From: david@coent.demon.co.uk (David Coe) Subject: Re: B17: Compiler-Error 11 Dec 1996 00:22:05 -0800 Sender: daemon@cygnus.com Approved: cygnus.gnu-win32@cygnus.com Distribution: cygnus Message-ID: <32ADD7B7.520A.cygnus.gnu-win32@coent.demon.co.uk> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.0Gold (WinNT; I) Original-To: Max Gotlib Original-CC: gnu-win32@cygnus.com, db@med-in.uni-sb.de Original-Sender: owner-gnu-win32@cygnus.com Max Gotlib wrote: > > It seems to me that compiler chosed non-proper string.h header file (it uses > one designed for C and not for C++). > May be somebody will fix this bug, or point out how to withdraw with > it (i'm afraid that it will not be usable to type > "c:/cygnus/include/.../string.h" > instead of > ). > I don't think the problem is in the function definition via compiler choice of the c string.h or g++ String.h header. The construct: strch ( cwd, '\0' )[-1] != '/' turns up in Groff both in libbib/index.cc and three times in grops/ps.cc each time producing the "invalid types in int [int]" diagnostic. It isn't a trick I'm familiar with, but if it means something like: cwd[ strch (cwd, '\0' ) - 1] != '/' testing for a / at the end of the character string, then replacing the four occurences by the second form gets Groff compiled and running (at least so far -:). I can't find the first form of back-stepping a pointer to a string inside a function in any of my "cookbooks". Is it legal? Best wishes -- Dr David Coe \=\ 58 Fairlawn Drive, East Grinstead \=\ Tel +44 1342 326860 West Sussex, RH19 1NT, United Kingdom \=\ Fax +44 1342 316019 - For help on using this list, send a message to "gnu-win32-request@cygnus.com" with one line of text: "help".