Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

GHC 7.10 #378

Closed
snoyberg opened this issue Dec 24, 2014 · 121 comments
Closed

GHC 7.10 #378

snoyberg opened this issue Dec 24, 2014 · 121 comments

Comments

@snoyberg
Copy link
Contributor

Tested with RC1. I'm going to put in some hacks to allow testing the builds even with the restrictive upper bounds, but here are the bounds issues right now (pinging @hvr who asked for this information):

Loading settings for: Nightly
Writing build plan to: nightly-2014-12-24.yaml
Checking build plan
stackage:
Cabal-1.22.0.0 depended on by:

base-4.8.0.0 depended on by:

bytestring-0.10.6.0 depended on by:

  • bytestring-builder-0.10.4.0.1 (>=0.9 && < 0.10.1 || ==0.10.4.*). No maintainer. @lpsmith

deepseq-1.4.0.0 depended on by:

ghc-7.10.0.20141222 depended on by:

template-haskell-2.10.0.0 depended on by:

time-1.5.0.1 depended on by:

transformers-0.4.2.0 depended on by:

@ekmett
Copy link
Contributor

ekmett commented Dec 24, 2014

lens will be shipping a new version momentarily. lens HEAD currently builds with GHC 7.10, but it's a rats' nest of import warnings we want to fix before we call it an official release.

Note: the current version of lens will build, but using makeLenses will fail at compile time for downstream packages.

@ekmett
Copy link
Contributor

ekmett commented Dec 24, 2014

Fixing lens up on hackage will probably have to happen after Christmas though for me to dig in deeper.

I'll have limited availability the next couple of days, and @glguy is also traveling for family stuff on his side, so both of us are going to be out of commission for the next few days.

@Twinside
Copy link
Contributor

JuicyPixels-3.2.1 allows the last deepseq, it should be usable in stackage when diagrams-cairo will be updated on hackage (bound is ok on HEAD).

@jgm
Copy link
Contributor

jgm commented Dec 24, 2014

Try building pandoc with -f-old-locale -- this should force it to use
time 1.5. The trouble is, I wasn't able to test this because I couldn't
compile the dependencies with time 1.5.

+++ Vincent [Dec 24 14 03:31 ]:

JuicyPixels-3.2.1 allows the last deepseq, it should be usable in stackage when diagrams-cairo will be updated on hackage (bound is ok on HEAD).


Reply to this email directly or view it on GitHub:
#378 (comment)

@aslatter
Copy link
Contributor

The package uuid-1.3.8 builds on GHC 7.10.1 RC 1. The tests do not build yet - I'll let dependencies settle down before I look at that. Let me know if that's not an appreciated approach :-)

The package quickcheck-instances-0.3.10 also builds on GHC 7.10.1 RC 1.

@hvr
Copy link

hvr commented Dec 24, 2014

@aslatter fwiw, cabal test works just fine for uuid-1.3.8 w/ GHC 7.10.1 RC1 for me...

PS: Here's the package versions cabal picks for me:

$ cabal install  --enable-tests --dry
Resolving dependencies...
In order, the following would be installed (use -v for more details):
HUnit-1.2.5.2
ansi-terminal-0.6.2.1
ansi-wl-pprint-0.6.7.1
byteable-0.1.1
cryptohash-0.11.6
extensible-exceptions-0.1.1.4
hostname-1.0
mtl-2.2.1
network-info-0.2.0.5
old-locale-1.0.0.7
primitive-0.5.4.0
random-1.1
regex-base-0.93.2
regex-posix-0.95.2
text-1.2.0.3
hashable-1.2.3.1
tf-random-0.5
QuickCheck-2.7.6
xml-1.3.13
test-framework-0.8.1.0
test-framework-hunit-0.3.0.1
test-framework-quickcheck2-0.3.0.3
uuid-1.3.8

@hvr
Copy link

hvr commented Dec 24, 2014

@snoyberg why is scientific-0.3.3.3 (>=4.3 && < 4.8). mentioned above? That one seems wrong to me, unless I don't understand what the list above means exactly...

@geraldus
Copy link
Contributor

@hvr it have unresolved dependency on base, i.e. <4.8, while latter going to be exact 4.8.0.0.

@hvr
Copy link

hvr commented Dec 24, 2014

@geraldus I'm still confused, if you look at http://hackage.haskell.org/package/scientific-0.3.3.3/scientific.cabal it says

  build-depends:       base        >= 4.3   && < 4.9
                     , ghc-prim
                     , deepseq     >= 1.3   && < 1.5
                     , text        >= 0.8   && < 1.3
                     , hashable    >= 1.1.2 && < 1.3
                     , array       >= 0.1   && < 0.6

and cabal install scientific works just fine w/ GHC 7.10 (scientific-0.3.3.3 was in fact released specifically to add support for GHC 7.10)

@sol
Copy link
Contributor

sol commented Dec 24, 2014

For hspec-expectations I just uploaded a new version to Hackage which works.

For Doctest we have sol/doctest#94. If anybody wants to help with it, that would be much appreciated.

@geraldus
Copy link
Contributor

@hvr, oh, now I see.

@geraldus
Copy link
Contributor

By the way, does anyone have prepared docker image with GHC 7.10 RC1?

@snoyberg
Copy link
Contributor Author

@hvr it's the test suite that has a restrictive upper bound on base.

On Thu, Dec 25, 2014, 1:53 AM Артур Файзрахманов [email protected]
wrote:

By the way, does anyone have prepared docker image with GHC 7.10 RC1?


Reply to this email directly or view it on GitHub
#378 (comment).

@snoyberg
Copy link
Contributor Author

@jgm Thanks for that explanation, when it gets closer I can turn off that flag.

@byorgey Do you have an ETA for a new release of diagrams-cairo that allows the newest JuicyPixels?

@hadronized
Copy link
Contributor

Hm, I don’t have GHC 7.10 installed yet. I’ll try that after partying :) Enjoy Christmas Eve folks :)

@byorgey
Copy link
Contributor

byorgey commented Dec 25, 2014

Not sure, I've been somewhat out of touch with diagrams development
recently. Hopefully soon.
On Dec 24, 2014 11:47 PM, "Michael Snoyman" [email protected]
wrote:

@jgm https://github.com/jgm Thanks for that explanation, when it gets
closer I can turn off that flag.

@byorgey https://github.com/byorgey Do you have an ETA for a new
release of diagrams-cairo that allows the newest JuicyPixels?


Reply to this email directly or view it on GitHub
#378 (comment).

@bergey
Copy link

bergey commented Dec 25, 2014

diagrams-cairo-1.2.0.5 now allows JuicyPixels-3.2.*. I haven't tried to build with GHC-7.10 or HEAD.

@snoyberg
Copy link
Contributor Author

Thanks @bergey!

@geraldus
Copy link
Contributor

If someone need to have GHC-7.10-RC1 up and running quickly, here is my docker image: https://registry.hub.docker.com/u/geraldus/ghc-7.10-rc1/

@geraldus
Copy link
Contributor

I was able to build sodium with relexed upper bound. However there was mtl-2.2.1 dependency installed before.
mtl-2.1.3.1 which is listed as a constraint in 7.10-RC1's cabal.config file rejects to build because of transformers (as already mentioned in head), though 2.2.1 installs successfully.

@geraldus
Copy link
Contributor

Pinging @snoyberg
Michael, can you say something about mtl version? I'm confused little bit.

@snoyberg
Copy link
Contributor Author

I haven't released a 7.10 cabal.config file. Where are you seeing it?

@geraldus
Copy link
Contributor

@snoyberg
Copy link
Contributor Author

That's a config file for building the Stackage executable. The Stackage executable itself is still built with GHC 7.8 since not all of its dependencies build with GHC 7.10.

@geraldus
Copy link
Contributor

All I want to know is what version of mtl should I use to build sodium with GHC 7.10-RC1, does 2.2.1 acceptable?
mtl-2.1.3.1 is listed in issue description, but there is mtl-2.2.1 on Hackage, and it accepts new version of transformers. That's why I'm confused.

@snoyberg
Copy link
Contributor Author

You'll want to use mtl-2.2.1 on GHC 7.10.

@giogadi
Copy link
Contributor

giogadi commented Dec 28, 2014

Can anyone recommend a clean way to test my package on GHC 7.10 on OSX?

@d12frosted
Copy link
Contributor

@snoyberg
I'll test my package in few days.

@geraldus
Copy link
Contributor

Ok, sodium-0.11.0.3 should be fixed now!

@d12frosted
Copy link
Contributor

@giogadi
The same here. Can't find apple-darwin release on downloads page. Have I missed something?

@bergmark
Copy link
Member

snaplet-fay fixed with new revision

@chrisdone
Copy link
Member

Fixed pdfinfo in 1.5.3.

@bergmark
Copy link
Member

wl-pprint-1.2

@jgm
Copy link
Contributor

jgm commented Apr 15, 2015

pandoc-1.13.2.1 builds on ghc 7.10.1.

@RyanGlScott
Copy link
Contributor

Fixed wai-middleware-static in 0.7.0.0.

@gbaz
Copy link
Contributor

gbaz commented Apr 17, 2015

fixed mbox

@Daniel-Diaz
Copy link
Contributor

haskintex now accepts the new version of filepath.

@meteficha
Copy link
Member

fb and esqueleto should be fine.

@snoyberg
Copy link
Contributor Author

I've just pushed c67909b that blocks a bunch of packages that do not have GHC 7.10 support. As of now, all bounds issues are resolved, though a number of packages had to be removed. I'm going to move ahead with doing a test run of actually building the Stackage package set. I'll report the results of that, likely in a blog post to get more attention on fixing it.

@meteficha
Copy link
Member

Great work, @snoyberg.

@PierreR
Copy link
Contributor

PierreR commented Apr 30, 2015

Is there a way to build hruby successfully ? It does not work with ruby1.9 (a ruby version that should have never been the default on debian 7).

It builds fine with nix.

PS: here is the relevant commit on nixpkg.

@bergmark
Copy link
Member

@PierreR latest versions build fine for me.
screen shot 2015-04-30 at 17 33 29

@PierreR
Copy link
Contributor

PierreR commented Apr 30, 2015

@bergmark thanks that's great ! I wrongly assume it would fail because it does so on hackage.

How do you get such information ? I can't find it on https://www.stackage.org/ for instance by looking at here ?

@bergmark
Copy link
Member

I built this locally outside of stackage with https://github.com/hvr/hackage-matrix-builder

@bergmark
Copy link
Member

... so I'm not sure why you say it fails on hackage?

@PierreR
Copy link
Contributor

PierreR commented Apr 30, 2015

@bergmark I just meant it is marked as failing here:
http://hackage.haskell.org/package/hruby-0.3.1.2/reports/

@bergmark
Copy link
Member

Oh of course, sorry.

That just seems to be because that build server doesn't have ruby installed.

@PierreR
Copy link
Contributor

PierreR commented Apr 30, 2015

Yep. My question really is: can hruby be built on the stackage infrastructure given the fact it requires a version of ruby that is not 1.9.

@snoyberg
Copy link
Contributor Author

@PierreR The Stackage build server actually uses a Docker image, specified here:

https://github.com/fpco/stackage/blob/master/Dockerfile

If we can get it installed in the Docker image, then we can build it.

@snoyberg
Copy link
Contributor Author

It's official: we have a build with GHC 7.10! Details in the blog post here:

https://www.fpcomplete.com/blog/2015/04/update-ghc-7-10-stackage

Once this branch is merged to master and becomes the official Stackage Nightly (discussed in the post), we can close this issue.

@tmcdonell
Copy link
Contributor

Ah, I was just notified of this ticket so might be too late to the party, but I have just uploaded accelerate-0.15.1.0, which compiles with 7.8 and 7.10.

@snoyberg
Copy link
Contributor Author

snoyberg commented May 1, 2015

@tmcdonell Not too late at all, I'm running a new build with accelerate back in.

@snoyberg
Copy link
Contributor Author

snoyberg commented May 5, 2015

ghc7.10 is now merged to master (w00t!), closing.

@snoyberg snoyberg closed this as completed May 5, 2015
@nikita-volkov
Copy link
Contributor

"neat-interpolation" is now fixed

@phadej
Copy link
Contributor

phadej commented May 15, 2015

edit-distance got an NMU haskell-infra/hackage-trustees#19

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests