[PD] save search path 0.43 OSX

[Roman Haefeli]

On Fri, 2012-02-24 at 12:14 -0800, Jonathan Wilkes wrote:
> ----- Original Message -----
> > From: Roman Haefeli <reduzent at gmail.com>
> > To: m.e.grimm <megrimm at gmail.com>
> > Cc: Jonathan Wilkes <jancsika at yahoo.com>; pd-list <pd-list at iem.at>
> > Sent: Friday, February 24, 2012 2:57 PM
> > Subject: Re: [PD] save search path 0.43 OSX
> > 
> > On Fri, 2012-02-24 at 11:37 -0500, m.e.grimm wrote:
> >>  > I think the better way to fix those help-files is to use an [import] 
> > or
> >>  > [declare] object in the help patch.
> >> 
> >>  one prob I have found ... in a lib such as "rtc" the objects are 
> > not
> >>  compiled externals but abstractions that rely on "list-abs". how 
> > to
> >>  deal with this? you can't really put [import list-abs] or [declare] in
> >>  an abstraction ... well I guess you can buts that's not the way it
> >>  should be done.
> > 
> > Interesting that you say that. I always thought it is the very goal of
> > using [import]: make any patch or abstraction resolve its own
> > dependencies.
> > 
> > In what way [import] shouldn't be used inside abstractions?
> > 
> > (I specifically mention only [import] now, since [declare] has its own
> > implications, though if it would be free of bugs, I'd mentioned it as
> > well.)
> 
> So we have [import] which isn't in vanilla, [declare] which you say has 
> bugs, and using "libname/" prefixes which works for both vanilla and 
> extended.  What am I missing?

Many libraries  come as multi-class externals, either because you
compiled  them yourself and this is the default setup designed by the
developer or you get them as package (in Debian, for instance). For all
those libraries, [libname/classname] will simply break. OTOH, [declare]
already works now (in both, Pd-vanilla and Pd-extended) [1], or you
could use [import] which you can easily install (for instance, in Debian
there is already a package). 

I think, it's much easier to find a way to load a certain library
(either with start-up flags, [declare] or [import]) than to have to edit
a patch and make it work by replacing all occurences of
[libname/classname] by [classname].

Roman



[1] The one known bug in [declare] I was thinking about doesn't affect
its functionality, it works fine, but when used within abstractions, it
will add a bogus line in the parents patch file, when the parent is
saved. People need to be aware of that when using [declare].