Date: Thu, 29 Apr 1999 17:27:45 -0400 Message-Id: <199904292127.RAA27712@envy.delorie.com> From: DJ Delorie To: djgpp-workers AT delorie DOT com In-reply-to: (salvador@inti.gov.ar) Subject: Re: fork/exec/wait -> system .... More problems. CygWin is that bad? References: Reply-To: djgpp-workers AT delorie DOT com X-Mailing-List: djgpp-workers AT delorie DOT com X-Unsubscribes-To: listserv AT delorie DOT com Precedence: bulk > Sandeep (author of SDCC) failed to understand why exactly but looks > like people running CygWin programs that uses system *must* have a > valid sh in the path. It looks very strange because according to a > post in the Allegro mailing list (problems with makefiles for > CygWin) looks like Make uses command.com if no shell is available, > but perhaps that's made by code in Make and no automatically by > system. cygwin's make uses command.com *if* MAKE_MODE is unset. If you set MAKE_MODE to "unix" it changes the way it interprets makefiles to match unix, but requires a /bin/sh.exe to be available to help it along. > I told Sandeep that Cygnus people should consider including a shell > in the .zip in the same way they are including cywin.dll and I'm > including cwsdpmi.exe. But if it is too much asking I will finally > forced to write a spawn equivalent using fork/exec/wait for UNIX and > CygWin. We do provide a shell, but it's in the user-tools zip instead of the development-tools zip. DJGPP actually has a *bigger* problem because we use more zips, so the chances of a critical zip not being installed are higher. The only thing in our favor is that less zips are critical.