Page MenuHomeFreeBSD
Differential D8416

Rethink %SUBDIR% for Xfce's MASTER_SITES
ClosedPublic
Actions

Authored by madpilot on Nov 1 2016, 7:13 PM.
Tags
None
Referenced Files
F114883494: D8416.id58738.diff
Fri, Apr 18, 2:34 AM
Unknown Object (File)
Thu, Apr 17, 2:19 AM
Unknown Object (File)
Tue, Apr 15, 10:30 PM
Unknown Object (File)
Tue, Apr 15, 6:32 PM
Unknown Object (File)
Fri, Apr 11, 12:01 AM
Unknown Object (File)
Sat, Apr 5, 10:26 AM
Unknown Object (File)
Mar 6 2025, 6:57 AM
Unknown Object (File)
Mar 1 2025, 11:06 PM
View All Files
Subscribers
duchateau.olivier_gmail.com
ehaupt
madpilot
xfce

Details

Reviewers
ehaupt
mat
olivierd
Group Reviewers
O5: Ports Framework(Owns No Changed Paths)
portmgr
Commits
rP504557: - Simplify XFCE MASTER_SITES usage
Summary

This patch simplify MASTER_SITES in each Xfce ports. Users can only override the default component (xfce), possible values are:

  • apps
  • art
  • bindings
  • thunar-plugins
  • panel-plugins
  • archive

'xfce' contains core components. Test is needed, for ports where ${PORTNAME} differ from official name.

Diff Detail

Lint
Lint Skipped
Unit
Tests Skipped

Event Timeline

olivierd updated this revision to Diff 21902.Nov 1 2016, 7:13 PM
olivierd retitled this revision from to Rethink %SUBDIR% for Xfce's MASTER_SITES.
olivierd updated this object.
olivierd edited the test plan for this revision. (Show Details)
olivierd added a reviewer: mat.
olivierd added subscribers: xfce, ehaupt.
ehaupt accepted this revision.Nov 2 2016, 8:36 AM
ehaupt added a reviewer: ehaupt.

Looks good to me.

This revision is now accepted and ready to land.Nov 2 2016, 8:36 AM
mat added inline comments.Nov 2 2016, 11:58 AM
Mk/bsd.sites.mk
1240–1245

I don't understand the difference in the if and else here.

if DISTNAME without DISTVERSION is PORTNAME, then use a lowercased version of PORTNAME, otherwise, use DISTNAME without DISTVERSION.
I had a quick check, I could not find any port in this diff where both are not the same thing.

olivierd added a comment.Nov 2 2016, 6:04 PM

Lowercase is for x11-fm/thunar, package name (and tarball) is spelled Thunar, but it is located in src/xfce/thunar/ directory.

Check is required, because I noticed, when ${DISTNAME} is defined in Makefile, path is well formed, we can fetch tarball.

For example x11/libexo

make fetch
===>  License GPLv2 LGPL21 accepted by the user
===>  Found saved configuration for libexo-0.11.2
===>   libexo-0.11.2 depends on file: /usr/local/sbin/pkg - found
=> exo-0.11.2.tar.bz2 doesn't seem to exist in /usr/ports/distfiles/xfce4.
=> Attempting to fetch http://archive.xfce.org/src/xfce/exo/0.11/exo-0.11.2.tar.bz2
exo-0.11.2.tar.bz2                            100% of 1282 kB 1591 kBps 00m01s
===> Fetching all distfiles required by libexo-0.11.2 for building

But, when ${DISTNAME} is not defined in Makefile, it does not work (e.g. x11/xfce4-terminal):

make fetch
===>  License GPLv2 accepted by the user
===>   xfce4-terminal-0.8.1 depends on file: /usr/local/sbin/pkg - found
=> xfce4-terminal-0.8.1.tar.bz2 doesn't seem to exist in /usr/ports/distfiles/xfce4.
=> Attempting to fetch http://archive.xfce.org/src/apps/xfce4-terminal-0.8.1/0.8/xfce4-terminal-0.8.1.tar.bz2
fetch: http://archive.xfce.org/src/apps/xfce4-terminal-0.8.1/0.8/xfce4-terminal-0.8.1.tar.bz2: Not Found

Path should be src/apps/xfce4-terminal/0.8/ not src/apps/xfce4-terminal-0.8.1/0.8/

mat edited edge metadata.Nov 3 2016, 3:32 PM
In D8416#175291, @olivierd wrote:

Lowercase is for x11-fm/thunar, package name (and tarball) is spelled Thunar, but it is located in src/xfce/thunar/ directory.

If there are no cases of the directory being uppercased, then it could always be based on DISTNAME with :tl, so it is a non issue.

Check is required, because I noticed, when ${DISTNAME} is defined in Makefile, path is well formed, we can fetch tarball.

For example x11/libexo

make fetch
===>  License GPLv2 LGPL21 accepted by the user
===>  Found saved configuration for libexo-0.11.2
===>   libexo-0.11.2 depends on file: /usr/local/sbin/pkg - found
=> exo-0.11.2.tar.bz2 doesn't seem to exist in /usr/ports/distfiles/xfce4.
=> Attempting to fetch http://archive.xfce.org/src/xfce/exo/0.11/exo-0.11.2.tar.bz2
exo-0.11.2.tar.bz2                            100% of 1282 kB 1591 kBps 00m01s
===> Fetching all distfiles required by libexo-0.11.2 for building

But, when ${DISTNAME} is not defined in Makefile, it does not work (e.g. x11/xfce4-terminal):

make fetch
===>  License GPLv2 accepted by the user
===>   xfce4-terminal-0.8.1 depends on file: /usr/local/sbin/pkg - found
=> xfce4-terminal-0.8.1.tar.bz2 doesn't seem to exist in /usr/ports/distfiles/xfce4.
=> Attempting to fetch http://archive.xfce.org/src/apps/xfce4-terminal-0.8.1/0.8/xfce4-terminal-0.8.1.tar.bz2
fetch: http://archive.xfce.org/src/apps/xfce4-terminal-0.8.1/0.8/xfce4-terminal-0.8.1.tar.bz2: Not Found

Path should be src/apps/xfce4-terminal/0.8/ not src/apps/xfce4-terminal-0.8.1/0.8/

Right. There is some lazy evaluation problem. This seems to work for me:

_PATH=        ${DISTNAME:S/-${DISTVERSIONFULL}//:tl}/${DISTVERSION:C/^([0-9]+\.[0-9]+).*/\1/}

(Also, can _PATH be renamed to, say, _XFCE_PATH or something, to make it less generic.)

olivierd added a comment.Nov 3 2016, 7:33 PM
In D8416#175494, @mat wrote:
In D8416#175291, @olivierd wrote:

Lowercase is for x11-fm/thunar, package name (and tarball) is spelled Thunar, but it is located in src/xfce/thunar/ directory.

If there are no cases of the directory being uppercased, then it could always be based on DISTNAME with :tl, so it is a non issue.

Check is required, because I noticed, when ${DISTNAME} is defined in Makefile, path is well formed, we can fetch tarball.

For example x11/libexo

make fetch
===>  License GPLv2 LGPL21 accepted by the user
===>  Found saved configuration for libexo-0.11.2
===>   libexo-0.11.2 depends on file: /usr/local/sbin/pkg - found
=> exo-0.11.2.tar.bz2 doesn't seem to exist in /usr/ports/distfiles/xfce4.
=> Attempting to fetch http://archive.xfce.org/src/xfce/exo/0.11/exo-0.11.2.tar.bz2
exo-0.11.2.tar.bz2                            100% of 1282 kB 1591 kBps 00m01s
===> Fetching all distfiles required by libexo-0.11.2 for building

But, when ${DISTNAME} is not defined in Makefile, it does not work (e.g. x11/xfce4-terminal):

make fetch
===>  License GPLv2 accepted by the user
===>   xfce4-terminal-0.8.1 depends on file: /usr/local/sbin/pkg - found
=> xfce4-terminal-0.8.1.tar.bz2 doesn't seem to exist in /usr/ports/distfiles/xfce4.
=> Attempting to fetch http://archive.xfce.org/src/apps/xfce4-terminal-0.8.1/0.8/xfce4-terminal-0.8.1.tar.bz2
fetch: http://archive.xfce.org/src/apps/xfce4-terminal-0.8.1/0.8/xfce4-terminal-0.8.1.tar.bz2: Not Found

Path should be src/apps/xfce4-terminal/0.8/ not src/apps/xfce4-terminal-0.8.1/0.8/

Right. There is some lazy evaluation problem. This seems to work for me:

_PATH=        ${DISTNAME:S/-${DISTVERSIONFULL}//:tl}/${DISTVERSION:C/^([0-9]+\.[0-9]+).*/\1/}

(Also, can _PATH be renamed to, say, _XFCE_PATH or something, to make it less generic.)

I tried your suggestion, but it fails with some ports.
For failing ports, if I replace ${DISTVERSIONFULL} by ${DISTVERSION} everything is fine, except for ports which were functional at first.

  • x11/xfce4-terminal (not working with ${DISTVERSION}, success with ${DISTVERSIONFULL})
  • x11/libexo (not working with ${DISTVERSIONFULL}, success with ${DISTVERSION})
  • misc/xfce4-wm-themes (not working with ${DISTVERSIONFULL}, success with ${DISTVERSION})
  • x11-fm/thunar (not working with ${DISTVERSION}, success with ${DISTVERSIONFULL})
mat added a comment.Nov 4 2016, 9:51 AM
In D8416#175534, @olivierd wrote:
  • x11/xfce4-terminal (not working with ${DISTVERSION}, success with ${DISTVERSIONFULL})
  • x11/libexo (not working with ${DISTVERSIONFULL}, success with ${DISTVERSION})
  • misc/xfce4-wm-themes (not working with ${DISTVERSIONFULL}, success with ${DISTVERSION})
  • x11-fm/thunar (not working with ${DISTVERSION}, success with ${DISTVERSIONFULL})

Ok, those can also be fixed.

The problem is the way DISTNAME is created.

By default, it is ${PORTNAME}-${DISTVERSIONFULL}. If you ask to match -${DISTVERSIONFULL} to do something, it does not have to be expanded, it can simply use the -${DISTVERSIONFULL} and be done with it.

If you set DISTNAME to foo-${PORTVERSION} and ask to match -${DISTVERSIONFULL} it needs to expand both sides. At that point, I figure DISTVERSIONFULL is not yet usable, and it so, it does not do what you want. On the other hand, if you change to foo-$(DISTVERSIONFULL} (which is the same, from your point of view) it does not need to expand it to match -${DISTVERSIONFULL} and it works.

So, for those two the following patch solves the problem:

--- misc/xfce4-wm-themes/Makefile
+++ misc/xfce4-wm-themes/Makefile
@@ -9 +9 @@ MASTER_SITES=       XFCE/art
-DISTNAME=      xfwm4-themes-${PORTVERSION}
+DISTNAME=      xfwm4-themes-${DISTVERSIONFULL}
--- x11/libexo/Makefile
+++ x11/libexo/Makefile
@@ -8 +8 @@ MASTER_SITES=       XFCE
-DISTNAME=      exo-${PORTVERSION}
+DISTNAME=      exo-${DISTVERSIONFULL}

Adding a comment in bsd.sites.mk in the XFCE block saying that DISTNAME must use DISTVERSIONFULL would be enough to steer people in the right way.

(Everything you definitively did not want to know about how make(1) variables work under the hood)

madpilot updated this revision to Diff 58362.Jun 7 2019, 2:41 PM
madpilot edited the summary of this revision. (Show Details)
madpilot added a subscriber: madpilot.

I merged in the suggestions from mat into the patch from olivierd.

It works fine for me. my testing has been making my machine perform "make checksum" on all affected ports after cleaning the distcache.

I ssee this simplification as useful and I'd like to get it in the tree soon, before XFCE 4.14 is released, so the two big commits don't overlap.

If any other changes are required or problems identified I'll try to addresss those.

This revision now requires review to proceed.Jun 7 2019, 2:41 PM
Owners added a reviewer: O5: Ports Framework.Jun 7 2019, 2:41 PM
Herald added a reviewer: portmgr. · View Herald TranscriptJun 7 2019, 2:41 PM
Harbormaster completed remote builds in B24751: Diff 58362.Jun 7 2019, 2:41 PM
madpilot added a comment.Jun 7 2019, 2:42 PM

BTW if an exp run is required let me know and I'll ask for one.

madpilot added inline comments.Jun 7 2019, 2:46 PM
Mk/bsd.sites.mk
1240–1245

I've addressed this using the code you suggested in D8416#175493

duchateau.olivier_gmail.com added a subscriber: duchateau.olivier_gmail.com.Jun 8 2019, 8:37 AM
In D8416#444072, @madpilot wrote:

I merged in the suggestions from mat into the patch from olivierd.

It works fine for me. my testing has been making my machine perform "make checksum" on all affected ports after cleaning the distcache.

I ssee this simplification as useful and I'd like to get it in the tree soon, before XFCE 4.14 is released, so the two big commits don't overlap.

If any other changes are required or problems identified I'll try to addresss those.

I use this simplification for a while (on my own ports). This should be committed quickly.

madpilot updated this revision to Diff 58738.Jun 17 2019, 4:03 PM

Update patch to latest tree.

Harbormaster completed remote builds in B24899: Diff 58738.Jun 17 2019, 4:03 PM
madpilot commandeered this revision.Jun 17 2019, 4:13 PM
madpilot added a reviewer: olivierd.

Commandeering this revision, since I've taken over with new patches.

madpilot marked an inline comment as done.Jun 19 2019, 12:46 PM
mat accepted this revision as: portmgr.Jun 19 2019, 1:24 PM

Assuming you did check that everything still works :-)

madpilot added a comment.Jun 19 2019, 2:06 PM
In D8416#447242, @mat wrote:

Assuming you did check that everything still works :-)

Thanks a lot. I did test but I'm going to do a further test right before committing.

This revision was not accepted when it landed; it landed in state Needs Review.Jun 19 2019, 4:14 PM
Closed by commit rP504557: - Simplify XFCE MASTER_SITES usage (authored by madpilot). · Explain Why
This revision was automatically updated to reflect the committed changes.
madpilot added a commit: rP504557: - Simplify XFCE MASTER_SITES usage.

Revision Contents
Changeset List

PathSize
Mk/
Uses/
2 lines
26 lines
archivers/
squeeze/
2 lines
thunar-archive-plugin/
5 lines
audio/
thunar-media-tags-plugin/
2 lines
xfce4-mixer/
2 lines
xfce4-mpc-plugin/
2 lines
xfce4-pulseaudio-plugin/
2 lines
xfmpc/
2 lines
deskutils/
orage/
2 lines
xfce4-notes-plugin/
2 lines
xfce4-notifyd/
2 lines
xfce4-volumed-pulse/
2 lines
xfce4-volumed/
2 lines
xfce4-xkb-plugin/
2 lines
devel/
thunar-vcs-plugin/
5 lines
xfce4-vala/
2 lines
editors/
mousepad/
2 lines
graphics/
ristretto/
2 lines
mail/
xfce4-mailwatch-plugin/
5 lines
math/
xfce4-calculator-plugin/
2 lines
misc/
xfce4-weather-plugin/
2 lines
xfce4-wm-themes/
4 lines
multimedia/
xfce4-parole/
2 lines
print/
xfce4-print/
2 lines
science/
xfce4-equake-plugin/
2 lines
sysutils/
gigolo/
2 lines
xfburn/
2 lines
xfce4-battery-plugin/
2 lines
xfce4-cpugraph-plugin/
2 lines
xfce4-diskperf-plugin/
5 lines
xfce4-fsguard-plugin/
4 lines
xfce4-genmon-plugin/
2 lines
xfce4-mount-plugin/
2 lines
xfce4-netload-plugin/
2 lines
xfce4-power-manager/
2 lines
xfce4-settings/
2 lines
xfce4-systemload-plugin/
2 lines
xfce4-wavelan-plugin/
2 lines
textproc/
xfce4-dict-plugin/
2 lines
www/
xfce4-smartbookmark-plugin/
2 lines
x11-clocks/
xfce4-datetime-plugin/
2 lines
xfce4-timer-out-plugin/
2 lines
xfce4-timer-plugin/
2 lines
x11-fm/
thunar-vfs/
2 lines
thunar/
2 lines
x11-themes/
gtk-xfce-engine/
2 lines
x11-toolkits/
libxfce4gui/
2 lines
x11-wm/
xfce4-desktop/
2 lines
xfce4-panel/
2 lines
xfce4-session/
2 lines
xfce4-wm/
2 lines
x11/
libexo/
2 lines
libxfce4menu/
2 lines
xfce4-clipman-plugin/
2 lines
xfce4-conf/
2 lines
xfce4-dashboard/
2 lines
xfce4-embed-plugin/
2 lines
xfce4-quicklauncher-plugin/
2 lines
xfce4-screenshooter-plugin/
2 lines
xfce4-taskmanager/
2 lines
xfce4-terminal/
2 lines
xfce4-verve-plugin/
2 lines
xfce4-whiskermenu-plugin/
2 lines
xfce4-wmdock-plugin/
2 lines

Diff 21902

View Options

Mk/Uses/xfce.mk

Show All 28 Lines
LDRSRC= ${BOOTSRC}/common LDRSRC= ${BOOTSRC}/common
LIBLUASRC= ${BOOTSRC}/liblua LIBLUASRC= ${BOOTSRC}/liblua
LUASRC= ${SRCTOP}/contrib/lua/src LUASRC= ${SRCTOP}/contrib/lua/src
SASRC= ${BOOTSRC}/libsa SASRC= ${BOOTSRC}/libsa
SYSDIR= ${SRCTOP}/sys SYSDIR= ${SRCTOP}/sys
UBOOTSRC= ${BOOTSRC}/uboot UBOOTSRC= ${BOOTSRC}/uboot
ZFSSRC= ${SASRC}/zfs ZFSSRC= ${SASRC}/zfs
LIBCSRC= ${SRCTOP}/lib/libc LIBCSRC= ${SRCTOP}/lib/libc
WORLDTMP= ${OBJTOP}/tmp
BOOTOBJ= ${OBJTOP}/stand BOOTOBJ= ${OBJTOP}/stand
# BINDIR is where we install # BINDIR is where we install
BINDIR?= /boot BINDIR?= /boot
# LUAPATH is where we search for and install lua scripts. # LUAPATH is where we search for and install lua scripts.
LUAPATH?= /boot/lua LUAPATH?= /boot/lua
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
.endif .endif
# Machine specific flags for all builds here # Machine specific flags for all builds here
# All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc # All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc
# or powerpc64. # or powerpc64.
.if ${MACHINE_ARCH} == "powerpc64" .if ${MACHINE_ARCH} == "powerpc64"
CFLAGS+= -m32 -mcpu=powerpc CFLAGS+= -m32 -mcpu=powerpc
# Use same linker used to link 32 bit binaries
.if "${COMPILER_TYPE}" == "clang" && "${LINKER_TYPE}" == "lld"
bdragonUnsubmitted
Done
Inline Actions

I'm seeing build failures at cleandir stage. I guess LINKER_TYPE isn't being defined at that point. So ${LINKER_TYPE} would need to be wrapped in quotes, or for LINKER_TYPE to be defined at all stages.

bdragon: I'm seeing build failures at cleandir stage. I guess LINKER_TYPE isn't being defined at that…
CFLAGS+= -fuse-ld=${LIB32LD}
impUnsubmitted
Done
Inline Actions

how does LIB32LD get defined? That's 100% opaque to me

imp: how does LIB32LD get defined? That's 100% opaque to me
alfredoAuthorUnsubmitted
Done
Inline Actions

It's defined by in Makefile.libcompat.inc1 and passed to 'stand' by Makefile.inc1 through CROSSENV variable.

alfredo: It's defined by in Makefile.libcompat.inc1 and passed to 'stand' by Makefile.inc1 through…
impUnsubmitted
Not Done
Inline Actions

This will fail when compiled stand alone outside of buildworld.

imp: This will fail when compiled stand alone outside of buildworld.
luporlUnsubmitted
Done
Inline Actions

LIB32LD will be passed to stand either through buildworld or by first performing a 'make buildenv' before changing to stand and trying to build it.

But when trying to build stand without calling buildenv first, LIB32LD will be undefined.
In this case I chose to report an error, informing that LIB32LD must be set when stand is built with clang on powerpc64.

Is that ok? If not, what would be the alternative? Put the LIB32LD detection code in a separate make file that could be included by stand/defs.mk?

luporl: LIB32LD will be passed to stand either through buildworld or by first performing a 'make…
impUnsubmitted
Not Done
Inline Actions

No. I'd say that's not OK. It's almost never OK to have .error statements based on things not being defined. there's too many use cases to know you covered them all.

imp: No. I'd say that's not OK. It's almost never OK to have .error statements based on things not…
impUnsubmitted
Done
Inline Actions

LIB32LD still isn't defined in a file that gets included in the build.
Defining it all in bsd.compat.mk seems fundamentally wrong to me. that's not really what that's for.
If you want to use those things in the rest of the build, you'd need to define it bsd.cpu.mk (which bsd.compat.mk has access to).

imp: LIB32LD still isn't defined in a file that gets included in the build. Defining it all in bsd.
.endif
.endif .endif
# For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is # For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is
# build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here # build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here
# and activate it when DO32 is explicitly defined to be 1. # and activate it when DO32 is explicitly defined to be 1.
.if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1 .if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1
CFLAGS+= -m32 CFLAGS+= -m32
# LD_FLAGS is passed directly to ${LD}, not via ${CC}: # LD_FLAGS is passed directly to ${LD}, not via ${CC}:
▲ Show 20 Lines • Show All 128 Lines • Show Last 20 Lines
View Options

Mk/bsd.sites.mk

Show All 28 Lines
LDRSRC= ${BOOTSRC}/common LDRSRC= ${BOOTSRC}/common
LIBLUASRC= ${BOOTSRC}/liblua LIBLUASRC= ${BOOTSRC}/liblua
LUASRC= ${SRCTOP}/contrib/lua/src LUASRC= ${SRCTOP}/contrib/lua/src
SASRC= ${BOOTSRC}/libsa SASRC= ${BOOTSRC}/libsa
SYSDIR= ${SRCTOP}/sys SYSDIR= ${SRCTOP}/sys
UBOOTSRC= ${BOOTSRC}/uboot UBOOTSRC= ${BOOTSRC}/uboot
ZFSSRC= ${SASRC}/zfs ZFSSRC= ${SASRC}/zfs
LIBCSRC= ${SRCTOP}/lib/libc LIBCSRC= ${SRCTOP}/lib/libc
WORLDTMP= ${OBJTOP}/tmp
BOOTOBJ= ${OBJTOP}/stand BOOTOBJ= ${OBJTOP}/stand
# BINDIR is where we install # BINDIR is where we install
BINDIR?= /boot BINDIR?= /boot
# LUAPATH is where we search for and install lua scripts. # LUAPATH is where we search for and install lua scripts.
LUAPATH?= /boot/lua LUAPATH?= /boot/lua
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
.endif .endif
# Machine specific flags for all builds here # Machine specific flags for all builds here
# All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc # All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc
# or powerpc64. # or powerpc64.
.if ${MACHINE_ARCH} == "powerpc64" .if ${MACHINE_ARCH} == "powerpc64"
CFLAGS+= -m32 -mcpu=powerpc CFLAGS+= -m32 -mcpu=powerpc
# Use same linker used to link 32 bit binaries
.if "${COMPILER_TYPE}" == "clang" && "${LINKER_TYPE}" == "lld"
bdragonUnsubmitted
Done
Inline Actions

I'm seeing build failures at cleandir stage. I guess LINKER_TYPE isn't being defined at that point. So ${LINKER_TYPE} would need to be wrapped in quotes, or for LINKER_TYPE to be defined at all stages.

bdragon: I'm seeing build failures at cleandir stage. I guess LINKER_TYPE isn't being defined at that…
CFLAGS+= -fuse-ld=${LIB32LD}
impUnsubmitted
Done
Inline Actions

how does LIB32LD get defined? That's 100% opaque to me

imp: how does LIB32LD get defined? That's 100% opaque to me
alfredoAuthorUnsubmitted
Done
Inline Actions

It's defined by in Makefile.libcompat.inc1 and passed to 'stand' by Makefile.inc1 through CROSSENV variable.

alfredo: It's defined by in Makefile.libcompat.inc1 and passed to 'stand' by Makefile.inc1 through…
impUnsubmitted
Not Done
Inline Actions

This will fail when compiled stand alone outside of buildworld.

imp: This will fail when compiled stand alone outside of buildworld.
luporlUnsubmitted
Done
Inline Actions

LIB32LD will be passed to stand either through buildworld or by first performing a 'make buildenv' before changing to stand and trying to build it.

But when trying to build stand without calling buildenv first, LIB32LD will be undefined.
In this case I chose to report an error, informing that LIB32LD must be set when stand is built with clang on powerpc64.

Is that ok? If not, what would be the alternative? Put the LIB32LD detection code in a separate make file that could be included by stand/defs.mk?

luporl: LIB32LD will be passed to stand either through buildworld or by first performing a 'make…
impUnsubmitted
Not Done
Inline Actions

No. I'd say that's not OK. It's almost never OK to have .error statements based on things not being defined. there's too many use cases to know you covered them all.

imp: No. I'd say that's not OK. It's almost never OK to have .error statements based on things not…
impUnsubmitted
Done
Inline Actions

LIB32LD still isn't defined in a file that gets included in the build.
Defining it all in bsd.compat.mk seems fundamentally wrong to me. that's not really what that's for.
If you want to use those things in the rest of the build, you'd need to define it bsd.cpu.mk (which bsd.compat.mk has access to).

imp: LIB32LD still isn't defined in a file that gets included in the build. Defining it all in bsd.
.endif
.endif .endif
# For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is # For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is
# build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here # build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here
# and activate it when DO32 is explicitly defined to be 1. # and activate it when DO32 is explicitly defined to be 1.
.if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1 .if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1
CFLAGS+= -m32 CFLAGS+= -m32
# LD_FLAGS is passed directly to ${LD}, not via ${CC}: # LD_FLAGS is passed directly to ${LD}, not via ${CC}:
▲ Show 20 Lines • Show All 128 Lines • Show Last 20 Lines
View Options

archivers/squeeze/Makefile

Show All 28 Lines
LDRSRC= ${BOOTSRC}/common LDRSRC= ${BOOTSRC}/common
LIBLUASRC= ${BOOTSRC}/liblua LIBLUASRC= ${BOOTSRC}/liblua
LUASRC= ${SRCTOP}/contrib/lua/src LUASRC= ${SRCTOP}/contrib/lua/src
SASRC= ${BOOTSRC}/libsa SASRC= ${BOOTSRC}/libsa
SYSDIR= ${SRCTOP}/sys SYSDIR= ${SRCTOP}/sys
UBOOTSRC= ${BOOTSRC}/uboot UBOOTSRC= ${BOOTSRC}/uboot
ZFSSRC= ${SASRC}/zfs ZFSSRC= ${SASRC}/zfs
LIBCSRC= ${SRCTOP}/lib/libc LIBCSRC= ${SRCTOP}/lib/libc
WORLDTMP= ${OBJTOP}/tmp
BOOTOBJ= ${OBJTOP}/stand BOOTOBJ= ${OBJTOP}/stand
# BINDIR is where we install # BINDIR is where we install
BINDIR?= /boot BINDIR?= /boot
# LUAPATH is where we search for and install lua scripts. # LUAPATH is where we search for and install lua scripts.
LUAPATH?= /boot/lua LUAPATH?= /boot/lua
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
.endif .endif
# Machine specific flags for all builds here # Machine specific flags for all builds here
# All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc # All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc
# or powerpc64. # or powerpc64.
.if ${MACHINE_ARCH} == "powerpc64" .if ${MACHINE_ARCH} == "powerpc64"
CFLAGS+= -m32 -mcpu=powerpc CFLAGS+= -m32 -mcpu=powerpc
# Use same linker used to link 32 bit binaries
.if "${COMPILER_TYPE}" == "clang" && "${LINKER_TYPE}" == "lld"
bdragonUnsubmitted
Done
Inline Actions

I'm seeing build failures at cleandir stage. I guess LINKER_TYPE isn't being defined at that point. So ${LINKER_TYPE} would need to be wrapped in quotes, or for LINKER_TYPE to be defined at all stages.

bdragon: I'm seeing build failures at cleandir stage. I guess LINKER_TYPE isn't being defined at that…
CFLAGS+= -fuse-ld=${LIB32LD}
impUnsubmitted
Done
Inline Actions

how does LIB32LD get defined? That's 100% opaque to me

imp: how does LIB32LD get defined? That's 100% opaque to me
alfredoAuthorUnsubmitted
Done
Inline Actions

It's defined by in Makefile.libcompat.inc1 and passed to 'stand' by Makefile.inc1 through CROSSENV variable.

alfredo: It's defined by in Makefile.libcompat.inc1 and passed to 'stand' by Makefile.inc1 through…
impUnsubmitted
Not Done
Inline Actions

This will fail when compiled stand alone outside of buildworld.

imp: This will fail when compiled stand alone outside of buildworld.
luporlUnsubmitted
Done
Inline Actions

LIB32LD will be passed to stand either through buildworld or by first performing a 'make buildenv' before changing to stand and trying to build it.

But when trying to build stand without calling buildenv first, LIB32LD will be undefined.
In this case I chose to report an error, informing that LIB32LD must be set when stand is built with clang on powerpc64.

Is that ok? If not, what would be the alternative? Put the LIB32LD detection code in a separate make file that could be included by stand/defs.mk?

luporl: LIB32LD will be passed to stand either through buildworld or by first performing a 'make…
impUnsubmitted
Not Done
Inline Actions

No. I'd say that's not OK. It's almost never OK to have .error statements based on things not being defined. there's too many use cases to know you covered them all.

imp: No. I'd say that's not OK. It's almost never OK to have .error statements based on things not…
impUnsubmitted
Done
Inline Actions

LIB32LD still isn't defined in a file that gets included in the build.
Defining it all in bsd.compat.mk seems fundamentally wrong to me. that's not really what that's for.
If you want to use those things in the rest of the build, you'd need to define it bsd.cpu.mk (which bsd.compat.mk has access to).

imp: LIB32LD still isn't defined in a file that gets included in the build. Defining it all in bsd.
.endif
.endif .endif
# For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is # For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is
# build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here # build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here
# and activate it when DO32 is explicitly defined to be 1. # and activate it when DO32 is explicitly defined to be 1.
.if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1 .if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1
CFLAGS+= -m32 CFLAGS+= -m32
# LD_FLAGS is passed directly to ${LD}, not via ${CC}: # LD_FLAGS is passed directly to ${LD}, not via ${CC}:
▲ Show 20 Lines • Show All 128 Lines • Show Last 20 Lines
View Options

archivers/thunar-archive-plugin/Makefile

Show All 28 Lines
LDRSRC= ${BOOTSRC}/common LDRSRC= ${BOOTSRC}/common
LIBLUASRC= ${BOOTSRC}/liblua LIBLUASRC= ${BOOTSRC}/liblua
LUASRC= ${SRCTOP}/contrib/lua/src LUASRC= ${SRCTOP}/contrib/lua/src
SASRC= ${BOOTSRC}/libsa SASRC= ${BOOTSRC}/libsa
SYSDIR= ${SRCTOP}/sys SYSDIR= ${SRCTOP}/sys
UBOOTSRC= ${BOOTSRC}/uboot UBOOTSRC= ${BOOTSRC}/uboot
ZFSSRC= ${SASRC}/zfs ZFSSRC= ${SASRC}/zfs
LIBCSRC= ${SRCTOP}/lib/libc LIBCSRC= ${SRCTOP}/lib/libc
WORLDTMP= ${OBJTOP}/tmp
BOOTOBJ= ${OBJTOP}/stand BOOTOBJ= ${OBJTOP}/stand
# BINDIR is where we install # BINDIR is where we install
BINDIR?= /boot BINDIR?= /boot
# LUAPATH is where we search for and install lua scripts. # LUAPATH is where we search for and install lua scripts.
LUAPATH?= /boot/lua LUAPATH?= /boot/lua
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
.endif .endif
# Machine specific flags for all builds here # Machine specific flags for all builds here
# All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc # All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc
# or powerpc64. # or powerpc64.
.if ${MACHINE_ARCH} == "powerpc64" .if ${MACHINE_ARCH} == "powerpc64"
CFLAGS+= -m32 -mcpu=powerpc CFLAGS+= -m32 -mcpu=powerpc
# Use same linker used to link 32 bit binaries
.if "${COMPILER_TYPE}" == "clang" && "${LINKER_TYPE}" == "lld"
bdragonUnsubmitted
Done
Inline Actions

I'm seeing build failures at cleandir stage. I guess LINKER_TYPE isn't being defined at that point. So ${LINKER_TYPE} would need to be wrapped in quotes, or for LINKER_TYPE to be defined at all stages.

bdragon: I'm seeing build failures at cleandir stage. I guess LINKER_TYPE isn't being defined at that…
CFLAGS+= -fuse-ld=${LIB32LD}
impUnsubmitted
Done
Inline Actions

how does LIB32LD get defined? That's 100% opaque to me

imp: how does LIB32LD get defined? That's 100% opaque to me
alfredoAuthorUnsubmitted
Done
Inline Actions

It's defined by in Makefile.libcompat.inc1 and passed to 'stand' by Makefile.inc1 through CROSSENV variable.

alfredo: It's defined by in Makefile.libcompat.inc1 and passed to 'stand' by Makefile.inc1 through…
impUnsubmitted
Not Done
Inline Actions

This will fail when compiled stand alone outside of buildworld.

imp: This will fail when compiled stand alone outside of buildworld.
luporlUnsubmitted
Done
Inline Actions

LIB32LD will be passed to stand either through buildworld or by first performing a 'make buildenv' before changing to stand and trying to build it.

But when trying to build stand without calling buildenv first, LIB32LD will be undefined.
In this case I chose to report an error, informing that LIB32LD must be set when stand is built with clang on powerpc64.

Is that ok? If not, what would be the alternative? Put the LIB32LD detection code in a separate make file that could be included by stand/defs.mk?

luporl: LIB32LD will be passed to stand either through buildworld or by first performing a 'make…
impUnsubmitted
Not Done
Inline Actions

No. I'd say that's not OK. It's almost never OK to have .error statements based on things not being defined. there's too many use cases to know you covered them all.

imp: No. I'd say that's not OK. It's almost never OK to have .error statements based on things not…
impUnsubmitted
Done
Inline Actions

LIB32LD still isn't defined in a file that gets included in the build.
Defining it all in bsd.compat.mk seems fundamentally wrong to me. that's not really what that's for.
If you want to use those things in the rest of the build, you'd need to define it bsd.cpu.mk (which bsd.compat.mk has access to).

imp: LIB32LD still isn't defined in a file that gets included in the build. Defining it all in bsd.
.endif
.endif .endif
# For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is # For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is
# build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here # build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here
# and activate it when DO32 is explicitly defined to be 1. # and activate it when DO32 is explicitly defined to be 1.
.if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1 .if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1
CFLAGS+= -m32 CFLAGS+= -m32
# LD_FLAGS is passed directly to ${LD}, not via ${CC}: # LD_FLAGS is passed directly to ${LD}, not via ${CC}:
▲ Show 20 Lines • Show All 128 Lines • Show Last 20 Lines
View Options

audio/thunar-media-tags-plugin/Makefile

Loading...
View Options

audio/xfce4-mixer/Makefile

Loading...
View Options

audio/xfce4-mpc-plugin/Makefile

Loading...
View Options

audio/xfce4-pulseaudio-plugin/Makefile

Loading...
View Options

audio/xfmpc/Makefile

Loading...
View Options

deskutils/orage/Makefile

Loading...
View Options

deskutils/xfce4-notes-plugin/Makefile

Loading...
View Options

deskutils/xfce4-notifyd/Makefile

Loading...
View Options

deskutils/xfce4-volumed-pulse/Makefile

Loading...
View Options

deskutils/xfce4-volumed/Makefile

Loading...
View Options

deskutils/xfce4-xkb-plugin/Makefile

Loading...
View Options

devel/thunar-vcs-plugin/Makefile

Loading...
View Options

devel/xfce4-vala/Makefile

Loading...
View Options

editors/mousepad/Makefile

Loading...
View Options

graphics/ristretto/Makefile

Loading...
View Options

mail/xfce4-mailwatch-plugin/Makefile

Loading...
View Options

math/xfce4-calculator-plugin/Makefile

Loading...
View Options

misc/xfce4-weather-plugin/Makefile

Loading...
View Options

misc/xfce4-wm-themes/Makefile

Loading...
View Options

multimedia/xfce4-parole/Makefile

Loading...
View Options

print/xfce4-print/Makefile

Loading...
View Options

science/xfce4-equake-plugin/Makefile

Loading...
View Options

sysutils/gigolo/Makefile

Loading...
View Options

sysutils/xfburn/Makefile

Loading...
View Options

sysutils/xfce4-battery-plugin/Makefile

Loading...
View Options

sysutils/xfce4-cpugraph-plugin/Makefile

Loading...
View Options

sysutils/xfce4-diskperf-plugin/Makefile

Loading...
View Options

sysutils/xfce4-fsguard-plugin/Makefile

Loading...
View Options

sysutils/xfce4-genmon-plugin/Makefile

Loading...
View Options

sysutils/xfce4-mount-plugin/Makefile

Loading...
View Options

sysutils/xfce4-netload-plugin/Makefile

Loading...
View Options

sysutils/xfce4-power-manager/Makefile

Loading...
View Options

sysutils/xfce4-settings/Makefile

Loading...
View Options

sysutils/xfce4-systemload-plugin/Makefile

Loading...
View Options

sysutils/xfce4-wavelan-plugin/Makefile

Loading...
View Options

textproc/xfce4-dict-plugin/Makefile

Loading...
View Options

www/xfce4-smartbookmark-plugin/Makefile

Loading...
View Options

x11-clocks/xfce4-datetime-plugin/Makefile

Loading...
View Options

x11-clocks/xfce4-timer-out-plugin/Makefile

Loading...
View Options

x11-clocks/xfce4-timer-plugin/Makefile

Loading...
View Options

x11-fm/thunar-vfs/Makefile

Loading...
View Options

x11-fm/thunar/Makefile

Loading...
View Options

x11-themes/gtk-xfce-engine/Makefile

Loading...
View Options

x11-toolkits/libxfce4gui/Makefile

Loading...
View Options

x11-wm/xfce4-desktop/Makefile

Loading...
View Options

x11-wm/xfce4-panel/Makefile

Loading...
View Options

x11-wm/xfce4-session/Makefile

Loading...
View Options

x11-wm/xfce4-wm/Makefile

Loading...
View Options

x11/libexo/Makefile

Loading...
View Options

x11/libxfce4menu/Makefile

Loading...
View Options

x11/xfce4-clipman-plugin/Makefile

Loading...
View Options

x11/xfce4-conf/Makefile

Loading...
View Options

x11/xfce4-dashboard/Makefile

Loading...
View Options

x11/xfce4-embed-plugin/Makefile

Loading...
View Options

x11/xfce4-quicklauncher-plugin/Makefile

Loading...
View Options

x11/xfce4-screenshooter-plugin/Makefile

Loading...
View Options

x11/xfce4-taskmanager/Makefile

Loading...
View Options

x11/xfce4-terminal/Makefile

Loading...
View Options

x11/xfce4-verve-plugin/Makefile

Loading...
View Options

x11/xfce4-whiskermenu-plugin/Makefile

Loading...
View Options

x11/xfce4-wmdock-plugin/Makefile

Loading...