Date: 3 Apr 2002 23:58:01 -0000
From: Mario Sergio Fujikawa Ferreira <lioux@FreeBSD.org>
To: FreeBSD-gnats-submit@FreeBSD.org
Cc: pat@FreeBSD.org
Subject: ports/36709: bsd.port.mk MASTER_SITES:n - add comma operator to {DIST,PATCH}FILES
Message-ID: <20020403235801.18411.qmail@exxodus.fedaykin.here>
next in thread | raw e-mail | index | archive | help
>Number: 36709
>Category: ports
>Synopsis: bsd.port.mk MASTER_SITES:n - add comma operator to {DIST,PATCH}FILES
>Confidential: no
>Severity: non-critical
>Priority: low
>Responsible: freebsd-ports
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: change-request
>Submitter-Id: current-users
>Arrival-Date: Wed Apr 03 16:10:00 PST 2002
>Closed-Date:
>Last-Modified:
>Originator: Mario Sergio Fujikawa Ferreira
>Release: FreeBSD 4.5-STABLE i386
>Organization:
>Environment:
System: FreeBSD exxodus.fedaykin.here 4.5-STABLE FreeBSD 4.5-STABLE #0: Thu Mar 21 17:13:58 BRT 2002 lioux@exxodus.fedaykin.here:/usr/obj/usr/src/sys/LIOUX i386
Current cvsup updated ports tree.
>Description:
Following discussion with aDe in an uncharted IRC channel.
Check
Message-ID: <20020402140619.A65560@exxodus.fedaykin.here>
http://docs.freebsd.org/cgi/getmsg.cgi?fetch=1112088+0+current/freebsd-ports
for proposed documentation which also contemplates the idea brought
here.
Recall MASTER_SITES:n? We have comma operator that works as an
additive group operator. Thus, one {MASTER,PATCH}SITES item can
belong to more than one group (A and B) if there's a comma, e.g.,
site1:A,B. The same applies to *SUBDIR, except that we require the
user to add a leading / so that we know for sure that : is not part
of the subdir name (suggested by sobomax), e.g., subdir/:A,B.
However, the comma operator does not exist for {MASTER,PATCH}FILES.
I think that comma should be orthogonal; thus, working inside *FILES
as well.
However, if I add comma awareness to *FILES... I cannot feasibly
know if file:a,b is not a <filename>file:a,b</filename> as opposed
to <filename>file</filename> belonging to groups both a and b
Currently, without comma, I test if a *SITES group exist... if
not, it's a filename...
aDe suggested: "so why not assume that file:a,b is 'file' belonging
to groups 'a' and 'b', try to fetch, and if that fails, it's a
filename. I think it's unlikely that we'd see a filename like that."
Hummm... defeats the main purpose of the patch which is to reduce
file search delay, not to mention brute force. Furthermore, I know
it's unlikely that such filenames exist but I'd better make sure
we have our bases covered.
I welcome suggestions like we did for *SUBDIR where trailing / does
not exist in dir names. However, is there such a character we can
trust does not exist in filenames?
There are 3 solutions:
1) keep current behavior: seriously suboptimal. Check the following
patch. It is a workaround adding additional groups to achieve similar
behavior as using the comma operator;
--- Makefile.orig Wed Apr 3 17:38:35 2002
+++ Makefile Wed Apr 3 20:34:34 2002
@@ -2,7 +2,7 @@
# Date created: Tue Jun 10 21:58:54 CEST 1997
# Whom: Andreas Klemm <andreas@klemm.gtn.com>
#
-# $FreeBSD: ports/print/ghostscript-gnu/Makefile,v 1.70 2002/04/02 21:38:47 lioux Exp $
+# $FreeBSD: ports/print/ghostscript-gnu/Makefile,v 1.71 2002/04/03 20:49:37 lioux Exp $
#
PORTNAME= ghostscript
@@ -16,23 +16,23 @@
http://home.t-online.de/home/Martin.Lottermoser/pcl3dist/:hpdj_pcl3 \
http://www.harsch.net/Download/:gdevdj \
ftp://mirror.cs.wisc.edu/pub/mirrors/ghost/gnu/gs652/ \
- ftp://mirror.cs.wisc.edu/pub/mirrors/ghost/gnu/fonts/:fonts \
- ftp://gatekeeper.dec.com/pub/GNU/ghostscript/:gatekeeper
+ ftp://mirror.cs.wisc.edu/pub/mirrors/ghost/gnu/fonts/:fonts,fonts_std \
+ ftp://gatekeeper.dec.com/pub/GNU/ghostscript/:gatekeeper,DEFAULT,fonts,gimp_print,hpijs
# gatekeeper is an exception, it is the only GNU site which carries
# some of the other files, so add it although it is listed inside
# MASTER_SITE_GNU
MASTER_SITE_SUBDIR= ${PORTNAME} gimp-print/:gimp_print hpinkjet/:hpijs
PKGNAMESUFFIX= -gnu
-DISTFILES= ${GS_SOURCES}:DEFAULT,gatekeeper \
- ${GS_FONTS_STD}:fonts \
- ${GS_FONTS_OTHER}:fonts,gatekeeper \
+DISTFILES= ${GS_SOURCES}:DEFAULT \
+ ${GS_FONTS_STD}:fonts_std \
+ ${GS_FONTS_OTHER}:fonts \
${DECRYPT_PDF}:pdf \
${HP8XX_DRV}:gdevcd \
${HPDJ_SRC}:hpdj_pcl3 \
${PCL3_SRC}:hpdj_pcl3 \
${HP970_DRV}:gdevdj \
- ${GPRINT_SRC}:gatekeeper,gimp_print \
- ${HPIJS_SRC}:gatekeeper,hpijs
+ ${GPRINT_SRC}:gimp_print \
+ ${HPIJS_SRC}:hpijs
DIST_SUBDIR= ghostscript
EXTRACT_ONLY= ${GS_SOURCES}
2) add additional character delimiters such as *SUBDIR: any?
3) add an example to Porter's Handbook on how to avoid the caveat...
For instance, we have files file1, file2, file3:a. file1 belongs to group b, file2 to c and file3:a to default
DISTFILES= file1:b file2:c file3:a:DEFAULT
Therefore, you will be safe if you mandatoraly add a group definition
(even if redundant DEFAULT) to a filename containing a trailing :
The patch adding comma awareness to *FILES follows.
>How-To-Repeat:
Check revision 1.70 of ports/print/ghostscript-gnu/Makefile
http://www.freebsd.org/cgi/cvsweb.cgi/ports/print/ghostscript-gnu/Makefile?rev=1.70&content-type=text/x-cvsweb-markup
for non-working example.
>Fix:
--- bsd.port.mk.orig Wed Apr 3 16:29:12 2002
+++ bsd.port.mk Wed Apr 3 17:22:05 2002
@@ -1443,26 +1443,44 @@
DISTFILES?= ${DISTNAME}${EXTRACT_SUFX}
_MASTER_SITES_ALL= ${_MASTER_SITES_DEFAULT}
_PATCH_SITES_ALL= ${_PATCH_SITES_DEFAULT}
+_G_TEMP= DEFAULT
.for _D in ${DISTFILES}
_D_TEMP= ${_D:S/^${_D:C/:[^:]+$//}//}
-. if !empty(_D_TEMP) && defined(_MASTER_SITES_${_D_TEMP:S/^://})
-#_MASTER_SITES_ALL+= ${MASTER_SITES_${_D_TEMP:S/^://}}
-_MASTER_SITES_ALL+= ${_MASTER_SITES_${_D:S/^${_D:C/:[^:]+$//}://}}
-#_DISTFILES+= ${_D:S/:${_D_TEMP:S/^://}$//}
+. if !empty(_D_TEMP)
+. for _group in ${_D_TEMP:S/^://:S/,/ /g}
+. if !defined(_MASTER_SITES_${_group})
+_G_TEMP_TEMP= ${_G_TEMP:M/${_group}/}
+. if empty(_G_TEMP_TEMP)
+_G_TEMP+= ${_group}
+_MASTER_SITES_ALL+= ${_MASTER_SITES_${_group}}
+. endif
+. endif
+. endfor
_DISTFILES+= ${_D:C/:[^:]+$//}
. else
_DISTFILES+= ${_D}
. endif
.endfor
+_G_TEMP= DEFAULT
.for _P in ${PATCHFILES}
_P_TEMP= ${_P:S/^${_P:C/:[^:]+$//}//}
-. if !empty(_P_TEMP) && defined(_PATCH_SITES_${_P_TEMP:S/^://})
-_PATCH_SITES_ALL+= ${_PATCH_SITES_${_P:S/^${_P:C/:[^:]+$//}://}}
+. if !empty(_P_TEMP)
+. for _group in ${_P_TEMP:S/^://:S/,/ /g}
+. if !defined(_PATCH_SITES_${_group})
+_G_TEMP_TEMP= ${_G_TEMP:M/${_group}/}
+. if empty(_G_TEMP_TEMP)
+_G_TEMP+= ${_group}
+_PATCH_SITES_ALL+= ${_PATCH_SITES_${_group}}
+. endif
+. endif
+. endfor
_PATCHFILES+= ${_P:C/:[^:]+$//}
. else
_PATCHFILES+= ${_P}
. endif
.endfor
+_G_TEMP=
+_G_TEMP_TEMP=
ALLFILES?= ${_DISTFILES} ${_PATCHFILES}
#
>Release-Note:
>Audit-Trail:
>Unformatted:
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ports" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020403235801.18411.qmail>