On Wed, 9 Feb 2005, Tim Blechmann wrote:
i've set up a wiki to collect topics for the developer meetings on irc: http://puredata.org/Members/timblech/developerwiki feel free to add topics, that are supposed to be discussed there ... it can also be used to propose a date ...
Apparently, in the meantime, this has moved to:
http://puredata.org/Members/timblech/TopicsForTheDeveloperMeetingsOnIrc/
******************************************************************* * NOTE: THE MEETING WILL BE HELD ON FRIDAY FEBRUARY 25, UTC 17.00 * *******************************************************************
This is the FINAL date, because the people we delayed the meeting for have, to my knowledge, not proposed any other date. (contact me if you believe i may have accidentally skipped a mail on this...)
Tim's list of topics (from the link above): -------------------------------------------
fft, position of the nyquist bin, fftw, sign of the complex part of rfft~ irfft~
behaviour of audio arrays (references to _array or a_vec?)
adding callback-based scheduling?
memory locking resulting in segmentation faults
locking mechanism for g_arrays to make a threaded soundfiler threadsafe
threadsafety of pd in general (e.g. improved idle callbacks)
alsa midi support (higgs: I'll even raise my hand and offer to start this if pd-dev likes.)
Matju's list of topics (from the link above): ---------------------------------------------
devel_0_38 bug in loading of externals (mystery and major hurdle)
distribution of devel_0_38 (tarballs, rpm's, etc.)
unit-testing in pd : how to ensure pd's functionalities work, and continue to work even when changes (major or minor) are made to pd itself or any supporting libraries
improving error-handling in pd (important if unit-testing of pd is done using pd patches); this would consist of a way for a pd patch to trap an error message and process it. For example, sending a message to [error]? would make that same message go out of left inlet, except that if ever pd_error() is called before returning, the message gets sent by right-outlet of [error]? instead of (or in addition to) being printed in the console.
fixing miller's console: how to make it optional? not all error messages are being rerouted to the Pd console, and when Pd dies early there is no way to see those messages. Matju's console is optional, but is found only in impd_0_37 and would cvs-conflict with Miller's. Also: control the urge to remove the term-window, or at least tell people how they're supposed to reenable it in case of need. (or figure a way to reroute stdout/stderr portably! no fork() in mingw)
integration of impd_0_37 to devel_0_38; especially, making sure the merge of miller_0_39 and devel_0_38 will be smooth enough, so that less efforts get wasted.
_____________________________________________________________________ Mathieu Bouchard -=- Montréal QC Canada -=- http://artengine.ca/matju