On 12/6/23 11:07, IOhannes m zmoelnig wrote:
now that i've written this, i notice that I haven't actually uploaded
any of my new packages to deken yet (that's mostly because i haven't
done any new release yet).
at least i've fixed the upload for zexy for now.
others might follow.
thanks.
fixed in PR https://github.com/pure-data/pure-data/pull/2156
--
Mensaje telepatico asistido por maquinas.
On 06/12/2023 12:45, IOhannes m zmölnig wrote:
my_lib.darwin-amd64-64.so
the .so extension is still common on macOS (being un*xy),so I/we went for a
minimum number of (native)
Am 6. Dezember 2023 12:16:44 MEZ schrieb Lucas Cordiviola
:
>@IOhannes in the HTML docs I filled the new "deken-scheme" naming for macOS
>with file extension ending with .dylib. is this incorrect?
>
>I mean `my_lib.darwin-amd64-64.dylib` will load? or should be
>`my_lib.darwin-amd64-64.so`
>
@IOhannes in the HTML docs I filled the new "deken-scheme" naming for
macOS with file extension ending with .dylib. is this incorrect?
I mean `my_lib.darwin-amd64-64.dylib` will load? or should be
`my_lib.darwin-amd64-64.so`
?
--
Mensaje telepatico asistido por maquinas.
On Wed, 2023-12-06 at 11:16 +0100, IOhannes m zmoelnig wrote:
> On 12/6/23 09:12, Roman Haefeli wrote:
> > Just compile it against the m_pd.h of the double-precision edition
> > of
> > Pd.
>
> this won't work, as the "m_pd.h" for Pd64 and Pd32 are exactly the
> same.
>
> instead set the
On 12/6/23 09:44, Lucas Cordiviola wrote:
On 06/12/2023 05:27, Alexandros Drymonitis wrote:
Sorry for the second question, now read the pd-lib-builder README. It
states that 64-bit externals will indeed get a different name.
essentialy you need to do 2 compilations (no need to use
On 12/6/23 09:12, Roman Haefeli wrote:
Just compile it against the m_pd.h of the double-precision edition of
Pd.
this won't work, as the "m_pd.h" for Pd64 and Pd32 are exactly the same.
instead set the "PD_FLOATSIZE" macro to 64 when building the external.
reasoning:
- just building with
On 12/6/23 08:43, Alexandros Drymonitis wrote:
Ahem, I should have guessed that...
The question now is how to have both versions run happily side by side?
for Pd on Debian the answer is:
- install both "puredata" and "puredata64"
- if you are a cmdline person, launch pd32 by typing "pd",
Darn,
I think I made a typo when compiling. I tried another external of mine
and it worked, and then I copied the `make` command to the neuralnet
directory and now [neuralnet] loads fine. Sorry for the noise. Here's
the command, for anyone interested:
`make floatsize=64
On 06/12/2023 06:19, Alexandros Drymonitis wrote:
Yes, I'm using the latest pd-lib-builder (0.7.0), and compiled with
floatsize=64 and extension=linux-amd64-64.so. I also use t_float
instead of float in my code. I just checked the .c file to make sure.
what does the console says if you
Yes, I'm using the latest pd-lib-builder (0.7.0), and compiled with
floatsize=64 and extension=linux-amd64-64.so. I also use t_float instead
of float in my code. I just checked the .c file to make sure.
On 12/6/23 11:06, Lucas Cordiviola wrote:
Oh and also: your code should use t_float instead
Oh and also: your code should use t_float instead of float and t_sample
for audio floats.
--
Mensaje telepatico asistido por maquinas.
On 06/12/2023 05:59, Lucas Cordiviola wrote:
On 06/12/2023 05:47, Alexandros Drymonitis wrote:
Actually I just tried that with my [neuralnet] and it didn't
On 06/12/2023 05:47, Alexandros Drymonitis wrote:
Actually I just tried that with my [neuralnet] and it didn't work for
Pd64. I do get a neuralnet.linux-arm64-64.so, but puredata64 can't
load it.
did you use "floatsize=64":
make floatsize=64 extension=linux-arm64-64.so
?
are you using
Actually I just tried that with my [neuralnet] and it didn't work for
Pd64. I do get a neuralnet.linux-arm64-64.so, but puredata64 can't load it.
On 12/6/23 10:44, Lucas Cordiviola wrote:
On 06/12/2023 05:27, Alexandros Drymonitis wrote:
Sorry for the second question, now read the
On 06/12/2023 05:27, Alexandros Drymonitis wrote:
Sorry for the second question, now read the pd-lib-builder README. It
states that 64-bit externals will indeed get a different name.
essentialy you need to do 2 compilations (no need to use different
m_pd.h as they are the same)
then you
What worries me most, with FluCoMa in head, is what is the horizon to support
this - in other words, when is 32 and 64 will be shipped as equal.
There is no more dev resources apart from some free time by yours truly for all
the stuff so I need to plan carefully… in years! So if anyone on the
Sorry for the second question, now read the pd-lib-builder README. It
states that 64-bit externals will indeed get a different name.
On 12/6/23 10:21, Lucas Cordiviola wrote:
there is a name convention to support both float types in the same path.
> You could configure different install and search paths for single- and
> double-precision versions of Pd. I don't know if they share the config
> file. If they do, you would have to adjust the config whenever you
> switch between them.
they do share the same config file on macOS at least. i
But this states that the binaries have different names, depending on the
architecture. When installing an external from deken, it doesn't seem to
be the case. Also, will this be the case when compiling for 64-bit, the
way Roman suggested?
On 12/6/23 10:21, Lucas Cordiviola wrote:
there is a
there is a name convention to support both float types in the same path.
https://msp.ucsd.edu/Pd_documentation/x4.htm#s1.2.1
If you are using lib-pd-builder see:
https://github.com/pure-data/pd-lib-builder/blob/master/tips-tricks.md#building-double-precision-externals
--
Mensaje
On Wed, 2023-12-06 at 09:43 +0200, Alexandros Drymonitis wrote:
>
> The question now is how to have both versions run happily side by
> side?
> On the 64-bit version I did find zexy (but some other libraries I
> searched for, including my own, neuralnet, were not available - not a
> surprise
And here keeping an eye too for simple double-support I hope...
> On 6 Dec 2023, at 07:43, Alexandros Drymonitis wrote:
>
> Ahem, I should have guessed that...
>
> The question now is how to have both versions run happily side by side? On
> the 64-bit version I did find zexy (but some other
Ahem, I should have guessed that...
The question now is how to have both versions run happily side by side?
On the 64-bit version I did find zexy (but some other libraries I
searched for, including my own, neuralnet, were not available - not a
surprise for my own, I haven't compiled it for
23 matches
Mail list logo
