xmp-devel — Discussions on xmp development

Re: [xmp] Libxmp 4.3.8 released

[Claudio M. <cma...@gm...>]

Uhm, we had a small problem uploading the package to sourceforge so please
make sure you're downloading the correct packages by checking their md5sum:

7bfa2c9dc2a0fca91e8500a29e49b734  libxmp-4.3.8.tar.gz
0bfb94efcca07cd9286c9072c9e99946  libxmp-lite-4.3.8.tar.gz


On Sat, Apr 4, 2015 at 8:17 AM, Claudio Matsuoka <cma...@gm...>
wrote:

> Hi,
>
> Libxmp 4.3.8 is available, and it fixes many issues related to envelopes.
> We finally changed envelope indexes to pre-incremented, allowing many IT
> test cases that fail with post-incremented indexes to work correctly.
> Envelope processing code was also rewritten to get rid of old legacy code
> that didn't make sense (but worked for some unknown reason).
>
> Envelope-related OpenMPT test cases that now work correctly include
> EnvLoops.it, EnvLoopEscape.it, EnvOffLength.it and s77.it. EnvLoops.it is
> pictured below (top: libxmp, bottom: Impulse Tracker).
>
> [image: Inline image 3]
>
> The Changelog (rather short this time):
>
> 4.3.8 (20150404):
>         Fix bugs caught in the OpenMPT test cases:
>         - fix pre-increment of envelope indexes
>         - fix IT note release at end of envelope sustain loop
>         - reset channel flags in case of delay effect
>         Other changes:
>         - fix MMD3 16-bit samples (reported by jbb666)
>         - refactor XM envelopes
>         - refactor IT envelopes
>
> And the packages:
>
> https://sourceforge.net/projects/xmp/files/libxmp/4.3.8/
>
> Please report if you find any regressions!
>
> Best regards,
> Claudio
>

[xmp] Libxmp 4.3.8 released

[Claudio M. <cma...@gm...>]

Hi,

Libxmp 4.3.8 is available, and it fixes many issues related to envelopes.
We finally changed envelope indexes to pre-incremented, allowing many IT
test cases that fail with post-incremented indexes to work correctly.
Envelope processing code was also rewritten to get rid of old legacy code
that didn't make sense (but worked for some unknown reason).

Envelope-related OpenMPT test cases that now work correctly include
EnvLoops.it, EnvLoopEscape.it, EnvOffLength.it and s77.it. EnvLoops.it is
pictured below (top: libxmp, bottom: Impulse Tracker).

[image: Inline image 3]

The Changelog (rather short this time):

4.3.8 (20150404):
        Fix bugs caught in the OpenMPT test cases:
        - fix pre-increment of envelope indexes
        - fix IT note release at end of envelope sustain loop
        - reset channel flags in case of delay effect
        Other changes:
        - fix MMD3 16-bit samples (reported by jbb666)
        - refactor XM envelopes
        - refactor IT envelopes

And the packages:

https://sourceforge.net/projects/xmp/files/libxmp/4.3.8/

Please report if you find any regressions!

Best regards,
Claudio

[xmp] Libxmp 4.3.7 released

[Claudio M. <cma...@gm...>]

Hi,

Libxmp 4.3.7 was released, focusing in Impulse Tracker fixes and
improvements. Thanks to the OpenMPT test cases, we're starting to have
decent support for IT modules (not nearly as good as OpenMPT itself, but I
believe we are, at this point, comparable to or even better than Schism
Tracker). Along with the fixes for corner cases explored in the OpenMPT
test cases, a few missing features (high offset, surround channel support
and sample pan support) were requested by honguito98 and implemented in
this version.

Here's the changelog:

4.3.7 (20150329):
        Fix bugs caught in the OpenMPT test cases:
        - fix IT sample mode note cut on invalid sample
        - fix IT sample mode note end detection
        - fix IT envelope handling with carry and fadeout
        - fix IT tone portamento with sample changes
        - fix IT initial global volume setting
        - fix IT keyoff with instrument in old effects mode
        - fix IT filter maximum values with resonance
        Other changes:
        - fix IT random volume variation
        - fix pattern initialization sanity check
        - fix ++ pattern handling in IT loader (reported by honguito98)
        - fix Soundtracker short rip loading (reported by Shlomi Fish)
        - add IT high offset command (SAx)
        - add IT surround command (S9x)
        - add IT surround channel support
        - add IT sample pan setting support

Packages:

https://sourceforge.net/projects/xmp/files/libxmp/4.3.7/

Enjoy!

[xmp] Libxmp 4.3.6 released

[Claudio M. <cma...@gm...>]

Hi,

Libxmp 4.3.6 is available with support to random volume and pan variation
in IT, Amiga split channel emulation in Oktalyzer, and many fixes to
prevent crashes caused by invalid input files. Here's the full changelog:

        Fix bugs caught in the OpenMPT test cases:
        - fix IT volume column volume slide effect memory
        - fix IT default filter cutoff on new note
        - fix IT filter envelope memory
        Fix crashes with fuzzed files (reported by Lionel Debroux):
        - add sanity check to MED2/3/4 loader
        - add sanity check to STIM/GDM/DBM/LIQ/ICE/PSM/PTM/MGT loader
        - add sanity check to MDL/RAD/MGT/IMF/RTM/DT/LIQ/DTM pattern loader
        - add sanity check to OKT/IMF/MMD/MDL sample loader
        - add sanity check to Archimedes Tracker format test
        - add sanity check to Digital Symphony track loader
        - add sanity checks to SQSH, bzip2, arc, lha, lzx and S404 depackers
        - add sanity check for AMD/STX number of patterns
        - add sanity check for DSYM/MMD1/MMD3 number of channels
        - add sanity check for MMD1/MMD3 instrument type
        - add sanity check for IT old instrument loading
        - add sanity checks and fix memory leaks in the Vorbis decoder
        Other changes:
        - fix instrument number in channel initialization
        - fix sample size limit (reported by Jochen Goernitz)
        - fix loading of OpenMPT 1.17 IT modules (reported by Dane Bush)
        - fix sample number limit (reported by Lionel Debroux)
        - fix Oktalyzer split channel replay (reported by Dennis Lindroos)
        - fix Oktalyzer sample loop (by Dennis Lindroos)
        - fix Oktalyzer note slide up/down effect
        - fix ThePlayer pattern decoding
        - fix XM loading for MED2XM modules (reported by Lorence Lombardo)
        - add support to Amiga split channel loop and volume setting
        - add IT random volume variation
        - add IT random pan variation

Source packages available at:

https://sourceforge.net/projects/xmp/files/libxmp/4.3.6/


Best regards,
Claudio

Re: [xmp] Libxmp 4.3.5 released

[Claudio M. <cma...@gm...>]

Thanks, will check and send a patch ASAP.

On Tue, Feb 10, 2015 at 12:10 PM, dev...@ol... <
dev...@ol...> wrote:

>  Thanks for this update, the more stable the better!
>
> However, it seems that one of the tracker files (XM) I am using for tests
> won't play anymore now with the new sanity checks. It's some XM to mp3
> conversion, I'll attach it for you.
>
> ~Smirftsch
>
>
> On 08.02.2015 00:37, Claudio Matsuoka wrote:
>
> Hi,
>
>  Libxmp 4.3.5 is now available with an important fix by Emmanuel Julien
> in sample loop rendering (which should result in a much cleaner sound for
> short looped samples), and significant improvements in robustness with
> corrupted input files as a result of fuzz tests conducted by Lionel
> Debroux. Additionally, the minimum BPM for MED files was fixed. Here's the
> full changelog for this version:
>
>  4.3.5 (20150207):
>  Fix crashes with fuzzed files (reported by Lionel Debroux):
>  - add sanity check for ST3 S3M maximum sample size
>  - add sanity check for sample loop start
>  - add sanity check for speed 0
>  - add sanity check for invalid XM effects
>  - add sanity check for maximum number of channels
>  - add sanity check for number of points in IT envelope
>  - add sanity check for S3M file format information
>  - add sanity check for maximum sample size
>  - add sanity check for invalid envelope points
>  - add sanity check for basic module parameters
>  - add sanity check for instrument release after load error
>  - add sanity check for XM header size
>  - add sanity check for XM/IT/S3M/MTM/RTM parameters and sample size
>  - add sanity checks to inflate and lha decompressors
>  - add more tests to 669 and NO file detection
>  - fix mixer index overflow with large samples
>  - fix prowizard data request response
>  - fix EU/NP1/NP2/NP3 prowizard depackers
>  - fix crash on attempt to play invalid sample
>  - fix infinite loop in break+delay quirk
>  - reset module data before loading module
>  Other changes:
>  - fix loop processing error in scan (reported by Lionel Debroux)
>  - fix minimum BPM value for MED (reported by cspiegel)
>  - fix sample loop adjustment (by Emmanuel Julien)
>
>  Packages can be downloaded from:
>
>  https://sourceforge.net/projects/xmp/files/libxmp/4.3.5/
>
>  Best regards,
> Claudio
>
>
> ------------------------------------------------------------------------------
> Dive into the World of Parallel Programming. The Go Parallel Website,
> sponsored by Intel and developed in partnership with Slashdot Media, is your
> hub for all things parallel software development, from weekly thought
> leadership blogs to news, videos, case studies, tutorials and more. Take a
> look and join the conversation now. http://goparallel.sourceforge.net/
>
>
>
> ---https://lists.sourceforge.net/lists/listinfo/xmp-devel
>
>
>

[xmp] Libxmp 4.3.5 released

[Claudio M. <cma...@gm...>]

Hi,

Libxmp 4.3.5 is now available with an important fix by Emmanuel Julien in
sample loop rendering (which should result in a much cleaner sound for
short looped samples), and significant improvements in robustness with
corrupted input files as a result of fuzz tests conducted by Lionel
Debroux. Additionally, the minimum BPM for MED files was fixed. Here's the
full changelog for this version:

4.3.5 (20150207):
Fix crashes with fuzzed files (reported by Lionel Debroux):
- add sanity check for ST3 S3M maximum sample size
- add sanity check for sample loop start
- add sanity check for speed 0
- add sanity check for invalid XM effects
- add sanity check for maximum number of channels
- add sanity check for number of points in IT envelope
- add sanity check for S3M file format information
- add sanity check for maximum sample size
- add sanity check for invalid envelope points
- add sanity check for basic module parameters
- add sanity check for instrument release after load error
- add sanity check for XM header size
- add sanity check for XM/IT/S3M/MTM/RTM parameters and sample size
- add sanity checks to inflate and lha decompressors
- add more tests to 669 and NO file detection
- fix mixer index overflow with large samples
- fix prowizard data request response
- fix EU/NP1/NP2/NP3 prowizard depackers
- fix crash on attempt to play invalid sample
- fix infinite loop in break+delay quirk
- reset module data before loading module
Other changes:
- fix loop processing error in scan (reported by Lionel Debroux)
- fix minimum BPM value for MED (reported by cspiegel)
- fix sample loop adjustment (by Emmanuel Julien)

Packages can be downloaded from:

https://sourceforge.net/projects/xmp/files/libxmp/4.3.5/

Best regards,
Claudio

[xmp] Libxmp 4.3.4 released

[Claudio M. <cma...@gm...>]

Hi,

Libxmp 4.3.4 is available with a few more fixes for XM replay based on the
OpenMPT test cases, fixes in PSM pattern parsing, support for ProPacker
3.0, and an important fix in the linear interpolator. The full changelog is:

4.3.4 (20150111):
        Fix bugs caught in the OpenMPT test cases:
        - fix XM keyoff+delay combinations
        - fix XM fine pitch slide with pattern delay
        - fix XM vibrato rampdown waveform
        - fix XM volume column pan with keyoff and delay
        - fix XM pan envelope position setting
        - fix channel volume and instrument initialization
        - fix end of module detection inside a loop
        Fix bugs reported by Francisco Pareja-Lecaros:
        - fix MASI (PSM) volume command
        - fix MASI (PSM) note number parsing
        - fix Noisetracker note limit detection
        Other changes:
        - fix overflow in linear interpolator (reported by Jochen Goernitz)
        - fix MTM invalid track load (reported by Douglas Carmichael)
        - add ProPacker 3.0 loader

Download the packages from
https://sourceforge.net/projects/xmp/files/libxmp/4.3.4/.

Best regards,
Claudio

Re: [xmp] more various questions: XMP_PLAYER_INTERP

[<dev...@ol...>]

Great news!

What do you think / know about this "Polyphase" interpolation modplug 
offers?
As said I was told it would be best in quality, although I have no idea 
or method to really compare that except my ears, which don't seem to 
provide me that information :)
Also I don't know how or if it can be compared with the interpolation 
methods xmp offers.

On 03.01.2015 12:11, Claudio Matsuoka wrote:
> Yep, it was a bug in the linear interpolator. Fixed.
>
> On Fri, Jan 2, 2015 at 11:31 PM, Claudio Matsuoka <cma...@gm... 
> <mailto:cma...@gm...>> wrote:
>
>     Very interesting. At this development stage I wouldn't expect this
>     kind of error, but it really seems that we have some kind of
>     interpolation error here (see the attached waveform plot). I'll
>     investigate and report back.
>
>     On Fri, Jan 2, 2015 at 5:43 AM, dev...@ol...
>     <mailto:dev...@ol...> <dev...@ol...
>     <mailto:dev...@ol...>> wrote:
>
>         Thanks for explanation! This finally makes sense to me ;)
>
>         Major reason I was asking is, because I have one IT track
>         (Impulse Tracker 2.14v5 IT 1.00) in which a noticeable
>         crackling can be heared when playing, except when using
>         SPLINE. The specific sound happens at around second 13 the
>         first time, its some high pitched effect and the crackling
>         sounds a bit like overmodulation. I attached the corresponding
>         file, maybe its a problem in the file, or some kind of bug, or
>         plain normal side-effect of interpolation,  I really can't
>         say, but maybe you want to verify this yourself.
>         Also while digging for information about this matter I
>         stumbled again over modplug, they seem to have another kind of
>         interpolation called "Polyphase" - I am not sure what this
>         means again, but an audio artist I have been talking with told
>         me it would sound the best.
>         Sorry for my comparison with modplug all the time, I hate to
>         do myself, but it's just what I am being reported and the
>         player most people seem to compare with.
>
>
>
>         On 01.01.2015 21:22, Claudio Matsuoka wrote:
>>         Hi again,
>>
>>         Interpolation is needed because the PCM sounds and our mixed
>>         sound output have different sampling rates. When no sample
>>         exists at that exact point in the input sample, an
>>         interpolated value is used. The following image illustrates
>>         what we're doing:
>>
>>         http://www.mathworks.com/matlabcentral/answers/uploaded_files/20673/File1.JPG
>>
>>         The dots are the PCM sample points. The square-ish solid line
>>         shows the nearest neighbor interpolation, and the dashed line
>>         shows the linear interpolation waveform. The cubic spline
>>         interpolation goes a bit further and connects the dots using
>>         curves. It sounds better for some modules, but it uses much
>>         more CPU to compute than linear interpolation. Linear
>>         interpolation is usually good enough for modules, and it's
>>         what we use in the Android player. The PC CLI player uses
>>         cubic splines by default.
>>
>>         On Thu, Jan 1, 2015 at 7:52 AM, dev...@ol...
>>         <mailto:dev...@ol...> <dev...@ol...
>>         <mailto:dev...@ol...>> wrote:
>>
>>             sorry, the different interpolation types
>>             XMP_INTERP_NEAREST /* Nearest neighbor */
>>             XMP_INTERP_LINEAR /* Linear (default) */
>>             XMP_INTERP_SPLINE /* Cubic spline */
>>
>>             are also very short in explanation, to short for me at
>>             least...
>>             hope this is not tooooo annoying for you :)
>>
>>             ~Jochen
>>
>>             ------------------------------------------------------------------------------
>>             Dive into the World of Parallel Programming! The Go
>>             Parallel Website,
>>             sponsored by Intel and developed in partnership with
>>             Slashdot Media, is your
>>             hub for all things parallel software development, from
>>             weekly thought
>>             leadership blogs to news, videos, case studies, tutorials
>>             and more. Take a
>>             look and join the conversation now.
>>             http://goparallel.sourceforge.net
>>             ---
>>             https://lists.sourceforge.net/lists/listinfo/xmp-devel
>>
>>
>
>
>

Re: [xmp] more various questions: XMP_PLAYER_INTERP

[Claudio M. <cma...@gm...>]

Yep, it was a bug in the linear interpolator. Fixed.

On Fri, Jan 2, 2015 at 11:31 PM, Claudio Matsuoka <cma...@gm...>
wrote:

> Very interesting. At this development stage I wouldn't expect this kind of
> error, but it really seems that we have some kind of interpolation error
> here (see the attached waveform plot). I'll investigate and report back.
>
> On Fri, Jan 2, 2015 at 5:43 AM, dev...@ol... <
> dev...@ol...> wrote:
>
>>  Thanks for explanation! This finally makes sense to me ;)
>>
>> Major reason I was asking is, because I have one IT track (Impulse
>> Tracker 2.14v5 IT 1.00) in which a noticeable crackling can be heared when
>> playing, except when using SPLINE. The specific sound happens at around
>> second 13 the first time, its some high pitched effect and the crackling
>> sounds a bit like overmodulation. I attached the corresponding file, maybe
>> its a problem in the file, or some kind of bug, or plain normal side-effect
>> of interpolation,  I really can't say, but maybe you want to verify this
>> yourself.
>> Also while digging for information about this matter I stumbled again
>> over modplug, they seem to have another kind of interpolation called
>> "Polyphase" - I am not sure what this means again, but an audio artist I
>> have been talking with told me it would sound the best.
>> Sorry for my comparison with modplug all the time, I hate to do myself,
>> but it's just what I am being reported and the player most people seem to
>> compare with.
>>
>>
>>
>> On 01.01.2015 21:22, Claudio Matsuoka wrote:
>>
>> Hi again,
>>
>>  Interpolation is needed because the PCM sounds and our mixed sound
>> output have different sampling rates. When no sample exists at that exact
>> point in the input sample, an interpolated value is used. The following
>> image illustrates what we're doing:
>>
>>
>> http://www.mathworks.com/matlabcentral/answers/uploaded_files/20673/File1.JPG
>>
>>  The dots are the PCM sample points. The square-ish solid line shows the
>> nearest neighbor interpolation, and the dashed line shows the linear
>> interpolation waveform. The cubic spline interpolation goes a bit further
>> and connects the dots using curves. It sounds better for some modules, but
>> it uses much more CPU to compute than linear interpolation. Linear
>> interpolation is usually good enough for modules, and it's what we use in
>> the Android player. The PC CLI player uses cubic splines by default.
>>
>> On Thu, Jan 1, 2015 at 7:52 AM, dev...@ol... <
>> dev...@ol...> wrote:
>>
>>> sorry, the different interpolation types
>>> XMP_INTERP_NEAREST /* Nearest neighbor */
>>> XMP_INTERP_LINEAR /* Linear (default) */
>>> XMP_INTERP_SPLINE /* Cubic spline */
>>>
>>> are also very short in explanation, to short for me at least...
>>> hope this is not tooooo annoying for you :)
>>>
>>> ~Jochen
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Dive into the World of Parallel Programming! The Go Parallel Website,
>>> sponsored by Intel and developed in partnership with Slashdot Media, is
>>> your
>>> hub for all things parallel software development, from weekly thought
>>> leadership blogs to news, videos, case studies, tutorials and more. Take
>>> a
>>> look and join the conversation now. http://goparallel.sourceforge.net
>>> ---
>>> https://lists.sourceforge.net/lists/listinfo/xmp-devel
>>>
>>
>>
>>
>

Re: [xmp] more various questions: XMP_PLAYER_INTERP

[Claudio M. <cma...@gm...>]

Very interesting. At this development stage I wouldn't expect this kind of
error, but it really seems that we have some kind of interpolation error
here (see the attached waveform plot). I'll investigate and report back.

On Fri, Jan 2, 2015 at 5:43 AM, dev...@ol... <
dev...@ol...> wrote:

>  Thanks for explanation! This finally makes sense to me ;)
>
> Major reason I was asking is, because I have one IT track (Impulse Tracker
> 2.14v5 IT 1.00) in which a noticeable crackling can be heared when playing,
> except when using SPLINE. The specific sound happens at around second 13
> the first time, its some high pitched effect and the crackling sounds a bit
> like overmodulation. I attached the corresponding file, maybe its a problem
> in the file, or some kind of bug, or plain normal side-effect of
> interpolation,  I really can't say, but maybe you want to verify this
> yourself.
> Also while digging for information about this matter I stumbled again over
> modplug, they seem to have another kind of interpolation called "Polyphase"
> - I am not sure what this means again, but an audio artist I have been
> talking with told me it would sound the best.
> Sorry for my comparison with modplug all the time, I hate to do myself,
> but it's just what I am being reported and the player most people seem to
> compare with.
>
>
>
> On 01.01.2015 21:22, Claudio Matsuoka wrote:
>
> Hi again,
>
>  Interpolation is needed because the PCM sounds and our mixed sound
> output have different sampling rates. When no sample exists at that exact
> point in the input sample, an interpolated value is used. The following
> image illustrates what we're doing:
>
>
> http://www.mathworks.com/matlabcentral/answers/uploaded_files/20673/File1.JPG
>
>  The dots are the PCM sample points. The square-ish solid line shows the
> nearest neighbor interpolation, and the dashed line shows the linear
> interpolation waveform. The cubic spline interpolation goes a bit further
> and connects the dots using curves. It sounds better for some modules, but
> it uses much more CPU to compute than linear interpolation. Linear
> interpolation is usually good enough for modules, and it's what we use in
> the Android player. The PC CLI player uses cubic splines by default.
>
> On Thu, Jan 1, 2015 at 7:52 AM, dev...@ol... <
> dev...@ol...> wrote:
>
>> sorry, the different interpolation types
>> XMP_INTERP_NEAREST /* Nearest neighbor */
>> XMP_INTERP_LINEAR /* Linear (default) */
>> XMP_INTERP_SPLINE /* Cubic spline */
>>
>> are also very short in explanation, to short for me at least...
>> hope this is not tooooo annoying for you :)
>>
>> ~Jochen
>>
>>
>> ------------------------------------------------------------------------------
>> Dive into the World of Parallel Programming! The Go Parallel Website,
>> sponsored by Intel and developed in partnership with Slashdot Media, is
>> your
>> hub for all things parallel software development, from weekly thought
>> leadership blogs to news, videos, case studies, tutorials and more. Take a
>> look and join the conversation now. http://goparallel.sourceforge.net
>> ---
>> https://lists.sourceforge.net/lists/listinfo/xmp-devel
>>
>
>
>

[xmp] source code for Impulse Tracker released

[Paul W. <pa...@bo...>]

Hi,

You probably heard about this but just in case:

The source code (assembler) of Impulse Tracker was released under a BSD license:

https://roartindon.blogspot.com/2014/12/20-years-of-impulse-tracker-part-4.html
https://bitbucket.org/jthlim/impulsetracker

-- 
bye,
pabs

http://bonedaddy.net/pabs3/

Re: [xmp] Heads-up: Audacious 3.6 API changes

[Claudio M. <cma...@gm...>]

Hi Michael,

Interesting news indeed. Sorry for not replying earlier, I was in a place
with a rather poor internet connection and only now I'm reading and
replying the emails received in the last three weeks or so.

I think the Audacious guys could be doing the right thing rewriting their
API. Sometimes we really have to drop all the legacy cruft and start from
scratch, and the XMMS-BMP-Audacious lineage was too old and patchy. It's
also an interesting test to see if libxmp works well linked to C++ code.

Regarding the old copyright notices, they're from the monolithic player so
they can be dropped (and since you're the plugin author, I think the only
copyright line that's important is the one with your name, no need to
recursively include library credits). And the 8-bit stuff -- is it still
relevant today? If not, I think it can be dropped as well.

Also since this is a major rewrite, it could be a good time to just set up
a new code repository and we can just flag the previous one as legacy.

Best regards,
Claudio


On Sat, Dec 13, 2014 at 7:44 AM, Michael Schwendt <msc...@gm...>
wrote:

> Hi everyone!
>
> Interesting times ahead.
>
> Audacious 3.6-alpha1 has been released and introduces usage of Qt as a
> future replacement of GTK+ as well as C++ as the primary programming
> language.
> This also affects the Plugin API. Plugin definitions are done in C++ now.
> The Preferences GUI API also has seen modifications, which required an
> update
> to the plugin.
>
> I've had a first go at porting it,
>
>
> https://github.com/mschwendt/xmp-plugin-audacious/blob/master/src/audxmp.cpp
>
> which has also shortened the code because I've dropped all old cruft I
> could
> find (including the separate guicfg and plugin_cfg struct sync stuff that
> is
> unnecessary nowadays). Clearly one can set the cfg values directly from
> the GUI
> and have play() evaluate them as appropriate for configuring libxmp.
>
> Questions and TODOs:
>
>  * The Copyright details for the About dialog. The top half refers to XMP
> unless I'm mistaken:
>
> https://github.com/mschwendt/xmp-plugin-audacious/blob/master/src/audxmp.cpp#L83
>
>  * The source file preamble still mentions the ancient XMMS/Beep/Audacious
> plugin credits, although hardly anything from that era is left in the
> current code. Especially the old play/stop/pause/seek functions are gone
> or have changed completely due to Audacious development. No strong feelings
> here, just that it is misleading to pretend such a relationship to an
> origin that has hardly anything in common with the current code anymore.
>
>  * The "force8bit" and "convert8bit" configuration settings. It doesn't
> work. In the older plugin versions, force8bit is not set by the GUI,
> and convert8bit and bits8 seem redundant. Activating 8bit, XMP here
> creates noise. How to fix that?
>
>
> ------------------------------------------------------------------------------
> Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
> from Actuate! Instantly Supercharge Your Business Reports and Dashboards
> with Interactivity, Sharing, Native Excel Exports, App Integration & more
> Get technology previously reserved for billion-dollar corporations, FREE
>
> http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk
> ---
> https://lists.sourceforge.net/lists/listinfo/xmp-devel
>

Re: [xmp] more various questions: XMP_PLAYER_INTERP

[Claudio M. <cma...@gm...>]

Hi again,

Interpolation is needed because the PCM sounds and our mixed sound output
have different sampling rates. When no sample exists at that exact point in
the input sample, an interpolated value is used. The following image
illustrates what we're doing:

http://www.mathworks.com/matlabcentral/answers/uploaded_files/20673/File1.JPG

The dots are the PCM sample points. The square-ish solid line shows the
nearest neighbor interpolation, and the dashed line shows the linear
interpolation waveform. The cubic spline interpolation goes a bit further
and connects the dots using curves. It sounds better for some modules, but
it uses much more CPU to compute than linear interpolation. Linear
interpolation is usually good enough for modules, and it's what we use in
the Android player. The PC CLI player uses cubic splines by default.

On Thu, Jan 1, 2015 at 7:52 AM, dev...@ol... <
dev...@ol...> wrote:

> sorry, the different interpolation types
> XMP_INTERP_NEAREST /* Nearest neighbor */
> XMP_INTERP_LINEAR /* Linear (default) */
> XMP_INTERP_SPLINE /* Cubic spline */
>
> are also very short in explanation, to short for me at least...
> hope this is not tooooo annoying for you :)
>
> ~Jochen
>
>
> ------------------------------------------------------------------------------
> Dive into the World of Parallel Programming! The Go Parallel Website,
> sponsored by Intel and developed in partnership with Slashdot Media, is
> your
> hub for all things parallel software development, from weekly thought
> leadership blogs to news, videos, case studies, tutorials and more. Take a
> look and join the conversation now. http://goparallel.sourceforge.net
> ---
> https://lists.sourceforge.net/lists/listinfo/xmp-devel
>

Re: [xmp] various questions PAN and MIX, DSP

[Claudio M. <cma...@gm...>]

Hi,

The difference is that XMP_PLAYER_MIX is applied in the PCM buffer
generation when the left/right channels are rendered, so it affects all
modules regardless if they have dynamic panning such as XM or IT, or
pre-assigned "default pan" values such as Protracker modules and friends
(LRRL). In those default pan modules, however, 100% LRRL is too aggressive
for headphones. If you set 50% XMP_PLAYER_MIX they'll sound much better,
but you'll render modules with dynamic pan incorrectly. So
XMP_PLAYER_DEFPAN was created to keep dynamic pan at a higher level and
allow control only for LRRL modules. The xmp command-line player sets
XMP_PLAYER_DEFPAN to 50% by default.

Best regards,
Claudio

On Thu, Jan 1, 2015 at 7:45 AM, dev...@ol... <
dev...@ol...> wrote:

> What is the actual difference between XMP_PLAYER_DEFPAN and XMP_PLAYER_MIX?
>
> The api docs are unclear about that, maybe it should be obvious, but I
> am not so deep in this matter, so please forgive me if these are stupid
> questions. :)
> I have to admit that a couple of functions are not entirely clear to me,
> f.e. what the different settings for XMP_PLAYER_DSP exactly do.
>
> On an entirely different topic, I was told that at least a specific IT
> track (Impulse Tracker 2.14v3 S3M) plays songs in reversed stereo
> (left/right) compared to modplug tracker. I can send the track if needed.
>
> Thanks for all the hard work you put in this!
>
> PS: Happy new year!
>
>
>
>
> ------------------------------------------------------------------------------
> Dive into the World of Parallel Programming! The Go Parallel Website,
> sponsored by Intel and developed in partnership with Slashdot Media, is
> your
> hub for all things parallel software development, from weekly thought
> leadership blogs to news, videos, case studies, tutorials and more. Take a
> look and join the conversation now. http://goparallel.sourceforge.net
> ---
> https://lists.sourceforge.net/lists/listinfo/xmp-devel
>

[xmp] more various questions: XMP_PLAYER_INTERP

[<dev...@ol...>]

sorry, the different interpolation types
XMP_INTERP_NEAREST /* Nearest neighbor */
XMP_INTERP_LINEAR /* Linear (default) */
XMP_INTERP_SPLINE /* Cubic spline */

are also very short in explanation, to short for me at least...
hope this is not tooooo annoying for you :)

~Jochen

[xmp] various questions PAN and MIX, DSP

[<dev...@ol...>]

What is the actual difference between XMP_PLAYER_DEFPAN and XMP_PLAYER_MIX?

The api docs are unclear about that, maybe it should be obvious, but I 
am not so deep in this matter, so please forgive me if these are stupid 
questions. :)
I have to admit that a couple of functions are not entirely clear to me, 
f.e. what the different settings for XMP_PLAYER_DSP exactly do.

On an entirely different topic, I was told that at least a specific IT 
track (Impulse Tracker 2.14v3 S3M) plays songs in reversed stereo 
(left/right) compared to modplug tracker. I can send the track if needed.

Thanks for all the hard work you put in this!

PS: Happy new year!

[xmp] Libxmp 4.3.3 released

[Claudio M. <cma...@gm...>]

Hi,

Libxmp 4.3.3 is now available with fixes for XM envelope loops (which
should address the regression in Elwood's "Shooting Star"), XM note delay
effect handling and fixes for big-endian detection. (Not many fixes in this
release -- I spent a few weeks without my laptop in a place with poor
internet connection, so I couldn't really work on the player as much as I
should have).

If you have contacted me and I didn't reply, I should finish checking my
email backlog in a few days :)

Here's the changelog for libxmp-4.3.3:

4.3.3 (20141231):
        Fix bugs caught in the OpenMPT test cases:
        - fix XM note delay volume with no note or instrument set
        - fix XM out-of-range note delays with pattern delays
        Other changes:
        - fix XM envelope loop length (reported by Per Törner)
        - fix big-endian detection in configuration (by Andreas Schwab)

Source packages for the regular and "lite" player libraries are available
at:

https://sourceforge.net/projects/xmp/files/libxmp/4.3.3/

Enjoy and please report any bugs you may find.

Re: [xmp] Heads-up: Audacious 3.6 API changes

[Michael S. <msc...@gm...>]

>  * The "force8bit" and "convert8bit" configuration settings. It doesn't
> work. In the older plugin versions, force8bit is not set by the GUI,
> and convert8bit and bits8 seem redundant. Activating 8bit, XMP here
> creates noise. How to fix that?

Oh, it seems the plugin simply has never before told XMP to use 8-bit
samples.

-    if (plugin_cfg.convert8bit == 0)
+    if (plugin_cfg.convert8bit == 0) {
         resol = 16;
+    } else {
+        flags |= XMP_FORMAT_8BIT;
+        flags |= XMP_FORMAT_UNSIGNED;
+    }
 

[xmp] Heads-up: Audacious 3.6 API changes

[Michael S. <msc...@gm...>]

Hi everyone!

Interesting times ahead.

Audacious 3.6-alpha1 has been released and introduces usage of Qt as a
future replacement of GTK+ as well as C++ as the primary programming language.
This also affects the Plugin API. Plugin definitions are done in C++ now.
The Preferences GUI API also has seen modifications, which required an update
to the plugin.

I've had a first go at porting it,

  https://github.com/mschwendt/xmp-plugin-audacious/blob/master/src/audxmp.cpp

which has also shortened the code because I've dropped all old cruft I could
find (including the separate guicfg and plugin_cfg struct sync stuff that is
unnecessary nowadays). Clearly one can set the cfg values directly from the GUI
and have play() evaluate them as appropriate for configuring libxmp.

Questions and TODOs:

 * The Copyright details for the About dialog. The top half refers to XMP
unless I'm mistaken:
https://github.com/mschwendt/xmp-plugin-audacious/blob/master/src/audxmp.cpp#L83

 * The source file preamble still mentions the ancient XMMS/Beep/Audacious
plugin credits, although hardly anything from that era is left in the
current code. Especially the old play/stop/pause/seek functions are gone
or have changed completely due to Audacious development. No strong feelings
here, just that it is misleading to pretend such a relationship to an
origin that has hardly anything in common with the current code anymore.

 * The "force8bit" and "convert8bit" configuration settings. It doesn't
work. In the older plugin versions, force8bit is not set by the GUI,
and convert8bit and bits8 seem redundant. Activating 8bit, XMP here
creates noise. How to fix that?

[xmp] Libxmp 4.3.2 and xmp 4.0.10 released

[Claudio M. <cma...@gm...>]

Hi,

After fixing a few more problems, new versions of libxmp and xmp are now
available. In the library side, the changelog is:

4.3.2 (20141130):
        Fix bugs caught in the OpenMPT test cases:
        - fix IT invalid instrument number recovery
        - fix IT note retrig on portamento with same sample
        - fix XM portamento target reset on new instrument
        - fix XM portamento with offset
        - fix XM pan slide memory
        - fix XM tremolo and vibrato waveforms
        - fix MOD pattern break with pattern delay
        - fix MOD Protracker offset bug emulation
        - fix tremolo rate
        Other changes:
        - fix IT portamento after keyoff and note end
        - fix IT fadeout reset on new note
        - fix IT pattern row delay scan
        - fix MOD/XM volume up+down priority (reported by Jason Gibson)
        - fix MOD fine volume slide memory (reported by Dennis Lindroos)
        - fix set sample offset effect (by Dennis Lindroos)
        - fix Windows temp file (reported by Andreas Argirakis & Eric
Lévesque)
        - add emulation of the FT2 pattern loop bug (by Eugene Toder)
        - allow loading of packed formats from memory
        - allow loading of OpenMPT MOD files with large samples
        - enable offset bug emulation by default for Protracker MODs
        - code cleanup

A few improvements were also made in the command-line player, including
fixing sound driver deinitialization issues, adjusting sound latency in OS
X, and adding a new AIFF file output driver. Here's the changelog:

4.0.10 (20141030):
        - Add AIFF file output driver (by Lorence Lombardo)
        - Add command 'h' to display help message (by Eugene Toder)
        - Fix sound driver deinitialization on signal (by Eugene Toder)
        - Adjust CoreAudio driver latency
        - Fix missing --all-sequences in help message

As usual, get the source tarballs from:

https://sourceforge.net/projects/xmp/files/libxmp/4.3.2/
https://sourceforge.net/projects/xmp/files/xmp/4.0.10/

Best regards,
Claudio

[xmp] Libxmp 4.3.1 released

[Claudio M. <cma...@gm...>]

Hi.

Libxmp 4.3.1 (and the corresponding "lite" version) is available with many
interesting fixes including one for a very old bug related to sample
retrigs that probably was lurking around since version 2.0. Other changes
include more fixes for OpenMPT test cases, and for module titles and
instrument names in certain BSDs and Mac OS X. The complete changelog is:

4.3.1 (20141111):
        Fix bugs caught in the OpenMPT test cases:
        - fix IT filter envelope range
        - fix IT envelope carry after envelope end
        - fix XM note off with volume command
        - fix XM K00 effect handling
        - fix XM portamento with volume column portamento
        - fix XM keyoff with instrument
        - fix XM note limits
        Fix bugs reported by Andreas Argirakis:
        - fix MOD false positive for UNIC Tracker modules
        - fix EMOD instrument finetune
        - fix UNIC Tracker instrument finetune test
        - fix NoisePacker1 loader
        Other changes:
        - fix IT tone portamento in first note (reported by Jan Engelhardt)
        - fix XM invalid memory access in event reader
        - fix STM empty note event read
        - fix ABK loader test in Win32
        - fix MOD period range enforcing (reported by Jason Gibson)
        - fix ST2.6 speed effect (reported by Saga Musix)
        - fix corner case memory leak in S3M loader
        - fix retrig of single-shot samples after the end of the sample
        - fix crash in envelope reset with invalid instrument
        - fix module titles and instrument names in Mac OS X
        - fix row delay initialization on new module
        - refactor depacking code
        - code cleanup

As usual, the release packages are available at
https://sourceforge.net/projects/xmp/files/libxmp/4.3.1/.

Enjoy!

Re: [xmp] unic tracker problems

[Claudio M. <cma...@gm...>]

Thanks, will check and report back.

On Sat, Nov 8, 2014 at 7:40 PM, Andreas Argirakis <an...@bl...> wrote:

> This is supposedly a unic tracker 1.4 (sometimes called v2), the module
> sad-song from the demo http://www.exotica.org.uk/wiki/Guardian_Dragon. It
> is not recognized by libxmp.
>
> The files at http://www.exotica.org.uk/wiki/Guardian_Dragon are
> identified as protracker and noisetracker and will thus not play correctly.
>
> Using libxmp 4.3.0.
>
>
> /Andreas
>
>
> ------------------------------------------------------------------------------
>
> ---
> https://lists.sourceforge.net/lists/listinfo/xmp-devel
>
>

[xmp] unic tracker problems

[Andreas A. <an...@bl...>]

This is supposedly a unic tracker 1.4 (sometimes called v2), the module 
sad-song from the demo http://www.exotica.org.uk/wiki/Guardian_Dragon. 
It is not recognized by libxmp.

The files at http://www.exotica.org.uk/wiki/Guardian_Dragon are 
identified as protracker and noisetracker and will thus not play correctly.

Using libxmp 4.3.0.


/Andreas

Re: [xmp] noisepacker 1 doesn't work

[Claudio M. <cma...@gm...>]

Fixed. It was reading an incorrect number of patterns.

On Wed, Nov 5, 2014 at 8:42 PM, Claudio Matsuoka <cma...@gm...>
wrote:

> Will check, thanks.
>
> On Wed, Nov 5, 2014 at 8:13 PM, Andreas Argirakis <an...@bl...>
> wrote:
>
>> Hi,
>>
>> I'm not sure how good the supprt for Noisepacker v1 is overall, but the
>> attached one doesn't work very well at all. I think this is the first v1
>> one I've ever tested.
>>
>> Using libxmp 4.3.0
>>
>>
>> ------------------------------------------------------------------------------
>>
>> ---
>> https://lists.sourceforge.net/lists/listinfo/xmp-devel
>>
>>
>

Re: [xmp] noisepacker 1 doesn't work

[Claudio M. <cma...@gm...>]

Will check, thanks.

On Wed, Nov 5, 2014 at 8:13 PM, Andreas Argirakis <an...@bl...> wrote:

> Hi,
>
> I'm not sure how good the supprt for Noisepacker v1 is overall, but the
> attached one doesn't work very well at all. I think this is the first v1
> one I've ever tested.
>
> Using libxmp 4.3.0
>
>
> ------------------------------------------------------------------------------
>
> ---
> https://lists.sourceforge.net/lists/listinfo/xmp-devel
>
>