Re: DEPENDS -- is it time to remove it?

2007-12-11 Thread Paul Schenkeveld
On Fri, Jan 05, 2007 at 08:28:10PM +, RW wrote:
 On Fri, 5 Jan 2007 20:52:50 +0300
 Andrew Pantyukhin [EMAIL PROTECTED] wrote:
 
  On 1/5/07, RW [EMAIL PROTECTED] wrote:
   Isn't DEPENDS still a sensible way of making
   one metaport depend on another. For example
   if someone wanted to create a personal desk-
   top metaport that depends on KDE, xorg etc.
  
  People need programs, not ports. 
 
 It's not that straightforward when you want to depend on a
 metaport like KDE. All of the binaries can be provided by individual
 sub-ports. The sensible thing to do is create a dependency on KDE and
 let KDE's options/knobs handle the lower dependencies. The ports tree
 doesn't need to have metaports depend on metaports, but some people
 find it useful to create their own. 
 
 It's more
  sensible to run_depend on files than just on
  ports.
 
 Looking at the porter's handbook it looks like the solution is to use
 RUN_DEPENDS with${NONEXISTENT}, so I guess DEPENDS is redundant.

So now I have a need for a metaport to depend on another metaport.
Without DEPENDS, how do I accomplish that.  Using RUN_DEPENDS with 
${NONEXISTENT} seems not appropriate here, the Porters Handbook says
that this should only be used to pull in source, not to install another
metaport (unless it is already installed) and the effect of using
something like RUN_DEPENDS=${NONEXISTENT}:${PORTSDIR}/lang/php5-extensions
causes make install to try installing php5-extensions even when this port
is already installed.

The two metaport I need to depend on are php5-extensions and
xorg-drivers and I really don't want to copy the OPTIONS processing of
these ports and maintain that in the future.

Thanks in advance for any help.

Paul Schenkeveld
___
freebsd-ports@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: DEPENDS -- is it time to remove it?

2007-12-11 Thread Scot Hetzel
On 12/11/07, Paul Schenkeveld [EMAIL PROTECTED] wrote:
 So now I have a need for a metaport to depend on another metaport.
 Without DEPENDS, how do I accomplish that.  Using RUN_DEPENDS with
 ${NONEXISTENT} seems not appropriate here, the Porters Handbook says
 that this should only be used to pull in source, not to install another
 metaport (unless it is already installed) and the effect of using
 something like RUN_DEPENDS=${NONEXISTENT}:${PORTSDIR}/lang/php5-extensions
 causes make install to try installing php5-extensions even when this port
 is already installed.

 The two metaport I need to depend on are php5-extensions and
 xorg-drivers and I really don't want to copy the OPTIONS processing of
 these ports and maintain that in the future.


For xorg-drivers you simply use:

RUN_DEPENDS= 
${LOCALBASE}/libdata/xorg/drivers:${PORTSDIR}/x11-drivers/xorg-drivers

to get this to work for php5-extensions, you would need to patch the
Makefile with:

PLIST_FILES= libdata/php5/extensions
PLIST_DIR= libdata/php5

do-install:
${MKDIR} ${PREFIX}/libdata/php5
${TOUCH} ${PREFIX}/libdata/php5/extensions

Then you could use:

RUN_DEPENDS= 
${LOCALBASE}/libdata/php5/extensions:${PORTSDIR}/lang/php5-extensions

Scot
___
freebsd-ports@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: DEPENDS -- is it time to remove it?

2007-12-11 Thread Paul Schenkeveld
On Tue, Dec 11, 2007 at 02:40:11PM +0100, Gabor Kovesdan wrote:
 Scot Hetzel escribió:
 On 12/11/07, Paul Schenkeveld [EMAIL PROTECTED] wrote:
   
 So now I have a need for a metaport to depend on another metaport.
 Without DEPENDS, how do I accomplish that.  Using RUN_DEPENDS with
 ${NONEXISTENT} seems not appropriate here, the Porters Handbook says
 that this should only be used to pull in source, not to install another
 metaport (unless it is already installed) and the effect of using
 something like RUN_DEPENDS=${NONEXISTENT}:${PORTSDIR}/lang/php5-extensions
 causes make install to try installing php5-extensions even when this port
 is already installed.
 
 The two metaport I need to depend on are php5-extensions and
 xorg-drivers and I really don't want to copy the OPTIONS processing of
 these ports and maintain that in the future.
 
 
 
 For xorg-drivers you simply use:
 
 RUN_DEPENDS= 
 ${LOCALBASE}/libdata/xorg/drivers:${PORTSDIR}/x11-drivers/xorg-drivers
 
 to get this to work for php5-extensions, you would need to patch the
 Makefile with:
 
 PLIST_FILES= libdata/php5/extensions
 PLIST_DIR= libdata/php5
 
 do-install:
 ${MKDIR} ${PREFIX}/libdata/php5
 ${TOUCH} ${PREFIX}/libdata/php5/extensions
 
 Then you could use:
 
 RUN_DEPENDS= 
 ${LOCALBASE}/libdata/php5/extensions:${PORTSDIR}/lang/php5-extensions
   
 Or what about just depending on the specific components of 
 php5-extensions? This would reasult in more lines, but would give you a 
 better way to fine-tune the dependencies and won't result in 
 unnecessarily installed ports for the user. Or do you really need the 
 whole bunch of extensions that php5-extensions offers?

No, I want to use the OPTIONS facility of php5-extensions so I cannot
depend on individual ports as each individual port may be disabled
during make config.

-- Paul Schenkeveld
___
freebsd-ports@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: DEPENDS -- is it time to remove it?

2007-12-11 Thread Paul Schenkeveld
On Tue, Dec 11, 2007 at 07:35:16AM -0600, Scot Hetzel wrote:
 On 12/11/07, Paul Schenkeveld [EMAIL PROTECTED] wrote:
  So now I have a need for a metaport to depend on another metaport.
  Without DEPENDS, how do I accomplish that.  Using RUN_DEPENDS with
  ${NONEXISTENT} seems not appropriate here, the Porters Handbook says
  that this should only be used to pull in source, not to install another
  metaport (unless it is already installed) and the effect of using
  something like RUN_DEPENDS=${NONEXISTENT}:${PORTSDIR}/lang/php5-extensions
  causes make install to try installing php5-extensions even when this port
  is already installed.
 
  The two metaport I need to depend on are php5-extensions and
  xorg-drivers and I really don't want to copy the OPTIONS processing of
  these ports and maintain that in the future.
 
 
 For xorg-drivers you simply use:
 
 RUN_DEPENDS= 
 ${LOCALBASE}/libdata/xorg/drivers:${PORTSDIR}/x11-drivers/xorg-drivers
 
 to get this to work for php5-extensions, you would need to patch the
 Makefile with:
 
 PLIST_FILES= libdata/php5/extensions
 PLIST_DIR= libdata/php5
 
 do-install:
 ${MKDIR} ${PREFIX}/libdata/php5
 ${TOUCH} ${PREFIX}/libdata/php5/extensions
 
 Then you could use:
 
 RUN_DEPENDS= 
 ${LOCALBASE}/libdata/php5/extensions:${PORTSDIR}/lang/php5-extensions

Thanks.

I was afraid it would turn out this way but this would make my port
non-portable to my customers unless I patch all their php5-extensions
Makefiles and make sure they won't get overwritten by a a future cvs
checkout.

Or try to ask the php5-extensions maintainer to add a dummy file or
dir to this meta-port for other meta-ports to depend on.

Anyway, it's a pity that the old DEPENDS functionality of depending on
a port instead of something installed by the port has gone.

Regards,

Paul Schenkeveld
___
freebsd-ports@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: DEPENDS -- is it time to remove it?

2007-12-11 Thread Pav Lucistnik
 On Fri, Jan 05, 2007 at 08:28:10PM +, RW wrote:

Blast from the past :)

 So now I have a need for a metaport to depend on another metaport.
 Without DEPENDS, how do I accomplish that.  Using RUN_DEPENDS with 
 ${NONEXISTENT} seems not appropriate here, the Porters Handbook says
 that this should only be used to pull in source, not to install another
 metaport (unless it is already installed) and the effect of using
 something like RUN_DEPENDS=${NONEXISTENT}:${PORTSDIR}/lang/php5-extensions
 causes make install to try installing php5-extensions even when this port
 is already installed.

Try

RUN_DEPENDS=some-metaport=0:${PORTSDIR}/dir/metaport

Documented at
http://www.freebsd.org/doc/en/books/porters-handbook/makefile-depend.html#AEN2283

-- 
Pav Lucistnik [EMAIL PROTECTED]
  [EMAIL PROTECTED]
It's ten o'clock; do you know where your processes are?


signature.asc
Description: Toto je digitálně	 podepsaná část	 zprávy


Re: DEPENDS -- is it time to remove it?

2007-01-05 Thread Vasil Dimov
On Thu, Jan 04, 2007 at 16:05:21 +0100, Erwin Lansing wrote:
 On Thu, Jan 04, 2007 at 02:01:46PM +, Matthew Seaman wrote:
  
  Dear all,
  
  The ports make infrastructure has a whole series of variables for showing
  how one port depends on another: LIB_DEPENDS, BUILD_DEPENDS, RUN_DEPENDS
  etc.  Amongst these is what looks to me like the vestigial remnant of
  an earlier phase of the ports: the plain, unmodified 'DEPENDS' variable.
  
 There is also a PR against this: ports/99724.  Unfortunately, it hasn't
 made it into an -exp run on pointyhat yet as other patches had higher
 priority.  It is on the TODO-list however, you are completely right that
 it should be removed.

One thing that can be done in the meantime is to remove DEPENDS from
those 3 ports that still use it. So things are nice and tidy when time
comes to edit bsd.port.mk.

-- 
Vasil Dimov
[EMAIL PROTECTED]
%
It is a simple task to make things complex, but a complex
task to make them simple.


pgpTgVEfA6dq7.pgp
Description: PGP signature


Re: DEPENDS -- is it time to remove it?

2007-01-05 Thread Andrew Pantyukhin

On 1/5/07, RW [EMAIL PROTECTED] wrote:

Isn't DEPENDS still a sensible way of making
one metaport depend on another. For example
if someone wanted to create a personal desk-
top metaport that depends on KDE, xorg etc.


People need programs, not ports. It's more
sensible to run_depend on files than just on
ports.
___
freebsd-ports@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: DEPENDS -- is it time to remove it?

2007-01-05 Thread RW
On Fri, 5 Jan 2007 20:52:50 +0300
Andrew Pantyukhin [EMAIL PROTECTED] wrote:

 On 1/5/07, RW [EMAIL PROTECTED] wrote:
  Isn't DEPENDS still a sensible way of making
  one metaport depend on another. For example
  if someone wanted to create a personal desk-
  top metaport that depends on KDE, xorg etc.
 
 People need programs, not ports. 

It's not that straightforward when you want to depend on a
metaport like KDE. All of the binaries can be provided by individual
sub-ports. The sensible thing to do is create a dependency on KDE and
let KDE's options/knobs handle the lower dependencies. The ports tree
doesn't need to have metaports depend on metaports, but some people
find it useful to create their own. 

It's more
 sensible to run_depend on files than just on
 ports.

Looking at the porter's handbook it looks like the solution is to use
RUN_DEPENDS with${NONEXISTENT}, so I guess DEPENDS is redundant.
___
freebsd-ports@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to [EMAIL PROTECTED]


DEPENDS -- is it time to remove it?

2007-01-04 Thread Matthew Seaman

Dear all,

The ports make infrastructure has a whole series of variables for showing
how one port depends on another: LIB_DEPENDS, BUILD_DEPENDS, RUN_DEPENDS
etc.  Amongst these is what looks to me like the vestigial remnant of
an earlier phase of the ports: the plain, unmodified 'DEPENDS' variable.

This seems to have an effect apparently like all of the other DEPENDS
variants rolled together, but unlike the others you can't give it a
filename or a shlib to use as a test that whatever it references has
been installed, so it always causes its target to be installed.

The Porter's Handbook says (Section 5.7.10):

   Do not use DEPENDS unless there is no other way the behaviour you want
can be accomplished. It will cause the other port to always be built
(and installed, by default), and the dependency will go into the
packages as well. If this is really what you need, you should probably
write it as BUILD_DEPENDS and RUN_DEPENDS instead--at least the
intention will be clear.

Plus it seems that there is nowadays a grand total of just 3 ports out
of 16,300 or so that actually uses this variable:

happy-idiot-talk:/usr/ports:% grep -re '^DEPENDS\' .
./audio/klira/Makefile:DEPENDS= ${PORTSDIR}/misc/kdehier[*]
./www/caudium12/Makefile:DEPENDS=   ${PORTSDIR}/lang/pexts/
./x11/kdelibs3/Makefile:DEPENDS=${PORTSDIR}/misc/kdehier

Practically, removing DEPENDS would simplify (slightly) the logic in
bsd.port.mk and make generating the INDEX-n file a bit faster.

Cheers,

Matthew

[*] audio/klira is an oddity: all but two other KDE apps inherit their
dependency on kdehier through depending on kdelibs3.  The two other
exceptions, both of which have a RUN_DEPENDS directly on kdehier, are
math/proofgeneral (deprecated, due for removal on 2007-02-07) and
sysutils/kports (which also depends on kdelibs, so doesn't need the
direct dependency as well).

-- 
Dr Matthew J Seaman MA, D.Phil.   7 Priory Courtyard
  Flat 3
PGP: http://www.infracaninophile.co.uk/pgpkey Ramsgate
  Kent, CT11 9PW



signature.asc
Description: OpenPGP digital signature


Re: DEPENDS -- is it time to remove it?

2007-01-04 Thread Erwin Lansing
On Thu, Jan 04, 2007 at 02:01:46PM +, Matthew Seaman wrote:
 
 Dear all,
 
 The ports make infrastructure has a whole series of variables for showing
 how one port depends on another: LIB_DEPENDS, BUILD_DEPENDS, RUN_DEPENDS
 etc.  Amongst these is what looks to me like the vestigial remnant of
 an earlier phase of the ports: the plain, unmodified 'DEPENDS' variable.
 
There is also a PR against this: ports/99724.  Unfortunately, it hasn't
made it into an -exp run on pointyhat yet as other patches had higher
priority.  It is on the TODO-list however, you are completely right that
it should be removed.

Cheers,
-erwin

-- 
Erwin Lansing http://droso.org
Security is like an onion.  (o_ _o)
It's made up of several layers   \\\_\   /_///[EMAIL PROTECTED]
And it makes you cry.) ([EMAIL PROTECTED]


pgptQ5Eb8ONbt.pgp
Description: PGP signature