Am 19.09.2013 17:50, schrieb IOhannes m zmoelnig:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 2013-09-19 17:18, hermann meyer wrote:

Am 19.09.2013 16:50, schrieb IOhannes m zmoelnig:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 2013-09-19 05:31, hermann meyer wrote:

I'm sad to hear that. :-( Please don't let you lead from the
things you didn't like, let you lead from the things you like
instead. I guess then it's necessary to let you know that we
use /as well/ a fork of your work, the zita-convolver library,
in the guitarix project. But we leave your copyright untouched,
and the fork will only come in use, when the user set a
explicit compile flag. We didn't promote it, or force the fork.
Ordinary your original code is in use. We do it to use ffmpeg
instead fftw3 FFT, which perform better on ARM devices.

but this sounds like the perfect opportunity to not do a simple
fork, but to send patches to upstream so fons' aeolus could
support both fftw3 and ffmpeg FFTs. it might be a win-win
situation, where not only more than just the original aeolus
users can profit from fons' work (because you use his code) but
also more than just your users can profit from your work (because
you changes are included into upstream aeolus).

Sorry, you miss-match aeolus and zita-convolver here.

ah yeah, sorry. pleas do s/aeolus/zita-convolver/g in my answer.

As well, we didn't "add" ffmpeg support to zita-convolver, we
replace fftw3 with ffmpeg, and only use this under special
circumstances (ARM support),

i'm not sure what you mean.
zita-convolver depends on FFTW3 for doing the FFTs.
if you "replace fftw3 with ffmpeg", you create modifications to
zita-convolves to not use fftw3 anymore (in some special
circumstances), and instead use ffmpeg's FFT (whether as a library or
statically compiled in).

in any case, this seems to be a substantial change to some parts of
zita-convolver, so you should definitely add your copyright to the
files you modified (substantially).

No, a disclaimer in the source tree which clarify this will be enough. 

we didn't distribute it as "zita-convolver library"

so?
it seems that you did distribute it "in some form" (else what are we
talking about?).

MuseScore doesn't distribute a disfigured version of "Aeolus" in order
to pile praise upon themselves and/or to denounce fons. they
distribute "MuseScore" which comes bundled with a version of Aeolus
that integrates into their system.
similar goes for Maurizio: he has a project called "oscAeolus" which
makes it quite clear that it is a derivative of Aeolus (and not the
real thing itself).

That's why I wrote

On 2013-09-19 05:31, hermann meyer wrote:

.... I guess then it's necessary to let you know that we use /as well/ a
fork of your work, the zita-convolver library, in the guitarix
project.

we do exactly the same, we modify the source to our needs.

We have done the same with zita-resampler for some time, as long Fons takes to include the provided patch.

No useful patch is available.

i didn't say that contributing is easy.
more often than not it requires a lot of work and interaction with
upstream until you have prepared a patch that can be accepted.

what i did say is that you should make a useful patch available.

This is pretty new, and we have a lot stuff where we are working on. ;-)
I could provide a diff, which shows the changes.

i'd love to see a zita-reverb that performs "ideally" on all platforms
(and that unfortunately includes ARM-systems by now)

I'm not aware of any relation between zita-reverb and zita-convolver.

greets
hermann