hello gang! i ve made some externals for a project of mine on my imac with Xcode following the "how to" on puredata info site : http://puredata.info/docs/developer...

all worked well..

I now want to port what I did on linux (ubuntu studio 10.04 amd64), but i cant seem to fin d the corresponding "how to" for linux.. would the procedure be different for each distro of linux? where could i find help?

thanks in advance..

Hello! I would like to zero-pad a signal that I fourrier transform with rfft~ and block~. I tried cutting the signal in pieces by using tabwrite~ and del objects outside block~s subpatch but its very complicated just for the sake of zero-padding, plus I cant use block~s overlapp functionality anymore... Does anyone have a clue?? I must be missing something on this... I want to zero-pad to get some interpolation in frequency between the normal zero-pad free fourrier transform bins.. I know its the best solution for what I want to do caus I ve tested multiples methods in octave, including linear interpolation...

I m desperate to get my external working on a PC platform under linux or windows XP... I ve installed ubuntu studio but my M-audio fast track pro doesnt seem to work in duplex.. so I guess I ll try to port my external from OS X to windows XP.. i found info on this forum for that matter:

old post for how to under windows:
http://puredata.hurleur.com/sujet-1029-problem-compiling-external-windows
http://puredata.hurleur.com/viewtopic.php?pid=3706#p3706
http://www.youngmusic.org/wiki/index.php/PD-linking-tutorial

those doesnt work caus there is no pd.lib to be found under the pd/bin directory...

does any one know where to find an up to date tutorial for externals under windows??

for info, embeding two subpatches with different blocksize doesnt do the trick it seems.. Maelstrom's solution works

I want to add samples... but I guess your idea could work by doubling the window lenth and overlapp specified by block~... then i could use a window that cancel out half of the window as you said.. thanks for the hint..

I wonder if embeding two pd subpatchs with different blocksize could do the trick... for example a first pd subpatch with a blocksize of 2048 and a second pd subpatch within the first pd subpatch with a blocksize of 4096... I think this could imply less floating point multipy operations.. i ll keep the post informed if this is successfull.. block~ object is not extensively documented so its really try and error..