From owner-freebsd-ports Thu Aug 20 12:11:37 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id MAA03114 for freebsd-ports-outgoing; Thu, 20 Aug 1998 12:11:37 -0700 (PDT) (envelope-from owner-freebsd-ports@FreeBSD.ORG) Received: from localhost.my.domain (ppp1646.on.bellglobal.com [206.172.249.110]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id MAA03107 for ; Thu, 20 Aug 1998 12:11:29 -0700 (PDT) (envelope-from hoek@FreeBSD.ORG) Received: (from tim@localhost) by localhost.my.domain (8.8.8/8.8.8) id PAA01289; Thu, 20 Aug 1998 15:10:16 -0400 (EDT) (envelope-from tim) Message-ID: <19980820151016.A320@zappo> Date: Thu, 20 Aug 1998 15:10:16 -0400 From: Tim Vanderhoek To: Martin Cracauer Cc: freebsd-ports@FreeBSD.ORG Subject: Re: cvs commit: ports/devel/ilu Makefile References: <199808200947.CAA05643@freefall.freebsd.org> <19980820151620.A28383@cons.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.91.1i In-Reply-To: <19980820151620.A28383@cons.org>; from Martin Cracauer on Thu, Aug 20, 1998 at 03:16:20PM +0200 Sender: owner-freebsd-ports@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Thu, Aug 20, 1998 at 03:16:20PM +0200, Martin Cracauer wrote: > > I'm the maintainer. I need to build PLIST on the fly, unless I want > the port's build to depend on many other big language packages that > are irrelevant to 99% of the users (who want to use C, C++ or maybe > Java). Okay, I thought it might be something like that. If it's just a few packages, it's possible to make a second PLIST that specifies the additional files and concat them together. However, from the configure output, I assume that its installation is changed by more than "a few" packages. I do remember you sending a mail to the ports list about this when you committed it... :-) > I spent quite some time to make it work in the cases of [...] > - building a binary package and installing it on a machine with less > languages installed. Cool. This is really useful on the package-building machine. [...] > so please touch this port only with great care! Relax. I have pretty good record for the past week, actually. AFAIK, I broke only modula-3 for an extended period of time (a couple days). You could have had someone like vanilla touch the port! ;-) ;-) Almost all of the changes I've been making lately have been to what might be termed the "port substructure" as opposed to the port itself. The port had been broken at the install stage by bsd.port.mk r.1.282, r.1.227.2.44, Aug. 12. It probably escaped attention because it's marked NO_PACKAGE. To package correctly it needs r.1.285 of bsd.port.mk which has not been committed, yet. > I noticed other ports have a compile-time PLIST also, if > someone developed a better scheme to do so, please let me know. Yes, there is a better scheme, but it doesn't seem to apply to your port. > Regarding the manpages, there is a number of fixed manpages and a > number of language-specific manpages. The others are easy to identify > by -stubber.1. Ideally, what would happen is that you will be struck by bolt of lightening and think of some way to get all the manpages listed in the MANx variables. For the fixed manpages, this is easy. Unless you can identify what files are going to be installed at the configure stage, this is not so easy for the non-fixed manpages... Next-to-ideal would be just leaving the port as it is. :-) > I also chose to let the port install the manpages to its own base dir > and link them to $(PREFIX)/man. That is easy to revert, if there's > desire to do so. Nope. This is fine. I'm pretty sure there are other ports that do this. > I missed the rationale about removing manpages from PLIST. If > someone forwared me the new policy, I could take care of the manpages > for this port myself. "Any manpages listed in the Makefile with the MANx variables shall be delisted from pkg/PLIST. They are now automatically added to the plist." > As I'm going to touch the port anyway, any other suggestions? No, I I have trouble reading pkg/DESCRs that use almost a whole screen for one paragraph. My eyes start to go blurry. ;-) portint(1) reccomends keeping pkg/DESCR =< 24 lines. > couldn't locate the reason why it doesn't work when compiled with -O, > although I tried quite hard, so the reason is that I'm too stupid :-/ Maybe with a different cc? If -O does work with a different cc, it could probably be hacked to build the package with that cc, but use system cc when special cc is not installed. -- This .sig is not innovative, witty, or profund. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ports" in the body of the message