Mailing-List: contact cygwin-developers-help AT cygwin DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-developers-owner AT cygwin DOT com Delivered-To: mailing list cygwin-developers AT cygwin DOT com Message-ID: <3C6AB747.6000306@ece.gatech.edu> Date: Wed, 13 Feb 2002 13:58:15 -0500 From: Charles Wilson User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:0.9.4) Gecko/20011019 Netscape6/6.2 X-Accept-Language: en-us MIME-Version: 1.0 To: cygwin-developers AT cygwin DOT com Subject: Re: Release 1.3.10 and then merge daemon code References: <20020212192227 DOT GA27947 AT redhat DOT com> <3C6A0961 DOT 8040609 AT ece DOT gatech DOT edu> <3C6A1DC0 DOT 1030307 AT ece DOT gatech DOT edu> <20020213181858 DOT GB32148 AT redhat DOT com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Christopher Faylor wrote: > Both of those makefiles are rather old so it would be hard to believe > that something this basic is required. The INSTALL_DATA script is > supposed to create the directories automatically. AFAICT, it is working > correctly. Okay, whatever. I'm just reporting my experience -- yes, $(INSTALL_DATA) [ == "$(SHELL) $(updir)/install-sh -c" ] *should* do this, but in certain cases it is not doing so on my system. It has not been working properly for months, for me. I've had to create a "template" tree of emtpy installation directories by hand before running 'make install' (I even have a script...) Since nobody else seemed to complain, I figured it was just me -- but I finally got tired of it, and provided a patch. If you don't want it, fine; but I'm keeping it in my local CVS archive because creating directories by hand (or script) is a drag. Perhaps my work style is weird: I gather that most folks compile a new DLL and either (a) copy only the DLL into /usr/bin, or (b) do a complete direct install into /usr. I tend to create "fake" 'cygwin', 'mingw', and 'w32api' packages and then use setup to install the new packages. This requires doing a 'make install' into an empty temp dir, and running a build script to "package" the result. Note that in either the (a) or (b) scenario, the problem of non-pre-existing installation directories never comes up. My guess is that ONLY you, me, and Earnie *ever* do anything like "my" procedure. You because you're the cygwin package king, Earnie because he maintains the mingw and w32api packages, and me because I'm weird. So, if my guess is right, it isn't surprising that nobody has complained -- except that it seems to work for you (and Earnie). Since ya'll are the maintainers of the packages in question, I suppose that's the important thing; as long as it's working for you... > I just tried it for a couple of things but I don't have > anything working well enough right now to try a full install. I'm > still trying to track down a gcc 3.1 bug and my dll is really screwed > up. Sorry to hear that -- but I'm glad somebody is working on cygwin+modern gcc. :-) --Chuck