On 2012-01-06, Stuart Henderson <[email protected]> wrote:
> On 2012-01-06, Pavel Korovin <[email protected]> wrote:
>> Unfortunately I was unable to simplify the Makefile diff.
>
> Sure you can.
>
> +.if !${BUILD_PACKAGES:M-pgsql}
> +BUILD_PACKAGES := ${BUILD_PACKAGES:N-pgsql}
> +.endif
>
> I don't know where you got this and a bunch of similar lines
> but they do nothing useful.
>
>> Although
>> people are supposed to use binary packages, I'm modestly taking the
>> view that it's the matter of taste. Ports now have a wonderful
>> mechanism of PSEUDO_FLAVORS, why not use it?
>
> Ports have had pseudo flavours for ages. It complicates testing and
> makes the Makefile harder to read. Sometimes it is useful enough that
> it's worth the extra pain. Sure, let some of the bigger stuff have a
> switch. But rrdtool? dbi? what's the point?
>
>> I moved all this "non-official" stuff into a single patch and added
>> additional PACKAGE/PSEUDO_FLAVOR for it.
>
> I don't see this. Wrong diff attached? Sounds nasty though, the issue

Ah, munpack automatically renamed the attachment to avoid overwriting
the existing file from when I looked at this before. (Inline plaintext
diffs are preferred...)

Might it make more sense for this plugin to be a separate port?



> with adding any more intrusive unofficial patches to any port is that
> it can really get in the way of updates, hiding them behind a switch
> doesn't change this.
>
> And the more options you have, the more things have to be tested
> *each update*.
>
>
>

Reply via email to