smith> > > <ollieparanoid@bitmessage.ch> wrote: > > >> Sorry for not reporting it yet.> > >>> > >> In our case the bug was triggered by a "dbus-nox11" package we created - probably because it ended up providing the same "so:..." dependency as the official "dbus" package. It worked fine, and at some point (maybe after an apk upgrade?) the package didn't work anymore - after we removed it, it worked again.> > >>> > >> If it helps with debugging/fixing this issue, I could try to reproduce the way the bug affected us, and provide apk's installed DB. > > >> > > I think I can create a reproducer. I will do it after I finish cooking dinner. > > > > Unfortunately based on this data, I was unable to create a testcase.> > > > Adding two packages resulted in an unsatisfiable constraints problem,> > but it explained what the problem was (that it wanted the user to> > choose a provider).> > I did add those testcases to apk-tools git if you would like to look at them...> > > > Could somebody reproduce the failure with apk-tools git compiled with> > debug printing enabled?> > > > Thanks in advance! > > I have attached output from debug print.
The problem is nvme-cli.
ERROR PKG: nvme-cli: conflict: same name provided
It seems that it comes from a local repository /home/ncopa/packages/testing where I have an older version of nvme-cli:
$ tar -O -zxf /home/ncopa/packages/testing/x86_64/APKINDEX.tar.gz APKINDEX | grep -w -A 15 P:nvme-cli
P:nvme-cli
V:1.1-r0
A:x86_64
S:68384
I:192512
T:NVM-Express user space tooling for Linux
U:https://github.com/linux-nvme/nvme-cli
L:GPL
o:nvme-cli
m:Florian Heigl <florian.heigl@gmail.com>
t:1482791353
c:3f5fcf3c09c78baca6ac5bde33ca8210aab75251
D:so:libc.musl-x86_64.so.1
p:nvme-cli
C:Q135cWo/MbmAnhinSUHunVd/nBCcM=
--
...
So it looks like I have a package named nvme-cli in community repo and
an old version of nvme-cli in my local testing repo, which has a
provides=nvme-cli for some reason.
Some things that fixes my machine:
- apk del nvme-cli
- commenting out /home/ncopa/packages/testing in /etc/apk/repositories
> > Thanks!> > -nc
---
Unsubscribe: alpine-devel+unsubscribe@lists.alpinelinux.org
Help: alpine-devel+help@lists.alpinelinux.org
---
Looks like the trigger for the issue is: provides="$pkgname".
This was also in the "dbus-nox11" case and was earlier reported here:
https://bugs.alpinelinux.org/issues/7586
Thanks,
Oliver
Natanael Copa:
> On Tue, 2 Jan 2018 10:55:40 +0100> Natanael Copa <ncopa@alpinelinux.org> wrote:> >> On Mon, 1 Jan 2018 22:49:07 -0600>> William Pitcock <nenolod@dereferenced.org> wrote:>>>>> Hello,>>>>>> On Mon, Jan 1, 2018 at 8:26 PM, William Pitcock>>> <nenolod@dereferenced.org> wrote: >>>> Hello,>>>>>>>> On Mon, Jan 1, 2018 at 7:19 PM, Oliver Smith>>>> <ollieparanoid@bitmessage.ch> wrote: >>>>> Sorry for not reporting it yet.>>>>>>>>>> In our case the bug was triggered by a "dbus-nox11" package we created - probably because it ended up providing the same "so:..." dependency as the official "dbus" package. It worked fine, and at some point (maybe after an apk upgrade?) the package didn't work anymore - after we removed it, it worked again.>>>>>>>>>> If it helps with debugging/fixing this issue, I could try to reproduce the way the bug affected us, and provide apk's installed DB. >>>>>>>> I think I can create a reproducer. I will do it after I finish cooking dinner. >>>>>> Unfortunately based on this data, I was unable to create a testcase.>>>>>> Adding two packages resulted in an unsatisfiable constraints problem,>>> but it explained what the problem was (that it wanted the user to>>> choose a provider).>>> I did add those testcases to apk-tools git if you would like to look at them...>>>>>> Could somebody reproduce the failure with apk-tools git compiled with>>> debug printing enabled?>>>>>> Thanks in advance! >>>> I have attached output from debug print.> > The problem is nvme-cli.> > ERROR PKG: nvme-cli: conflict: same name provided> > It seems that it comes from a local repository /home/ncopa/packages/testing where I have an older version of nvme-cli:> > $ tar -O -zxf /home/ncopa/packages/testing/x86_64/APKINDEX.tar.gz APKINDEX | grep -w -A 15 P:nvme-cli> P:nvme-cli> V:1.1-r0> A:x86_64> S:68384> I:192512> T:NVM-Express user space tooling for Linux> U:https://github.com/linux-nvme/nvme-cli> L:GPL> o:nvme-cli> m:Florian Heigl <florian.heigl@gmail.com>> t:1482791353> c:3f5fcf3c09c78baca6ac5bde33ca8210aab75251> D:so:libc.musl-x86_64.so.1> p:nvme-cli> > C:Q135cWo/MbmAnhinSUHunVd/nBCcM=> --> ...> > So it looks like I have a package named nvme-cli in community repo and> an old version of nvme-cli in my local testing repo, which has a> provides=nvme-cli for some reason.> > Some things that fixes my machine:> - apk del nvme-cli> - commenting out /home/ncopa/packages/testing in /etc/apk/repositories> > >>>> Thanks!>>>> -nc> > > > ---> Unsubscribe: alpine-devel+unsubscribe@lists.alpinelinux.org> Help: alpine-devel+help@lists.alpinelinux.org> ---> >
---
Unsubscribe: alpine-devel+unsubscribe@lists.alpinelinux.org
Help: alpine-devel+help@lists.alpinelinux.org
---