Re: toolchain/42242 (tools build broken)

[christos%zoulas.com@localhost (Christos Zoulas)]

The following reply was made to PR toolchain/42242; it has been noted by GNATS.

From: christos%zoulas.com@localhost (Christos Zoulas)
To: Ignatios Souvatzis <ignatios%cs.uni-bonn.de@localhost>, 
gnats-bugs%NetBSD.org@localhost
Cc: toolchain-manager%NetBSD.org@localhost, netbsd-bugs%NetBSD.org@localhost, 
        gnats-admin%NetBSD.org@localhost, is%NetBSD.org@localhost
Subject: Re: toolchain/42242 (tools build broken)
Date: Thu, 29 Oct 2009 09:04:09 -0400

 On Oct 29, 12:47pm, ignatios%cs.uni-bonn.de@localhost (Ignatios Souvatzis) 
wrote:
 -- Subject: Re: toolchain/42242 (tools build broken)
 
 | The patch below fixes it.
 | 
 | Short:
 | 
 | 1. The new lex needs m4. Our old m4 wasn't GNUish enough, so a workaround
 | was added using gm4 from pkgsrc, which wasn't removed when our m4 was
 | upgraded. (Thanks to Alan Barret for finding this.)
 | 
 | (My original problem was caused because I had a broken pkgsrc gm4
 | installation. pkg_delete-ing m4 didn't help, as lex still insisted on
 | /usr/pkg/bin/m4.)
 
 Oops I forgot to commit this.
 
 | 2. Fixing m4 didn't make toolchain lex work, as toolchain m4 wasn't
 | installed in time. It seems to me that the .WAIT magic used in the
 | tools Makefiles doesn't work as expected, but only forces early
 | installation of the tool immediately before the .WAIT in the list.
 | (This is from observation, not from fully understanding the
 |  higher-order Makefile magic used.)
 | 
 | Giving m4 its own .WAIT fixed this for me; maybe a full rewrite
 | of the Makefile magic would be better in the long run.
 
 Are you sure this is needed. I will add it anyway, but it looks fishy
 to me.
 
 christos