cvs commit: ports/mail/p5-Mail-SpamAssassin Makefile distinfo pkg-plist

[Doug Barton]

Erwin Lansing wrote:
> On Mon, Jan 19, 2004 at 07:51:29PM +0100, Mathieu Arnold wrote:
> 
>>+-Le 19/01/2004 10:45 -0800, David O'Brien écrivait :
>>| 
>>| Well, I guess that opens up a discussion then.  I don't use many "p5"
>>| ports, but I certainly do use SpamAssassin daily.  So I hadn't noticed
>>| that they go against the Porters Handbook and 'portlint'.  It is
>>| important to keep things all in the proper section so a user like me can
>>| know what dependencies there are.  Often I install dependencies from a
>>| precompiled package before I add local patches to the "leaf" port that I
>>| ultimately want installed.
>>
>>Well, portlint will complain if you do things badly, but it won't complain
>>when you do just that.
>>perl ports have been done this way for ages, I mean, the normal
>>dependencies at the right place, and the dependencies depending on perl
>>version at the end. I was told it was the right way to do it when I began
>>updating ports, so I though it was the right way.
>>
> 
> I'd say it's the only way, so unless someone finds a better way to make
> dependencies conditional, this is the right way.

Um, saying this is "the only way" is just plain silly. It worked just 
fine the way it was, so what we're talking about is style, not 
functionality.

What overwhelming problem is caused by including bsd.port.pre.mk early 
that needs to be solved by violating style guidelines followed by the 
rest of the ports tree?

> make -V RUN_DEPENDS will give you the dependencies you want after
> evaluating the conditions. I would recommend using this instead of
> reading the raw Makefile.

Personally I find that 'make clean' does a pretty good job most of the time.

Doug