Message-ID: <326C15C2.3E7@gbrmpa.gov.au> Date: Tue, 22 Oct 1996 08:30:58 +0800 From: Leath Muller Reply-To: leathm AT gbrmpa DOT gov DOT au Organization: Great Barrier Reef Marine Park Authority MIME-Version: 1.0 To: Eli Zaretskii CC: djgpp AT delorie DOT com Subject: Re: HK: v 2.1 References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit > I probably failed to make myself clear enough, but your DIR output just > confirms what I suspected: it's Windows 95 fault! Look at the listing > above: the short 8+3 alias for `psbook.exe0' is `psbook.exe', not > something like `psbook~1.exe. Therefore, when the linker opens a file > called `psbook.exe', Windows overwrites `psbook.exe0' since as far as > Windows is concerned, a file called `psbook.exe' already exists! The bug > is in Windows 95 rename file function: it should have changed the 8+3 > alias when renaming the file, or else the file was not really renamed. > The new v2.01 library works around this bug in its `rename' function, so > if you use a program such as `mv' to rename files, this problem won't > happen. My understanding was the Win95 simply looked at the long filenames when doing all its file I/O. It places the 8.3 format in the left half of the file description for DOS people who are used to seeing it there, and all the file descriptions that it uses on the right hand side of the dir listing. I have only ever seen DOS (ie, when booting straight into DOS or rebooting to DOS) put the ~ in the filename when: i) There are two copies of a file with the same filename (8.3) ii) The filename is longer than 8 characters Does this make sense? > > 1. Make a text file, x.abc0 > > 2. open the file with Edit x.abc0 > > 3. While in Edit, do a FileSaveAs with the name x.abc > > 4. Exit Edit > > > > Both x.abc0 and x.abc exist, so it is not a Win 95 problem. > This just means that Edit (whatever that is) works around this bug. > Emacs compiled with DJGPP v2.01 also works in the above way on Windows > 95, but that doesn't imply that the bug doesn't exist, just that the > application was smart enough to get its thing right, the bug > notwithstanding. If my previous paragraph is true and Win95 only looks at the long filename, then both edit and emacs would appear to behave correctly. Leathal.