From: jazz AT softway DOT com (Jason Zions) Subject: Re: OpenNT & GNU-Win32 Comparison? 23 Jul 1997 17:04:29 -0700 Approved: cygnus DOT gnu-win32 AT cygnus DOT com Distribution: cygnus Message-ID: <33D510C3.700E.cygnus.gnu-win32@softway.com> References: <199707220820 DOT KAA19574 AT digicash DOT com> Reply-To: jazz AT softway DOT com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.0 (WinNT; I) Original-To: Jan Nieuwenhuizen Original-CC: Andreas Bischoff , gnu-win32 AT cygnus DOT com Original-Sender: owner-gnu-win32 AT cygnus DOT com > Are you suggesting microsoft would put any extra effort into their > broken posix subsystem? I personally doubt they will, since Softway seems to be doing it for them. > All unix on nt software simply(?) translates calls to the windows32 > api, is it not? No, it is not. OpenNT is not a translation layer to Win32; it is a native subsystem, a peer to the Win32 subsystem, and makes direct NT kernel calls (an API which is mostly undocumented). A beta release of gcc for OpenNT (intel and alpha) should be available on our web site (http://www.opennt.com) by the end of the week. The major reason "autoconfigure" failed: MSVC *cannot* be told to "shut up" in the case of a no-errors compile; CL.EXE always prints the name of the file being compiled to stdout, even when no errors are generated. "configure" doesn't deal well with this. Internally, we added some cruft to the cc script to filter out the noise, and configure worked much better. We expect it'll run pretty cleanly with gcc. One caveat with tcl: the tcl test suite, run under NT 4.0 earlier than SP3, will blue-screen NT. Microsoft fixed the relevant NT kernel bug in SP3. (tcl should be appearing on our web site in the relative near term as well.) > The usage of MSVC creates a mess in my opinion. > They tried to create a wrapper, which converts unix like path names to > windoze like ones. At least in my case this didn't seem to work very well. The headers we ship for use with MSVC do indeed wrap the MS-provided headers in many cases. I'd be interested in hearing what problems you had, Andreas, with the wrapping mechanism; the only reported troubles we'd seen occurred when users tried to somehow reprocess or post-process the headers we provide, which did indeed include MS headers using absolute, DOS-style paths. The gcc port completely eliminates the use of MSVC headers and DOS-style paths in source code. Jason Zions Softway Systems Inc. - For help on using this list (especially unsubscribing), send a message to "gnu-win32-request AT cygnus DOT com" with one line of text: "help".