Thank you, Daniel. The problem is solved! I found out that both global and local package db get directory-1.1.0.0 and process-1.0.1.5. I unregistered them and other caused broken packages in the local package db. Then I successfully installed back hmatrix, ghc-mod, happy and etc. It seems that xmobar will install directory-1.1.0.0 again in the local package db. Did you succeed in installing haddock 2.9? Mine got some syntax oriented error duing cabal install haddock. But it's fine. Daniel Fischer wrote:

On Wednesday 09 March 2011 12:16:30, Qi Qi wrote:

...

I think this is the same problem as I got during installing some other packages via cabal such as hmatrix, ghc-mod, happy and etc.

Does anyone have any idea about how to solve it?

I think it may be a problem with your GHC. I have no problems cabal install'ing packages using

$ cabal --version cabal-install version 0.10.0 using version 1.10.1.0 of the Cabal library

$ ghc --version The Glorious Glasgow Haskell Compilation System, version 7.0.2

Does compiling a simple module using the Cabal library work? (for example

module Main (main) where

import Distribution.Simple.Setup

main :: IO () main = print defaultGlobalFlags

)

If your Cabal package is really unusable, that should break too (but then, why doesn't ghc-pkg check report it as broken?). Re-installing Cabal would be an option then. If it works, on the other hand, the problem may be with your cabal and re- installing that may be worth a try.

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- View this message in context: http://haskell.1045720.n5.nabble.com/some-problem-with-Cabal-tp3414662p34148... Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

I used cabal install xmobar --flags="all_extensions" . Is it true that every package installed from hackage should not install any package, which already exists in the global with the same version number , in the local db? Daniel Fischer wrote:

On Wednesday 09 March 2011 00:34:39, qiqi wrote:

...

The problem is solved! I found out that both global and local package db get directory-1.1.0.0 and process-1.0.1.5. I unregistered them and other caused broken packages in the local package db.

D'oh, I should've seen that.

...

Then I successfully installed back hmatrix, ghc-mod, happy and etc. It seems that xmobar will install directory-1.1.0.0 again in the local package db.

That's not good. And it doesn't seem to do that here:

$ cabal install xmobar --dry-run Resolving dependencies... In order, the following would be installed (use -v for more details): stm-2.2.0.1 syb-0.3 X11-1.5.0.0 utf8-string-0.3.6 xmobar-0.12

What does

$ cabal install xmobar --dry-run -v3 say for you about directory?

...

Did you succeed in installing haddock 2.9?

That was built with the compiler (I built from source), no problems.

$ haddock --version Haddock version 2.9.2, (c) Simon Marlow 2006 Ported to use the GHC API by David Waern 2006-2008

...

Mine got some syntax oriented error duing cabal install haddock. But it's fine.

The same as http://haskell.org/pipermail/haskell-cafe/2011-March/089902.html ? I don't know if that's been solved.

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- View this message in context: http://haskell.1045720.n5.nabble.com/some-problem-with-Cabal-tp3414662p34149... Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.

Daniel Fischer wrote:

On Wednesday 09 March 2011 02:51:35, qiqi wrote:

...

I used cabal install xmobar --flags="all_extensions" .

Whoa:

In order, the following would be installed: containers-0.3.0.0 (new version) filepath-1.1.0.4 (new version) directory-1.1.0.0 (reinstall) changes: filepath-1.2.0.0 -> 1.1.0.4 hinotify-0.3.1 (new package) mtl-1.1.1.1 (new version) network-2.2.1.10 (new version) process-1.0.1.5 (reinstall) changes: filepath-1.2.0.0 -> 1.1.0.4 stm-2.2.0.1 (new package) syb-0.3 (new package) X11-1.5.0.0 (new package) time-1.1.4 (new version) random-1.0.0.3 (reinstall) changes: time-1.2.0.3 -> 1.1.4 haskell98-1.1.0.1 (reinstall) utf8-string-0.3.6 (new package) X11-xft-0.3 (new package) libmpd-0.5.0 (new package) xmobar-0.12 (new package)

I haven't looked further, but libmpd is at least partially responsible, it depends on filepath < 1.2, time < 1.2 and a couple other old packages.

...

Is it true that every package installed from hackage should not install any package, which already exists in the global with the same version number , in the local db?

Generally, it shouldn't (and doesn't), but here some packages have to be reinstalled because they have to be built against different dependencies.

I'd prefer cabal install to refuse here (overridable with a --force flag) because reinstalling boot libs (containers, filepath, directory, process, time) tends to be a bad idea.

However, xmobar is an executable and no library, so if you unregister the packages (from the user db!) after installing before they can cause trouble, you should be fine (but no guarantees).

This is exactly what I am doing. I unregistered those packages but kept the xmobar's executable under ~/.cabal/bin. It works fine.

Thanks Daniel. I appreciate it!

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- View this message in context: http://haskell.1045720.n5.nabble.com/some-problem-with-Cabal-tp3414662p34149... Sent from the Haskell - Haskell-Cafe mailing list archive at Nabble.com.