"No child processes" in 6.6-candidate
Serge D. Mechveliani
mechvel at botik.ru
Thu Sep 14 07:10:26 EDT 2006
Dear GHC developers,
Sometimes ghc-6.6-candidate reports
--------------------------------------------------
runhaskell Setup.hs install --user
...
*** Exception: waitForProcess: does not exist (No child processes)
--------------------------------------------------
under Linux.
This happens as follows.
I build my large project (it uses Cabal, and it has some import loops)
and install it.
Then, I edit a bit, for example, the source module Dumatel.KBC,
remove by rm -rf all the files from the installation directory
./inst/,
and command make build (runhaskell Setup.hs build)
make install (runhaskell Setup.hs install --user).
It reports
---------------------------------------------------
scico:~/dm/1.06/dm/source> make build
runhaskell Setup.hs build
Preprocessing library dumatel-1.6...
Building dumatel-1.6...
[10 of 38] Compiling Bool0[boot] ( Bool0.hs-boot, nothing )
[22 of 38] Compiling Dumatel.KBC ( Dumatel/KBC.hs, dist/build/Dumatel/KBC.o )
/usr/bin/ar: creating dist/build/libHSdumatel-1.6.a
(looks all right: it re-compiles only the changed module and something
which also may be necessary)
scico:~/dm/1.06/dm/source> make install
runhaskell Setup.hs install --user
Installing:
/home/mechvel/dm/1.06/dm/source/inst/lib/dumatel-1.6/ghc-6.5.20060831 &
/home/mechvel/dm/1.06/dm/source/inst/bin dumatel-1.6...
*** Exception: waitForProcess: does not exist (No child processes)
make: *** [install] Error 1
------------------------------------------------------------------
ghc-6.4.1 did `install' all right in approximately the same situation
(the source modules where a bit different).
Then, I remove all the files also from the `build' directory
./dist/ and do by new `build' and `install'. And then, it works.
The matter is that compiling all modules by new slows down the process
of developing a project.
Can the above effect present a bug in the ghc + Cabal system?
-----------------
Serge Mechveliani
mechvel at botik.ru
More information about the Glasgow-haskell-users
mailing list