Fedora Account System
Red Hat Associate
Red Hat Customer
Description of problem: The current version of pandoc in rawhide fails to install with the following error (e.g., on x86_64): Error: Problem: conflicting requests - nothing provides libHSth-abstraction-0.2.6.0-KdJUlqAwMfUIgxZVMgLLVd-ghc8.2.2.so()(64bit) needed by pandoc-2.0.6-5.fc29.x86_64 Version-Release number of selected component (if applicable): pandoc-2.0.6-5.fc29 How reproducible: Always Additional info: I encountered this when trying to build bup, which has a BuildRequires on pandoc, in Rawhide today: https://koji.fedoraproject.org/koji/taskinfo?taskID=28491999
I've got: - nothing provides libHSth-abstraction-0.2.6.0-KdJUlqAwMfUIgxZVMgLLVd-ghc8.2.2.so()(64bit) needed by pandoc-citeproc-0.12.2.5-4.fc29.x86_64 Also, when I try to bump build pandoc-citeproc: Error: Problem 1: conflicting requests - nothing provides ghc-devel(th-abstraction-0.2.6.0-KdJUlqAwMfUIgxZVMgLLVd) needed by ghc-aeson-devel-1.2.3.0-3.fc29.x86_64 Problem 2: conflicting requests - nothing provides libHSth-abstraction-0.2.6.0-KdJUlqAwMfUIgxZVMgLLVd-ghc8.2.2.so()(64bit) needed by ghc-aeson-pretty-devel-0.8.5-3.fc29.x86_64 Problem 3: conflicting requests - nothing provides ghc-devel(http-client-0.5.7.1-4kyJ4nMDdL3BONGiqBslly) needed by ghc-pandoc-devel-2.0.6-5.fc29.x86_64 Problem 4: conflicting requests - nothing provides libHSth-abstraction-0.2.6.0-KdJUlqAwMfUIgxZVMgLLVd-ghc8.2.2.so()(64bit) needed by ghc-yaml-devel-0.8.25.1-3.fc29.x86_64 Problem 5: package ghc-pandoc-types-devel-1.17.3-3.fc29.x86_64 requires ghc-pandoc-types(x86-64) = 1.17.3-3.fc29, but none of the providers can be installed - conflicting requests - nothing provides libHSth-abstraction-0.2.6.0-KdJUlqAwMfUIgxZVMgLLVd-ghc8.2.2.so()(64bit) needed by ghc-pandoc-types-1.17.3-3.fc29.x86_64 This seems to be rather large scaled :(
This seems to need just a chain rebuild of all the deps. Resolving and trying in mock.
Got it will rebuild (in this order): ghc-aeson ghc-aeson-pretty ghc-doctemplates ghc-http-client ghc-asn1-types ghc-tasty-quickcheck ghc-asn1-encoding ghc-cryptonite ghc-pem ghc-asn1-parse ghc-x509 ghc-x509-store ghc-x509-validation ghc-tls ghc-x509-system ghc-connection ghc-http-client-tls ghc-yaml ghc-pandoc-types ghc-texmath ghc-skylighting ghc-doctemplates pandoc pandoc-citeproc
https://koji.fedoraproject.org/koji/taskinfo?taskID=28539556
ghc-aeson-pretty failed in Koji, yet it works in my mock :(
So does ghc-asn1-encoding, same error: https://koji.fedoraproject.org/koji/taskinfo?taskID=28541879 Linking dist/build/tests-asn1-encoding/tests-asn1-encoding ... /tmp/ghc8eea_0/ghc_2.o(.gnu.build.attributes..text.startup+0x18): error: relocation refers to local symbol "" [11], which is defined in a discarded section section group signature: "(null)" /tmp/ghc8eea_0/ghc_2.o(.gnu.build.attributes..text.startup+0x20): error: relocation refers to local symbol "" [2], which is defined in a discarded section section group signature: "(null)" collect2: error: ld returned 1 exit status `gcc' failed in phase `Linker'. (Exit code: 1)
Switching back to new, as this won't be that easy.
ghc-aeson-pretty ghc-asn1-encoding ghc-yaml ghc-doctemplates ghc-texmath all failed with the same error. architectures with skipped tests succeeded. if no proper fix is found, a quick rebuild with tests skipped might fix this.
%undefine _annotated_build also seems to solve this.
I'm running builds with `%undefine _annotated_build` now. We can always rebuild without that workaround if whatever ld or annobin bug is causing the build failures gets fixed, but for now, it'd be nice to have the broken deps fixed.
Thanks for working on this, guys: in the meantime I have been working on a bigger Haskell packages update for F29.
(In reply to Adam Williamson from comment #10) > I'm running builds with `%undefine _annotated_build` now. We can always > rebuild without that workaround if whatever ld or annobin bug is causing the > build failures gets fixed, but for now, it'd be nice to have the broken deps > fixed. Has it been reported?
Not sure if there's a specific bug for the remaining issue, no. Miro might have filed one?
Not in bugzilla, but in devel ML. Nick Clifton was later CCed in the thread. https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/message/KH4QQBW5PJBL6HMOOAJWHTF5W2MMPUAX/
Hi Guys, This problem should be fixed in binutils-2.31.1-3.fc29 Cheers Nick
Indeed, ghc-aeson-pretty now builds fine without "%undefine _annotated_build". Should we revert the commits and leave it be, or bump the release and do another rebuild? I'm not sure how important it is to have the builds annotated.
Removing it would be nice, since otherwise it's likely that the temporary workaround will persist forever.
We'll remove it for sure, yet I don't know whether we should just revert the commit or bump release and rebuild again as well.
Don't see any harm in rebuilding them. I added it to I think four or five packages, I'll rebuild those today if no-one else gets to it.
For the record, Miro re-enabled annotated build and rebuilt all the packages we had temporarily disabled it on - so everything here should be addressed now. Thanks Miro!
Thanks a lot Miro and Adam, this really helped me a lot! Since I am a bit swamped at the moment.
It's all self-interest really, I need pandoc working or else I can't cut releases of any of my projects :P
👍
(In reply to Adam Williamson from comment #22) > It's all self-interest really, I need pandoc working or else I can't cut > releases of any of my projects :P Same for me, I was just blocked by this :)