31.03.2013, в 23:54, Keith Marshall <keithmarshall@...> написал(а):
> On 31/03/13 18:46, Earnie Boyd wrote:
>>> Yes I have working strace.
>> So can you use it to determine why dash is causing uname an issue? It
>> seems strange that it works with bash and not dash. Maybe the bad
>> address is a direct result of some library used by one and not the
> Just thinking aloud; the following may, or may not be relevant...
> Perhaps I'm missing something, but IIRC Chuck built dash such that it
> explicitly would *not* link to msys-1.0.dll. Doesn't uname require the
> uname() service provided by that very dll? msys-1.0.dll would already
> be in memory, when running bash as shell, but not when running dash; is
> there some issue with uname loading msys-1.0.dll on demand, when it
> isn't already in core?
Yes I think msys-2.0.dll don't loading when I call 'uname'.
> Rebase also doesn't require msys-1.0.dll; indeed, it *mustn't*; the
> primary reason for providing dash at all, is to permit rebase to be run,
> possibly via Bourne shell scripts, *without* loading msys-1.0.dll, so
> that msys-1.0.dll itself may be rebased.
I agree that rebase doesn't require msys-2.0.dll. Calling 'uname' needed by rebaseall script. I can hardcode 'MSYS' in this script but it wrong way.