cvs commit: ports/multimedia/acidrip Makefile ports/math/algotutor Makefile ports/sysutils/apt Makefile ports/graphics/asciio Makefile ports/devel/autodia Makefile ports/www/awffull Makefile ports/devel/bazaar Makefile ports/www/bigbluebutton Mak

[Olli Hauer]

On 2012-01-22 17:00, Eitan Adler wrote:
> 2012/1/22 Alexey Dokuchaev <danfe at freebsd.org>:
>> On Sat, Jan 21, 2012 at 05:40:15PM +0000, Eitan Adler wrote:
>>> eadler      2012-01-21 17:40:15 UTC
>>>
>>>   Log:
>>>   At the moment 1385 ports use BUILD_DEPENDS= ${RUN_DEPENDS} and 450
>>>   ports use BUILD_DEPENDS:= ${RUN_DEPENDS}. This patch fixes ports that are
>>>   currently broken. This is a temporary measure until we organically stop
>>>   using := or someone(s) spend a lot of time changing all the ports over.
>>
>> Thanks for using correct way of assigning rdeps to bdeps.  Can you elaborate
>> on the last sentence of the log?  Why should we stop using :=?  

Hm, if I read the diffs Eitan has changed from `=' to `:=' so I think you mean
 Why should we stop using `='?


>> Instead, we should fix portlint(1) to offer this obvious thing instead of 
>> duplicating two variables like it bogusly does now.
> 
> The concern is that someone would add something extra to BUILD_DEPENDS
> without thinking about it and thus add extra bogus RUN_DEPENDS. I'm
> uncertain if there other additional reasons that portmgr dislikes :=


It doesn't matter if you have RUN_DEPENDS=${BUILD_DEPENDS} or RUN_DEPENDS:=${BUILD_DEPENDS},
if someone adds additional `something' to BUILD_DEPENDS then it is in both cases also in RUN_DEPENDS but `:=` preserves including additional dependencies from `something'.

If I see RUN_DEPENDS=${BUILD_DEPENDS} the first I do is a cross check with `make -V RUN_DEPENDS' and a second check if I change to `:='
Most every time the usage of `:=' stops double including perl and others to the RUN_DEPENDS list and I change then to `:='.

Maybe the following section in the PH will make things clearer.
http://www.freebsd.org/doc/en_US.ISO8859-1/books/porters-handbook/makefile-depend.html#AEN2173

Quote:
A quite common situation is when RUN_DEPENDS is literally the same as BUILD_DEPENDS, especially if ported software is written in a scripted language or if it requires the same build and run-time environment. In this case, it is both tempting and intuitive to directly assign one to the other:

RUN_DEPENDS=   ${BUILD_DEPENDS}

However, such assignment can pollute run-time dependencies with entries not defined in the port's original BUILD_DEPENDS. This happens because of make(1)'s lazy evaluation of variable assignment. Consider a Makefile with USE_* variables, which are processed by ports/Mk/bsd.*.mk to augment initial build dependencies. For example, USE_GMAKE=yes adds devel/gmake to BUILD_DEPENDS. To prevent such additional dependencies from polluting RUN_DEPENDS, take care to assign with expansion, i.e., expand the value before assigning it to the variable:

RUN_DEPENDS:=  ${BUILD_DEPENDS}