Insufficient buffer size to read current working directory: 0

[bstjean]

The primitive primitiveGetCurrentWorkingDirectory seems to fail as tit fallse back to the Smalltalk code in Win32Platform>>currentWorkingDirectoryPathWithBufferSize:

To Reproduce Steps to reproduce the behavior:

1. Create a pharo 9 image (32 or 64 bit) on Windows 10
2. Launch the image
3. The error occurs (see stack below). This error does NOT happen with Pharo 8 images!

Win32Platform(Object)>>error: Win32Platform(OSPlatform)>>currentWorkingDirectoryPathWithBuffer: Win32Platform(WinPlatform)>>currentWorkingDirectoryPathWithBufferSize: Win32Platform(WinPlatform)>>currentWorkingDirectoryPath DiskStore class>>defaultWorkingDirectory DiskStore class>>startUp: ClassSessionHandler>>startup: [ :arg2 | arg2 startup: arg1 ] in WorkingSession>>runStartup: in Block: [ :arg2 | arg2 startup: arg1 ] [ arg2 value: arg3 ] in [ :arg3 | [ arg2 value: arg3 ] on: Exception do: [ :arg4 | self errorHandler handleError: arg4 ] ] in WorkingSession>>runList:do: in Block: [ arg2 value: arg3 ] FullBlockClosure(BlockClosure)>>on:do: [ :arg3 | [ arg2 value: arg3 ] on: Exception do: [ :arg4 | self errorHandler handleError: arg4 ] ] in WorkingSession>>runList:do: in Block: [ :arg3 | ... Array(SequenceableCollection)>>do: WorkingSession>>runList:do: WorkingSession>>runStartup: WorkingSession>>start: SessionManager>>launchSnapshot:andQuit: [ tmp3 := self launchSnapshot: arg1 andQuit: arg2. tmp1 signal ] in SessionManager>>snapshot:andQuit: in Block: [ ... [ self value. Processor terminateActive ] in FullBlockClosure(BlockClosure)>>newProcess in Block: [ ...

[guillep]

Thanks for the report, this is not a Pharo launcher problem, this is related to some changes we are doing on windows to have better native support. I'll take care of fixing it over the week.

Thanks again!

[guillep]

I've migrated the issue to the Pharo repository, I'll follow it up in there!