Date: Sun, 19 Apr 1998 13:51:33 +0300 (IDT) From: Eli Zaretskii To: Paul Derbyshire cc: djgpp AT delorie DOT com Subject: Re: RHIDE and pentium cc1plus In-Reply-To: <9hAZ.3264$sN3.543198@news21.bellglobal.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Precedence: bulk On Fri, 17 Apr 1998, Paul Derbyshire wrote: > When run from the command prompt, cc1plus works fine, even > though it has exactly the same memory available (since RHIDE, like all > DJGPP-compiled software, swaps itself out entirely when it launches > a subtask). This is incorrect. DJGPP programs do NOT swap themselves out of memory when running child programs. This aspect of memory management is left entirely up to the DPMI server; most of the servers (I think all of them) only swap out the parent when they don't have enough physical memory for the child. > So what is wrong with PGCC? PGCC is an experimental version. Several messages posted to this forum, including some by me, explained that PGCC trades stability for support of advanced features. When you read these messages, please take them at face value, and consider your options carefully. Once again, I advise DJGPP users to not use PGCC unless (1) you *really* need the advanced features not supported by GCC; and (2) you are prepared to pay the price when it happens to have bugs or misfeatures (the ``price'' being that you will need to rearrange or rewrite your code, play with different optimization options, ask people for help, report bugs, etc.). Free software doesn't mean you are freed from the need to think and be responsible for the consequences of your actions.