Edit: This post was originally written with Pd-extended 0.43 in mind. These days there is a nice little plugin called Deken that you can use to easily download and install most of the popular libraries in Pd. It comes bundled with recent versions of Pd-vanilla. You can still use namespaces to load up the objects for most libraries, but for some libraries like zexy that have all the objects bundled into a single file, you will still need to use [declare] to load it up.
Installing libraries and plug-ins isn't so obvious. This guide will hopefully clear up the process, and may even explain more than you need to know.
Recent changes
This first section is mainly for those upgrading to Pd-0.43 from previous versions. If you're new to Pd and are starting on 0.43, you can just skip this section and get on with loading libraries.
It use to be that installing a library was done by putting it wherever you wanted on your system and adding its path to the preferences. While this wasn't particularly difficult, it did have some problems. Recent efforts have been made to standardize this a bit so those problems can be avoided.
In Pd-extended 0.43, you can't add paths from within Pd any more like you used to. You can open the preferences and add a path, but it won't save it. This means that non-default libraries won't load on start-up any more; you have to load them from within the patch. While this sounds like a hassle, it does have the advantage that patches are much easier to share. You don't have to worry about others having different start-up settings than you. It also forces the patch to document what libraries are being used, so if an object isn't loading, you don't have to go on some crazy hunt to find out where it came from or what it's supposed to do.
Putting them in the right folder
To make sure Pd can easily find the library, you need to put them in the right folder. This is OS dependent. The following folders are your options (taken from here). I personally recommend putting them in the user-specific folders since they are likely easier to migrate when updating your OS.
Linux:
User-specific:
~/pd-externals
Global:
/usr/local/lib/pd-externals
Mac OSX
User-specific:
~/Library/Pd
Global:
/Library/Pd
Windows
User-specific:
%AppData%/Pd
Global:
%CommonProgramFiles%/Pd
Linux and OSX are pretty straight-forward. "~" means your home folder.
Windows is a bit trickier. %AppData% and %CommonProgramFiles% are environment variables that contain specific directories. What the exact directory is depends on which version of Windows you are on. You can find out what it is by opening up the command line and typing this:
echo %AppData%
You might end up with something like C:\Users\username\AppData\Roaming or C:\Documentes and Settings\username\Application Data. Also, to see it in the file browser, you might need to show hidden files. At any rate, once you find the directory, create a Pd folder if there isn't one, and put your library, objects, or plug-ins there.
Loading whole libraries
To load a library you have a few options. The one that works on both vanilla and extended is to use [declare -stdpath libraryname]. The libraryname is the name of the folder that all the objects/abstractions are in.
[declare] doesn't always work as expected in abstractions, though. In Pd-extended, you can also use [import], which works well in abstractions. I have found that sometimes it doesn't work with libraries of abstractions that don't have a meta file, though. But that can be fixed by just making one. Simply create an empty Pd patch and save it as libraryname-meta.pd in the library's folder. That's all the meta patch is.
A third option (also Pd-extended only) is to use [path]. This is new to Pd-extended 0.43 and works similarly to [declare]. I haven't needed it, yet. But it's there.
Once the library is loaded, you should be able to load an object or abstraction just by creating it by name.
Loading objects or abstractions using namespaces
While this method means a little more typing, it is also the safest, and it makes it easier to find out what objects are from which library. The way you do it is add the library name (i.e. the namespace) to the beginning of the object when you call it, like this:
[zexy/multiplex~] <-- load the [multiplex~] object from zexy
[cyclone/comb~] <-- load [comb~] from cyclone
You don't have to load the library with [import] or [declare] for this to work. You just call the object that way.
Why is this safest? Because it's possible for different libraries to have objects with the same name. For example, both cyclone and zexy have an object called [mean]. But they don't work exactly the same. zexy's takes a list, while cyclone's takes in a continuous stream. If you have both libraries loaded (and in Pd-extended, these two libraries happen to be loaded by default), you can't be sure which one you're getting if you just type [mean]. However, if you type [cyclone/mean], you know exactly which one you are getting.
