I was thinking about this and it should be quite simple
to check for a '/' at the start of the string, indicating
a full-path (on Linux, OSX, or a "*:/" on Windows), in
which case the current dir is not used. Otherwise, paths
are relative to the current dir.

Victor

>
> Hi Victor, Chuck...
>
> The current behavior of csoundapi CVS seems to be the best
> for my   purposes.  Rarely do I use csoundapi to open
> multiple csds within one   patch.  I would assume that
> most would use the object to create   custom interfaces
> for specific csound orchestras, as I do.  For this
> purpose, relative paths seems to be the best solution.
>
> I have not the skills in c to adequately read the behavior
> of   csoundapi, nor do I know much about the internals of
> PD.  The current   implementation, which I requested of
> Victor, best matches my uses of   the object.  (Thanks
> Victor-- to my ends it works great!).  It might   not be
> the most "PD" of implementations, however.
>
> PD has a built-in path mechanism for the locating of
> objects and   abstractions.  A quick test shows that the
> built-in textfile object   also uses this path mechanism
> to search for files (ie, the command   [read test.txt(
> locates the file if it is located in the same dir as   the
> parent patch or if it is located in any of the paths
> specified by   the --path command line flag).
>
> I would think that an ideal implementation of csoundapi
> would harness   this same built-in mechanism; however,
> since I don't know the full   details of how this is
> implemented internally, I have no sense of   weather or
> not this is difficult or even possible.
>
> I hate to send more work your way Victor, since I am
> completely happy   and satisfied with the current
> solution.  It just seems that if there   are to be more
> changes, it should be to further the object's
> compatibility with PD's current system.
>
> Thanks, all, for your continued work to improve csound.
> It's perhaps   not said enough how much we all appreciate
> the developers' time and   devotion.
>
> JP
>
>
>
> On Aug 3, 2007, at 12:11 PM, Victor Lazzarini wrote:
>
> > But that changes the behaviour of csoundapi~. I have
> > done that following advice by Julian Peterson on  how
> > PD searches path. By the way, with that behaviour you
> > can still use full paths, but they are now relative
> > to the current directory (so you would do something
> > like ../../topdir/csds/etc.csd).
> >
> > You have reverted basically the old behaviour which
> > I was asked to change.
> >
> > Victor
> >
> >
> >
> >
> >>
> >>
> >> I'm proud of myself.  Instead of bothering you with
> >> something, I fixed it. Sending an "open" message to
> >> csoundapi~ would say that
> >> /home/chuckk//home/chuckk/newpd.csd was not found.  I
> went >> into the source and commented out everything about
> strcpy >> of curdir->s_name and "/", and changed strcat on
> the next >> line to strcpy, in two places (I don't have it
> in front of >> me).  After trying a few things a few times
> , it now loads >> correctly.  I don't mind typing a whole
> path name into the >> object.
> >>
> >> -Chuckk
> >>
> >>
> >> On 8/2/07, Victor Lazzarini 
> >> wrote: >
> >>> I'm glad it's fixed.
> >>> No worries about bothering me. Nothing much to do
> >>> between walks...
> >>>
> >>> Victor
> >>>
> >>>>
> >>>>
> >>>> On 8/2/07, Victor Lazzarini
>  >>>> wrote: >
> >>>>> I think I found the bug: in csoundapi_tilde.c
> >>>>> lines 84 & 263
> >>>>>
> >>>>> static t_int *csoundapi_perform(int *w)
> >>>>>
> >>>>> should be
> >>>>>
> >>>>> static t_int *csoundapi_perform(t_int *w)
> >>>>>
> >>>>> and line
> >>>>>
> >>>>> int     i, n, end = x->end, run = x->run;
> >>>>>
> >>>>> should be
> >>>>>
> >>>>> t_int     i, n, end = x->end, run = x->run;
> >>>>>
> >>>>> I am positive this is the bug (it is A bug in
> >>>>> 64bit CPUs, so I expect it to be the cause of
> >>>>> your crashes). I can't change in CVS or test it
> >>>>> now, will do when I am back to work.
> >>>>>
> >>>>> Chuck, can you test it for me?
> >>>>
> >>>>
> >>>> You are correct, it works.  I see that it has been
> >>>> changed in CVS, too. Thanks, Victor, I shan't bother
> >>>> you again during your break.  I look forward to
> >>>> bothering you afterward.
> >>>>
> >>>> -Chuckk
> >>>>
> >>> --
> >>> Send bugs reports to this list.
> >>> To unsubscribe, send email to
> >> csound-unsubscribe@lists.bath.ac.uk >
> >>
> >>
> >>
> >> --
> >> http://www.badmuthahubbard.com
> >>
> > --
> > Send bugs reports to this list.
> > To unsubscribe, send email to
> csound-unsubscribe@lists.bath.ac.uk
>
> --
> Send bugs reports to this list.
> To unsubscribe, send email to
> csound-unsubscribe@lists.bath.ac.uk