On 02/05/2013 04:30 AM, Roman Haefeli wrote:
On Mon, 2013-02-04 at 19:58 +0100, Òscar Martínez Carmona wrote:
hey, still having problems with that, by now I'm doing it with the absolute filepath... maybe the solution it'll be making the main applicattion finding out the f*cking path and sending the whole thing to pd via OSC, or maybe trying it another day!
If I am not mistaken, it hasn't been mentioned yet (though IOhannes assumed it very early) in this thread that [oggread~ ] oddly reads relative to Pd's start location (unlike many other classes like [textfile] or [readsf~ ] which read relative to the patch's location).
IMHO, this makes it very difficult for a patch writer to use relative paths as the patch doesn't have any notion of where Pd was started from. I consider the whole idea of reading relative to Pd's start location flawed. A similar case is the 'open patch.pd path' message to [s pd]. Also this one reads relative to Pd's start location. However, considering that it was implemented this way, because it probably originates from the '-open' commandline flag, where it makes sense to use a path relative to the current working directory for loading a patch, this one is excused.
For you, this means if your OSC application knows where Pd was started from, you need to make it use a path relative to that location. Otherwise you you're left with using absolute paths. When dealing with objects like [oggread~], I'd go for absolute paths, it's just seems safer and saner to deal with.
(or someone fixes [oggread~ ], which I even wouldn't consider to break backwards compatibility as the current implementation doesn't really allow to use relativ paths in meaningful way)
Roman
I agree, relative should be relative to the patch. Please file a bug report on that.
.hc