propagate git-annex -c on to transferrer child process
git -c was already propagated via environment, but need this for consistency. Also, notice it does not use gitAnnexChildProcess to run the transferrer. So nothing is done about avoid it taking the pid lock. It's possible that the caller is already doing something that took the pid lock, and if so, the transferrer will certianly fail, since it needs to take the pid lock too. This may prevent combining annex.stalldetection with annex.pidlock, but I have not verified it's really a problem. If it was, it seems git-annex would have to take the pid lock when starting a transferrer, and hold it until shutdown, or would need to take pid lock when starting to use a transferrer, and hold it until done with a transfer and then drop it. The latter would require starting the transferrer with pid locking disabled for the child process, so assumes that the transferrer does not do anyting that needs locking when not running a transfer.
This commit is contained in:
parent
00526a6739
commit
74c1e0660b
8 changed files with 65 additions and 45 deletions
|
@ -38,12 +38,12 @@ import Config
|
|||
-}
|
||||
pidLockChildProcess
|
||||
:: FilePath
|
||||
-> [String]
|
||||
-> [CommandParam]
|
||||
-> (CreateProcess -> CreateProcess)
|
||||
-> (Maybe Handle -> Maybe Handle -> Maybe Handle -> ProcessHandle -> IO a)
|
||||
-> Annex a
|
||||
pidLockChildProcess cmd ps f a = do
|
||||
let p = f (proc cmd ps)
|
||||
let p = f (proc cmd (toCommand ps))
|
||||
let gonopidlock = withCreateProcess p a
|
||||
#ifndef mingw32_HOST_OS
|
||||
pidLockFile >>= liftIO . \case
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue