Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT sources DOT redhat DOT com Delivered-To: mailing list cygwin AT sources DOT redhat DOT com Date: Tue, 4 Sep 2001 15:52:03 -0400 From: Christopher Faylor To: cygwin AT cygwin DOT com Subject: Re: press for cygwin Message-ID: <20010904155203.C7509@redhat.com> Reply-To: cygwin AT cygwin DOT com Mail-Followup-To: cygwin AT cygwin DOT com References: <17B78BDF120BD411B70100500422FC6309E334 AT IIS000> <3B939A12 DOT 5040009 AT ece DOT gatech DOT edu> <20010903111442 DOT D2024 AT redhat DOT com> <3B9506C5 DOT C6FAD015 AT etr-usa DOT com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <3B9506C5.C6FAD015@etr-usa.com> User-Agent: Mutt/1.3.21i On Tue, Sep 04, 2001 at 10:52:21AM -0600, Warren Young wrote: >Christopher Faylor wrote: >> >> >Somebody else mentioned this earlier -- and explained that Debian did >> >exactly that. ".deb" files are just ar archives, but .deb implies that >> >they obey some sort of internal format standard ("CYGWIN-PATCHES" ? >> >/etc/postinstall? ) >> > >> >I actually think this is a pretty good idea. >> >> It does have merits except for the fact that you lose the ability to >> distinguish between a .gz and .bz2 compressed archive. > >A quick peek at /etc/magic tells me that it's simple to distinguish the >two, and in fact it's more reliable, since magic numbers don't lie, but >file names can. Also, we might say from the outset that Cygwin packages >are all packed with bzip2, not gzip. Yes. As I keep saying (but somehow it seems to consistently be lost in the translation), I already knew that you could detect the difference when I sent my original suggestion of using magic numbers. >I realize the file name detection code exists, so it's easier to leave >it be than to add more code for magic number checking, but it _is_ >trivial to add magic number checking. For gzip, a matter of checking >the first 2 bytes of the file, and for bzip2, the first 3 bytes. It actually is not trivial at all. As I also keep mentioning, it requires that someone who cares about this will have to actually do it rather than report that it is possible to do it. >I think it's more likely that WinZip will add bzip2 decompression than a >rule to handle *.cgw files as .tar.gz. In other words, bzip2 confusing >Winzip might be a short-lived state of affairs. If you doubt that, ask >yourself if, a few years ago, if you would have guessed that WinZip >would have added tar and gzip code? I somehow doubt that WinZip would be incapable of uncompressing files merely becase they have a "non-standard" extension. This may solve the "Ah. It has a .tar.gz extension so I can use WinZip" problem, but it does not solve the "Ah. This is some kind of archive, I wonder if WinZip understands it" problem. Continuing to move towards .bz2 compression works around the problem temporarily but neither that nor naming the archives something different are a long term foolproof solution. cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/