[Csnd-dev] Release 6.12 is imminent

classic Classic list List threaded Threaded
10 messages Options
Reply | Threaded
Open this post in threaded view
|

[Csnd-dev] Release 6.12 is imminent

jpff
All being well we intend releasing v6.12 next week, probably Monday.

Below are the draft release notes.  Any comments or corrections welcome

==John ffitch

DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT



# CSOUND VERSION 6.12 RELEASE NOTES

Many changes including the removal of vst2cs functionality due to a copyright issue.

The changes made in 6.11 to raw format reading have been modified so
gen1 and diskin ignore positive file formats and use the file header,
unless the format is negative when it uses the absolute value in a raw
audio file.  This should preserve most compatibility issues.

There are a number of new and improved opcodes, new facilities in
scores and many bug fixes.


-- The Developers

## USER-LEVEL CHANGES

### New opcodes

- fluidinfo retrieves program information from a currently loaded soundfont.

- New opcode ftaudio writes a ftable to an audio file; irate and triggered
    k-rate version exist and k-rate version supports sync or async writing.

- Version of OSClisten that writes the data to a k-rate array now exists.

- OSCcount returns the number of OSC incoming messages pending.

- faustplay and faustdsp are new opcodes splitting the faust DSP
  instantiation and performance.

- OSCbundle sends a collection of similar OSC messages as a single packet
  for efficiency.

- beosc and beadsyn are band enhanced oscillator and oscillator bank.
  bpfcos is for breakpoint files with interpolation.  Similarly lincos
  adds cosine interpolation to linlin.

- printarray and reshapearray act on arrays.

- trim and trim_i adjust the size of a 1 dimensional array, either
  bigger or smaller, preserving data and/or padding with zeros.

### New Gen and Macros

-

### Orchestra

- New preprocessor option #includestr.  This is like #include but has
  macro expansion in the double-quote delimited string.

- Use of tied notes in subinstr fixed.

- Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.

- runtime error message now (usually) include a line number and a file/macro trace.

- multiple assignments such as ka,kb=1,2 are again supported by the
  parser; it had inadvertently got lost.

- the problematic use of i() with an array element directly is now
  flagged as an error.

- If an included file name has a "/" in it then any nested include is
  relative to that directory. (Issue #973)

### Score

- New preprocessor option #includestr.  This is like #include but has
  macro expansion in the double-quote delimited string

- 'd' score opcode for real-time performance.  This was issue #966.

- bug in np operation fixed.

- Use of [] syntax in a score could lead to a loss of precision for
  numbers over about 1 million; older version restored.

- Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.

- The forms for delayed ending of sections (e 5 or s 5) now work with
  fractional delays; previously only read the integer part.

- If an included file name has a "/" in it then any nested include is
  relative to that directory. (Issue #973)

### Options

-

### Modified Opcodes and Gens

- Add optional argument to ftom for rounding answer to integer.

- flooper2 and syncgrain etc  now allow resampling.

- chnclear can now take a list of channels to clear instead of just one.

- printf and printf_i now are like the manual: all arguments beyond the
  format and trigger are optional.

- prints and printks can take string arguments printed with %s.

- GEN2 can now take a size of zero, which is interpreted as size
  sufficient for the number of values provided.

- faustcompile now includes a new optional parameter to allow it to be
run in a blocking mode. Defaults (as before) to non-blocking.

- fillarray can be run at k-rate if any of the argument/values are k-rate.

- slicearray for other than i-rate arrays runs at k-rate only.

### Utilities

- A coding error in mixer was fixed.  It was very broken.

### Frontends

- Belacsound:

- CsoundQt:

### General Usage

## Bugs Fixed

- A typo in p5glove meant that the command to read the button status
  as a bitmap only returned state of button A.

- diskin to array fixed and also use with small ksmps.

- in loscil it sometimes failed to deal with the ibas argument; this
  has now been reworked to be correct.

- madsr could overflow an internal counter when given a negative p3.

- Fixed mapping from threads to lua_States (issue #959).

- the time calculation in flooper2, flooper, and syncgrain was corrected.

- resampling and pitch fixed in pvstanal.

- Rare buffer overflow case in faust opcodes fixed.

- pvsftw had an incorrect check for fft format which led to incorrect
  claim of bad format; fixed.

- if ksmps was 1 the opcode linenr at arate failed to work; fixed.

- window opcode fixed.

- The test for compatible subtypes of f-values in a number of pvs
  opcodes was wrong, causing spurious error messages.

- cosseg was broken for more than one segment; now OK.

- monitor opcode did not work correctly when using multiple processes; fixed.

- in the linenr opcode if the release stage was entered before the end
  of the rise the output value jumped to the end value and then
  decayed, causing a glitch.  This is fixed. (#1048)

# SYSTEM LEVEL CHANGES

- the various -zN options now reports the number of opcodes for the request, so
differs with respect to deprecated and polymorphic opcodes.

### System Changes

- Recompilation of named instruments totally reworked to avoid errors
  and memory leaks.

- The allocation of instrument names to internal numbers has been
  rewritten and should now be usable with replacements in live coding.

- Printing the number allocated to a named instrument now behaves the
  manual, not just for debug.

- if liblo version 0.29 is available csound can be built to use it (with
  a compiler flag LIBLO29) and this fixes some bugs related to
  heavy/complex use of OSClisten.  Unfortunately the older 0.28 version
  is being distributed by some Linux distros.

- The orchestra compiler has a number of new optimisations, avoiding
unnecessary assignments and doing some more expression optimisations.

### Translations

### API

- csound->ReadScore was changed so it behaves the same as a score in a
  csd or sco file.  This could incorrectly give an infinite score or not
  in unexpected cases.

### Platform Specific

- WebAudio:

- iOS

- Android

- Windows

- MacOS

- GNU/Linux

- Bela
 - allow analog in and out with different channel numbers.

==END==
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

joachim-3
looks great.

under "Modified Opcodes and Gens" we can add:

- pvstrace now has the option to also return an array containing the bin
numbers

(victor can sy better ...)

        joachim



On 18/10/18 15:18, John wrote:

> All being well we intend releasing v6.12 next week, probably Monday.
>
> Below are the draft release notes.  Any comments or corrections welcome
>
> ==John ffitch
>
> DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT
>
>
>
> # CSOUND VERSION 6.12 RELEASE NOTES
>
> Many changes including the removal of vst2cs functionality due to a copyright issue.
>
> The changes made in 6.11 to raw format reading have been modified so
> gen1 and diskin ignore positive file formats and use the file header,
> unless the format is negative when it uses the absolute value in a raw
> audio file.  This should preserve most compatibility issues.
>
> There are a number of new and improved opcodes, new facilities in
> scores and many bug fixes.
>
>
> -- The Developers
>
> ## USER-LEVEL CHANGES
>
> ### New opcodes
>
> - fluidinfo retrieves program information from a currently loaded soundfont.
>
> - New opcode ftaudio writes a ftable to an audio file; irate and triggered
>     k-rate version exist and k-rate version supports sync or async writing.
>
> - Version of OSClisten that writes the data to a k-rate array now exists.
>
> - OSCcount returns the number of OSC incoming messages pending.
>
> - faustplay and faustdsp are new opcodes splitting the faust DSP
>   instantiation and performance.
>
> - OSCbundle sends a collection of similar OSC messages as a single packet
>   for efficiency.
>
> - beosc and beadsyn are band enhanced oscillator and oscillator bank.
>   bpfcos is for breakpoint files with interpolation.  Similarly lincos
>   adds cosine interpolation to linlin.
>
> - printarray and reshapearray act on arrays.
>
> - trim and trim_i adjust the size of a 1 dimensional array, either
>   bigger or smaller, preserving data and/or padding with zeros.
>
> ### New Gen and Macros
>
> -
>
> ### Orchestra
>
> - New preprocessor option #includestr.  This is like #include but has
>   macro expansion in the double-quote delimited string.
>
> - Use of tied notes in subinstr fixed.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.
>
> - runtime error message now (usually) include a line number and a file/macro trace.
>
> - multiple assignments such as ka,kb=1,2 are again supported by the
>   parser; it had inadvertently got lost.
>
> - the problematic use of i() with an array element directly is now
>   flagged as an error.
>
> - If an included file name has a "/" in it then any nested include is
>   relative to that directory. (Issue #973)
>
> ### Score
>
> - New preprocessor option #includestr.  This is like #include but has
>   macro expansion in the double-quote delimited string
>
> - 'd' score opcode for real-time performance.  This was issue #966.
>
> - bug in np operation fixed.
>
> - Use of [] syntax in a score could lead to a loss of precision for
>   numbers over about 1 million; older version restored.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.
>
> - The forms for delayed ending of sections (e 5 or s 5) now work with
>   fractional delays; previously only read the integer part.
>
> - If an included file name has a "/" in it then any nested include is
>   relative to that directory. (Issue #973)
>
> ### Options
>
> -
>
> ### Modified Opcodes and Gens
>
> - Add optional argument to ftom for rounding answer to integer.
>
> - flooper2 and syncgrain etc  now allow resampling.
>
> - chnclear can now take a list of channels to clear instead of just one.
>
> - printf and printf_i now are like the manual: all arguments beyond the
>   format and trigger are optional.
>
> - prints and printks can take string arguments printed with %s.
>
> - GEN2 can now take a size of zero, which is interpreted as size
>   sufficient for the number of values provided.
>
> - faustcompile now includes a new optional parameter to allow it to be
> run in a blocking mode. Defaults (as before) to non-blocking.
>
> - fillarray can be run at k-rate if any of the argument/values are k-rate.
>
> - slicearray for other than i-rate arrays runs at k-rate only.
>
> ### Utilities
>
> - A coding error in mixer was fixed.  It was very broken.
>
> ### Frontends
>
> - Belacsound:
>
> - CsoundQt:
>
> ### General Usage
>
> ## Bugs Fixed
>
> - A typo in p5glove meant that the command to read the button status
>   as a bitmap only returned state of button A.
>
> - diskin to array fixed and also use with small ksmps.
>
> - in loscil it sometimes failed to deal with the ibas argument; this
>   has now been reworked to be correct.
>
> - madsr could overflow an internal counter when given a negative p3.
>
> - Fixed mapping from threads to lua_States (issue #959).
>
> - the time calculation in flooper2, flooper, and syncgrain was corrected.
>
> - resampling and pitch fixed in pvstanal.
>
> - Rare buffer overflow case in faust opcodes fixed.
>
> - pvsftw had an incorrect check for fft format which led to incorrect
>   claim of bad format; fixed.
>
> - if ksmps was 1 the opcode linenr at arate failed to work; fixed.
>
> - window opcode fixed.
>
> - The test for compatible subtypes of f-values in a number of pvs
>   opcodes was wrong, causing spurious error messages.
>
> - cosseg was broken for more than one segment; now OK.
>
> - monitor opcode did not work correctly when using multiple processes; fixed.
>
> - in the linenr opcode if the release stage was entered before the end
>   of the rise the output value jumped to the end value and then
>   decayed, causing a glitch.  This is fixed. (#1048)
>
> # SYSTEM LEVEL CHANGES
>
> - the various -zN options now reports the number of opcodes for the request, so
> differs with respect to deprecated and polymorphic opcodes.
>
> ### System Changes
>
> - Recompilation of named instruments totally reworked to avoid errors
>   and memory leaks.
>
> - The allocation of instrument names to internal numbers has been
>   rewritten and should now be usable with replacements in live coding.
>
> - Printing the number allocated to a named instrument now behaves the
>   manual, not just for debug.
>
> - if liblo version 0.29 is available csound can be built to use it (with
>   a compiler flag LIBLO29) and this fixes some bugs related to
>   heavy/complex use of OSClisten.  Unfortunately the older 0.28 version
>   is being distributed by some Linux distros.
>
> - The orchestra compiler has a number of new optimisations, avoiding
> unnecessary assignments and doing some more expression optimisations.
>
> ### Translations
>
> ### API
>
> - csound->ReadScore was changed so it behaves the same as a score in a
>   csd or sco file.  This could incorrectly give an infinite score or not
>   in unexpected cases.
>
> ### Platform Specific
>
> - WebAudio:
>
> - iOS
>
> - Android
>
> - Windows
>
> - MacOS
>
> - GNU/Linux
>
> - Bela
>  - allow analog in and out with different channel numbers.
>
> ==END==
>
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

Doctor B
In reply to this post by jpff
Wonderful new things and (for me especially) some super important bug fixes!  Can’t wait.  Thanks to all the developers for all the great work.

Dr. Richard Boulanger
Professor
Electronic Production and Design
Berklee College of Music

> On Oct 18, 2018, at 9:18 AM, John <[hidden email]> wrote:
>
> All being well we intend releasing v6.12 next week, probably Monday.
>
> Below are the draft release notes.  Any comments or corrections welcome
>
> ==John ffitch
>
> DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT
>
>
>
> # CSOUND VERSION 6.12 RELEASE NOTES
>
> Many changes including the removal of vst2cs functionality due to a copyright issue.
>
> The changes made in 6.11 to raw format reading have been modified so
> gen1 and diskin ignore positive file formats and use the file header,
> unless the format is negative when it uses the absolute value in a raw
> audio file.  This should preserve most compatibility issues.
>
> There are a number of new and improved opcodes, new facilities in
> scores and many bug fixes.
>
>
> -- The Developers
>
> ## USER-LEVEL CHANGES
>
> ### New opcodes
>
> - fluidinfo retrieves program information from a currently loaded soundfont.
>
> - New opcode ftaudio writes a ftable to an audio file; irate and triggered
>    k-rate version exist and k-rate version supports sync or async writing.
>
> - Version of OSClisten that writes the data to a k-rate array now exists.
>
> - OSCcount returns the number of OSC incoming messages pending.
>
> - faustplay and faustdsp are new opcodes splitting the faust DSP
>  instantiation and performance.
>
> - OSCbundle sends a collection of similar OSC messages as a single packet
>  for efficiency.
>
> - beosc and beadsyn are band enhanced oscillator and oscillator bank.
>  bpfcos is for breakpoint files with interpolation.  Similarly lincos
>  adds cosine interpolation to linlin.
>
> - printarray and reshapearray act on arrays.
>
> - trim and trim_i adjust the size of a 1 dimensional array, either
>  bigger or smaller, preserving data and/or padding with zeros.
>
> ### New Gen and Macros
>
> -
>
> ### Orchestra
>
> - New preprocessor option #includestr.  This is like #include but has
>  macro expansion in the double-quote delimited string.
>
> - Use of tied notes in subinstr fixed.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.
>
> - runtime error message now (usually) include a line number and a file/macro trace.
>
> - multiple assignments such as ka,kb=1,2 are again supported by the
>  parser; it had inadvertently got lost.
>
> - the problematic use of i() with an array element directly is now
>  flagged as an error.
>
> - If an included file name has a "/" in it then any nested include is
>  relative to that directory. (Issue #973)
>
> ### Score
>
> - New preprocessor option #includestr.  This is like #include but has
>  macro expansion in the double-quote delimited string
>
> - 'd' score opcode for real-time performance.  This was issue #966.
>
> - bug in np operation fixed.
>
> - Use of [] syntax in a score could lead to a loss of precision for
>  numbers over about 1 million; older version restored.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.
>
> - The forms for delayed ending of sections (e 5 or s 5) now work with
>  fractional delays; previously only read the integer part.
>
> - If an included file name has a "/" in it then any nested include is
>  relative to that directory. (Issue #973)
>
> ### Options
>
> -
>
> ### Modified Opcodes and Gens
>
> - Add optional argument to ftom for rounding answer to integer.
>
> - flooper2 and syncgrain etc  now allow resampling.
>
> - chnclear can now take a list of channels to clear instead of just one.
>
> - printf and printf_i now are like the manual: all arguments beyond the
>  format and trigger are optional.
>
> - prints and printks can take string arguments printed with %s.
>
> - GEN2 can now take a size of zero, which is interpreted as size
>  sufficient for the number of values provided.
>
> - faustcompile now includes a new optional parameter to allow it to be
> run in a blocking mode. Defaults (as before) to non-blocking.
>
> - fillarray can be run at k-rate if any of the argument/values are k-rate.
>
> - slicearray for other than i-rate arrays runs at k-rate only.
>
> ### Utilities
>
> - A coding error in mixer was fixed.  It was very broken.
>
> ### Frontends
>
> - Belacsound:
>
> - CsoundQt:
>
> ### General Usage
>
> ## Bugs Fixed
>
> - A typo in p5glove meant that the command to read the button status
>  as a bitmap only returned state of button A.
>
> - diskin to array fixed and also use with small ksmps.
>
> - in loscil it sometimes failed to deal with the ibas argument; this
>  has now been reworked to be correct.
>
> - madsr could overflow an internal counter when given a negative p3.
>
> - Fixed mapping from threads to lua_States (issue #959).
>
> - the time calculation in flooper2, flooper, and syncgrain was corrected.
>
> - resampling and pitch fixed in pvstanal.
>
> - Rare buffer overflow case in faust opcodes fixed.
>
> - pvsftw had an incorrect check for fft format which led to incorrect
>  claim of bad format; fixed.
>
> - if ksmps was 1 the opcode linenr at arate failed to work; fixed.
>
> - window opcode fixed.
>
> - The test for compatible subtypes of f-values in a number of pvs
>  opcodes was wrong, causing spurious error messages.
>
> - cosseg was broken for more than one segment; now OK.
>
> - monitor opcode did not work correctly when using multiple processes; fixed.
>
> - in the linenr opcode if the release stage was entered before the end
>  of the rise the output value jumped to the end value and then
>  decayed, causing a glitch.  This is fixed. (#1048)
>
> # SYSTEM LEVEL CHANGES
>
> - the various -zN options now reports the number of opcodes for the request, so
> differs with respect to deprecated and polymorphic opcodes.
>
> ### System Changes
>
> - Recompilation of named instruments totally reworked to avoid errors
>  and memory leaks.
>
> - The allocation of instrument names to internal numbers has been
>  rewritten and should now be usable with replacements in live coding.
>
> - Printing the number allocated to a named instrument now behaves the
>  manual, not just for debug.
>
> - if liblo version 0.29 is available csound can be built to use it (with
>  a compiler flag LIBLO29) and this fixes some bugs related to
>  heavy/complex use of OSClisten.  Unfortunately the older 0.28 version
>  is being distributed by some Linux distros.
>
> - The orchestra compiler has a number of new optimisations, avoiding
> unnecessary assignments and doing some more expression optimisations.
>
> ### Translations
>
> ### API
>
> - csound->ReadScore was changed so it behaves the same as a score in a
>  csd or sco file.  This could incorrectly give an infinite score or not
>  in unexpected cases.
>
> ### Platform Specific
>
> - WebAudio:
>
> - iOS
>
> - Android
>
> - Windows
>
> - MacOS
>
> - GNU/Linux
>
> - Bela
> - allow analog in and out with different channel numbers.
>
> ==END==
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

hlolli
Could someone look at and merge https://github.com/csound/csound/pull/1029 before releasing. Without this, it's impossible to specify more than 1 opcodedir with env-flags.

On Thu, 18 Oct 2018 at 16:06, Dr. Richard Boulanger <[hidden email]> wrote:
Wonderful new things and (for me especially) some super important bug fixes!  Can’t wait.  Thanks to all the developers for all the great work.

Dr. Richard Boulanger
Professor
Electronic Production and Design
Berklee College of Music

> On Oct 18, 2018, at 9:18 AM, John <[hidden email]> wrote:
>
> All being well we intend releasing v6.12 next week, probably Monday.
>
> Below are the draft release notes.  Any comments or corrections welcome
>
> ==John ffitch
>
> DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT
>
>
>
> # CSOUND VERSION 6.12 RELEASE NOTES
>
> Many changes including the removal of vst2cs functionality due to a copyright issue.
>
> The changes made in 6.11 to raw format reading have been modified so
> gen1 and diskin ignore positive file formats and use the file header,
> unless the format is negative when it uses the absolute value in a raw
> audio file.  This should preserve most compatibility issues.
>
> There are a number of new and improved opcodes, new facilities in
> scores and many bug fixes.
>
>
> -- The Developers
>
> ## USER-LEVEL CHANGES
>
> ### New opcodes
>
> - fluidinfo retrieves program information from a currently loaded soundfont.
>
> - New opcode ftaudio writes a ftable to an audio file; irate and triggered
>    k-rate version exist and k-rate version supports sync or async writing.
>
> - Version of OSClisten that writes the data to a k-rate array now exists.
>
> - OSCcount returns the number of OSC incoming messages pending.
>
> - faustplay and faustdsp are new opcodes splitting the faust DSP
>  instantiation and performance.
>
> - OSCbundle sends a collection of similar OSC messages as a single packet
>  for efficiency.
>
> - beosc and beadsyn are band enhanced oscillator and oscillator bank.
>  bpfcos is for breakpoint files with interpolation.  Similarly lincos
>  adds cosine interpolation to linlin.
>
> - printarray and reshapearray act on arrays.
>
> - trim and trim_i adjust the size of a 1 dimensional array, either
>  bigger or smaller, preserving data and/or padding with zeros.
>
> ### New Gen and Macros
>
> -
>
> ### Orchestra
>
> - New preprocessor option #includestr.  This is like #include but has
>  macro expansion in the double-quote delimited string.
>
> - Use of tied notes in subinstr fixed.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.
>
> - runtime error message now (usually) include a line number and a file/macro trace.
>
> - multiple assignments such as ka,kb=1,2 are again supported by the
>  parser; it had inadvertently got lost.
>
> - the problematic use of i() with an array element directly is now
>  flagged as an error.
>
> - If an included file name has a "/" in it then any nested include is
>  relative to that directory. (Issue #973)
>
> ### Score
>
> - New preprocessor option #includestr.  This is like #include but has
>  macro expansion in the double-quote delimited string
>
> - 'd' score opcode for real-time performance.  This was issue #966.
>
> - bug in np operation fixed.
>
> - Use of [] syntax in a score could lead to a loss of precision for
>  numbers over about 1 million; older version restored.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.
>
> - The forms for delayed ending of sections (e 5 or s 5) now work with
>  fractional delays; previously only read the integer part.
>
> - If an included file name has a "/" in it then any nested include is
>  relative to that directory. (Issue #973)
>
> ### Options
>
> -
>
> ### Modified Opcodes and Gens
>
> - Add optional argument to ftom for rounding answer to integer.
>
> - flooper2 and syncgrain etc  now allow resampling.
>
> - chnclear can now take a list of channels to clear instead of just one.
>
> - printf and printf_i now are like the manual: all arguments beyond the
>  format and trigger are optional.
>
> - prints and printks can take string arguments printed with %s.
>
> - GEN2 can now take a size of zero, which is interpreted as size
>  sufficient for the number of values provided.
>
> - faustcompile now includes a new optional parameter to allow it to be
> run in a blocking mode. Defaults (as before) to non-blocking.
>
> - fillarray can be run at k-rate if any of the argument/values are k-rate.
>
> - slicearray for other than i-rate arrays runs at k-rate only.
>
> ### Utilities
>
> - A coding error in mixer was fixed.  It was very broken.
>
> ### Frontends
>
> - Belacsound:
>
> - CsoundQt:
>
> ### General Usage
>
> ## Bugs Fixed
>
> - A typo in p5glove meant that the command to read the button status
>  as a bitmap only returned state of button A.
>
> - diskin to array fixed and also use with small ksmps.
>
> - in loscil it sometimes failed to deal with the ibas argument; this
>  has now been reworked to be correct.
>
> - madsr could overflow an internal counter when given a negative p3.
>
> - Fixed mapping from threads to lua_States (issue #959).
>
> - the time calculation in flooper2, flooper, and syncgrain was corrected.
>
> - resampling and pitch fixed in pvstanal.
>
> - Rare buffer overflow case in faust opcodes fixed.
>
> - pvsftw had an incorrect check for fft format which led to incorrect
>  claim of bad format; fixed.
>
> - if ksmps was 1 the opcode linenr at arate failed to work; fixed.
>
> - window opcode fixed.
>
> - The test for compatible subtypes of f-values in a number of pvs
>  opcodes was wrong, causing spurious error messages.
>
> - cosseg was broken for more than one segment; now OK.
>
> - monitor opcode did not work correctly when using multiple processes; fixed.
>
> - in the linenr opcode if the release stage was entered before the end
>  of the rise the output value jumped to the end value and then
>  decayed, causing a glitch.  This is fixed. (#1048)
>
> # SYSTEM LEVEL CHANGES
>
> - the various -zN options now reports the number of opcodes for the request, so
> differs with respect to deprecated and polymorphic opcodes.
>
> ### System Changes
>
> - Recompilation of named instruments totally reworked to avoid errors
>  and memory leaks.
>
> - The allocation of instrument names to internal numbers has been
>  rewritten and should now be usable with replacements in live coding.
>
> - Printing the number allocated to a named instrument now behaves the
>  manual, not just for debug.
>
> - if liblo version 0.29 is available csound can be built to use it (with
>  a compiler flag LIBLO29) and this fixes some bugs related to
>  heavy/complex use of OSClisten.  Unfortunately the older 0.28 version
>  is being distributed by some Linux distros.
>
> - The orchestra compiler has a number of new optimisations, avoiding
> unnecessary assignments and doing some more expression optimisations.
>
> ### Translations
>
> ### API
>
> - csound->ReadScore was changed so it behaves the same as a score in a
>  csd or sco file.  This could incorrectly give an infinite score or not
>  in unexpected cases.
>
> ### Platform Specific
>
> - WebAudio:
>
> - iOS
>
> - Android
>
> - Windows
>
> - MacOS
>
> - GNU/Linux
>
> - Bela
> - allow analog in and out with different channel numbers.
>
> ==END==
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

Tarmo Johannes-3
In reply to this post by jpff
Great news!

What about installers? If CsoundQt is included, the current stable version is
still 0.9.5 as in two previous releases. CsoundQT 0.9.6-beta is available, but
not ready for release, though.

Should I do maybe new builds against Csound 6.12 to be sure it works? Are
there binaries for MacOs and Windows available somewhere?

greetings,
tarmo

On Thursday, October 18, 2018 4:18:16 PM EEST you wrote:

> All being well we intend releasing v6.12 next week, probably Monday.
>
> Below are the draft release notes.  Any comments or corrections welcome
>
> ==John ffitch
>
> DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT  
> DRAFT
>
>
>
> # CSOUND VERSION 6.12 RELEASE NOTES
>
> Many changes including the removal of vst2cs functionality due to a
> copyright issue.
>
> The changes made in 6.11 to raw format reading have been modified so
> gen1 and diskin ignore positive file formats and use the file header,
> unless the format is negative when it uses the absolute value in a raw
> audio file.  This should preserve most compatibility issues.
>
> There are a number of new and improved opcodes, new facilities in
> scores and many bug fixes.
>
>
> -- The Developers
>
> ## USER-LEVEL CHANGES
>
> ### New opcodes
>
> - fluidinfo retrieves program information from a currently loaded soundfont.
>
> - New opcode ftaudio writes a ftable to an audio file; irate and triggered
>     k-rate version exist and k-rate version supports sync or async writing.
>
> - Version of OSClisten that writes the data to a k-rate array now exists.
>
> - OSCcount returns the number of OSC incoming messages pending.
>
> - faustplay and faustdsp are new opcodes splitting the faust DSP
>   instantiation and performance.
>
> - OSCbundle sends a collection of similar OSC messages as a single packet
>   for efficiency.
>
> - beosc and beadsyn are band enhanced oscillator and oscillator bank.
>   bpfcos is for breakpoint files with interpolation.  Similarly lincos
>   adds cosine interpolation to linlin.
>
> - printarray and reshapearray act on arrays.
>
> - trim and trim_i adjust the size of a 1 dimensional array, either
>   bigger or smaller, preserving data and/or padding with zeros.
>
> ### New Gen and Macros
>
> -
>
> ### Orchestra
>
> - New preprocessor option #includestr.  This is like #include but has
>   macro expansion in the double-quote delimited string.
>
> - Use of tied notes in subinstr fixed.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited
> nesting works.
>
> - runtime error message now (usually) include a line number and a file/macro
> trace.
>
> - multiple assignments such as ka,kb=1,2 are again supported by the
>   parser; it had inadvertently got lost.
>
> - the problematic use of i() with an array element directly is now
>   flagged as an error.
>
> - If an included file name has a "/" in it then any nested include is
>   relative to that directory. (Issue #973)
>
> ### Score
>
> - New preprocessor option #includestr.  This is like #include but has
>   macro expansion in the double-quote delimited string
>
> - 'd' score opcode for real-time performance.  This was issue #966.
>
> - bug in np operation fixed.
>
> - Use of [] syntax in a score could lead to a loss of precision for
>   numbers over about 1 million; older version restored.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited
> nesting works.
>
> - The forms for delayed ending of sections (e 5 or s 5) now work with
>   fractional delays; previously only read the integer part.
>
> - If an included file name has a "/" in it then any nested include is
>   relative to that directory. (Issue #973)
>
> ### Options
>
> -
>
> ### Modified Opcodes and Gens
>
> - Add optional argument to ftom for rounding answer to integer.
>
> - flooper2 and syncgrain etc  now allow resampling.
>
> - chnclear can now take a list of channels to clear instead of just one.
>
> - printf and printf_i now are like the manual: all arguments beyond the
>   format and trigger are optional.
>
> - prints and printks can take string arguments printed with %s.
>
> - GEN2 can now take a size of zero, which is interpreted as size
>   sufficient for the number of values provided.
>
> - faustcompile now includes a new optional parameter to allow it to be
> run in a blocking mode. Defaults (as before) to non-blocking.
>
> - fillarray can be run at k-rate if any of the argument/values are k-rate.
>
> - slicearray for other than i-rate arrays runs at k-rate only.
>
> ### Utilities
>
> - A coding error in mixer was fixed.  It was very broken.
>
> ### Frontends
>
> - Belacsound:
>
> - CsoundQt:
>
> ### General Usage
>
> ## Bugs Fixed
>
> - A typo in p5glove meant that the command to read the button status
>   as a bitmap only returned state of button A.
>
> - diskin to array fixed and also use with small ksmps.
>
> - in loscil it sometimes failed to deal with the ibas argument; this
>   has now been reworked to be correct.
>
> - madsr could overflow an internal counter when given a negative p3.
>
> - Fixed mapping from threads to lua_States (issue #959).
>
> - the time calculation in flooper2, flooper, and syncgrain was corrected.
>
> - resampling and pitch fixed in pvstanal.
>
> - Rare buffer overflow case in faust opcodes fixed.
>
> - pvsftw had an incorrect check for fft format which led to incorrect
>   claim of bad format; fixed.
>
> - if ksmps was 1 the opcode linenr at arate failed to work; fixed.
>
> - window opcode fixed.
>
> - The test for compatible subtypes of f-values in a number of pvs
>   opcodes was wrong, causing spurious error messages.
>
> - cosseg was broken for more than one segment; now OK.
>
> - monitor opcode did not work correctly when using multiple processes;
> fixed.
>
> - in the linenr opcode if the release stage was entered before the end
>   of the rise the output value jumped to the end value and then
>   decayed, causing a glitch.  This is fixed. (#1048)
>
> # SYSTEM LEVEL CHANGES
>
> - the various -zN options now reports the number of opcodes for the request,
> so differs with respect to deprecated and polymorphic opcodes.
>
> ### System Changes
>
> - Recompilation of named instruments totally reworked to avoid errors
>   and memory leaks.
>
> - The allocation of instrument names to internal numbers has been
>   rewritten and should now be usable with replacements in live coding.
>
> - Printing the number allocated to a named instrument now behaves the
>   manual, not just for debug.
>
> - if liblo version 0.29 is available csound can be built to use it (with
>   a compiler flag LIBLO29) and this fixes some bugs related to
>   heavy/complex use of OSClisten.  Unfortunately the older 0.28 version
>   is being distributed by some Linux distros.
>
> - The orchestra compiler has a number of new optimisations, avoiding
> unnecessary assignments and doing some more expression optimisations.
>
> ### Translations
>
> ### API
>
> - csound->ReadScore was changed so it behaves the same as a score in a
>   csd or sco file.  This could incorrectly give an infinite score or not
>   in unexpected cases.
>
> ### Platform Specific
>
> - WebAudio:
>
> - iOS
>
> - Android
>
> - Windows
>
> - MacOS
>
> - GNU/Linux
>
> - Bela
>  - allow analog in and out with different channel numbers.
>
> ==END==
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

Steven Yi
In reply to this post by jpff
Should add that:

- Redefinition of UDOs with zero output arguments fixed

and

WebAudio Csound updated: CsoundObj API should be backwards compatible
but additional methods added for channels, AudioWorklet and
ScriptProcessor Nodes encapsulated for individual reuse


On Thu, Oct 18, 2018 at 9:18 AM John <[hidden email]> wrote:

>
> All being well we intend releasing v6.12 next week, probably Monday.
>
> Below are the draft release notes.  Any comments or corrections welcome
>
> ==John ffitch
>
> DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT
>
>
>
> # CSOUND VERSION 6.12 RELEASE NOTES
>
> Many changes including the removal of vst2cs functionality due to a copyright issue.
>
> The changes made in 6.11 to raw format reading have been modified so
> gen1 and diskin ignore positive file formats and use the file header,
> unless the format is negative when it uses the absolute value in a raw
> audio file.  This should preserve most compatibility issues.
>
> There are a number of new and improved opcodes, new facilities in
> scores and many bug fixes.
>
>
> -- The Developers
>
> ## USER-LEVEL CHANGES
>
> ### New opcodes
>
> - fluidinfo retrieves program information from a currently loaded soundfont.
>
> - New opcode ftaudio writes a ftable to an audio file; irate and triggered
>     k-rate version exist and k-rate version supports sync or async writing.
>
> - Version of OSClisten that writes the data to a k-rate array now exists.
>
> - OSCcount returns the number of OSC incoming messages pending.
>
> - faustplay and faustdsp are new opcodes splitting the faust DSP
>   instantiation and performance.
>
> - OSCbundle sends a collection of similar OSC messages as a single packet
>   for efficiency.
>
> - beosc and beadsyn are band enhanced oscillator and oscillator bank.
>   bpfcos is for breakpoint files with interpolation.  Similarly lincos
>   adds cosine interpolation to linlin.
>
> - printarray and reshapearray act on arrays.
>
> - trim and trim_i adjust the size of a 1 dimensional array, either
>   bigger or smaller, preserving data and/or padding with zeros.
>
> ### New Gen and Macros
>
> -
>
> ### Orchestra
>
> - New preprocessor option #includestr.  This is like #include but has
>   macro expansion in the double-quote delimited string.
>
> - Use of tied notes in subinstr fixed.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.
>
> - runtime error message now (usually) include a line number and a file/macro trace.
>
> - multiple assignments such as ka,kb=1,2 are again supported by the
>   parser; it had inadvertently got lost.
>
> - the problematic use of i() with an array element directly is now
>   flagged as an error.
>
> - If an included file name has a "/" in it then any nested include is
>   relative to that directory. (Issue #973)
>
> ### Score
>
> - New preprocessor option #includestr.  This is like #include but has
>   macro expansion in the double-quote delimited string
>
> - 'd' score opcode for real-time performance.  This was issue #966.
>
> - bug in np operation fixed.
>
> - Use of [] syntax in a score could lead to a loss of precision for
>   numbers over about 1 million; older version restored.
>
> - Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.
>
> - The forms for delayed ending of sections (e 5 or s 5) now work with
>   fractional delays; previously only read the integer part.
>
> - If an included file name has a "/" in it then any nested include is
>   relative to that directory. (Issue #973)
>
> ### Options
>
> -
>
> ### Modified Opcodes and Gens
>
> - Add optional argument to ftom for rounding answer to integer.
>
> - flooper2 and syncgrain etc  now allow resampling.
>
> - chnclear can now take a list of channels to clear instead of just one.
>
> - printf and printf_i now are like the manual: all arguments beyond the
>   format and trigger are optional.
>
> - prints and printks can take string arguments printed with %s.
>
> - GEN2 can now take a size of zero, which is interpreted as size
>   sufficient for the number of values provided.
>
> - faustcompile now includes a new optional parameter to allow it to be
> run in a blocking mode. Defaults (as before) to non-blocking.
>
> - fillarray can be run at k-rate if any of the argument/values are k-rate.
>
> - slicearray for other than i-rate arrays runs at k-rate only.
>
> ### Utilities
>
> - A coding error in mixer was fixed.  It was very broken.
>
> ### Frontends
>
> - Belacsound:
>
> - CsoundQt:
>
> ### General Usage
>
> ## Bugs Fixed
>
> - A typo in p5glove meant that the command to read the button status
>   as a bitmap only returned state of button A.
>
> - diskin to array fixed and also use with small ksmps.
>
> - in loscil it sometimes failed to deal with the ibas argument; this
>   has now been reworked to be correct.
>
> - madsr could overflow an internal counter when given a negative p3.
>
> - Fixed mapping from threads to lua_States (issue #959).
>
> - the time calculation in flooper2, flooper, and syncgrain was corrected.
>
> - resampling and pitch fixed in pvstanal.
>
> - Rare buffer overflow case in faust opcodes fixed.
>
> - pvsftw had an incorrect check for fft format which led to incorrect
>   claim of bad format; fixed.
>
> - if ksmps was 1 the opcode linenr at arate failed to work; fixed.
>
> - window opcode fixed.
>
> - The test for compatible subtypes of f-values in a number of pvs
>   opcodes was wrong, causing spurious error messages.
>
> - cosseg was broken for more than one segment; now OK.
>
> - monitor opcode did not work correctly when using multiple processes; fixed.
>
> - in the linenr opcode if the release stage was entered before the end
>   of the rise the output value jumped to the end value and then
>   decayed, causing a glitch.  This is fixed. (#1048)
>
> # SYSTEM LEVEL CHANGES
>
> - the various -zN options now reports the number of opcodes for the request, so
> differs with respect to deprecated and polymorphic opcodes.
>
> ### System Changes
>
> - Recompilation of named instruments totally reworked to avoid errors
>   and memory leaks.
>
> - The allocation of instrument names to internal numbers has been
>   rewritten and should now be usable with replacements in live coding.
>
> - Printing the number allocated to a named instrument now behaves the
>   manual, not just for debug.
>
> - if liblo version 0.29 is available csound can be built to use it (with
>   a compiler flag LIBLO29) and this fixes some bugs related to
>   heavy/complex use of OSClisten.  Unfortunately the older 0.28 version
>   is being distributed by some Linux distros.
>
> - The orchestra compiler has a number of new optimisations, avoiding
> unnecessary assignments and doing some more expression optimisations.
>
> ### Translations
>
> ### API
>
> - csound->ReadScore was changed so it behaves the same as a score in a
>   csd or sco file.  This could incorrectly give an infinite score or not
>   in unexpected cases.
>
> ### Platform Specific
>
> - WebAudio:
>
> - iOS
>
> - Android
>
> - Windows
>
> - MacOS
>
> - GNU/Linux
>
> - Bela
>  - allow analog in and out with different channel numbers.
>
> ==END==
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

jpff
In reply to this post by Tarmo Johannes-3
A windows binary almost final is at
https://ci.appveyor.com/project/csound/csound/build/artifacts

No idea about OSX

On Thu, 18 Oct 2018, Tarmo Johannes wrote:

> Great news!
>
> What about installers? If CsoundQt is included, the current stable version is
> still 0.9.5 as in two previous releases. CsoundQT 0.9.6-beta is available, but
> not ready for release, though.
>
> Should I do maybe new builds against Csound 6.12 to be sure it works? Are
> there binaries for MacOs and Windows available somewhere?
>
> greetings,
> tarmo
>
k
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

jpff
In reply to this post by hlolli
I looked at that PR about search paths but it was not my code.
Hope Victor can review it
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

hlolli
Cool, I hope Victor finds time before release. Saves me time to remember to patch each time I compile from sources :)

But not the end of the world if csound is released without this. It seems that the lack of reports about this shows that not many users are dividing dynamic lib files into more than 1 directory, or do it in a different way I'm not familiar with.

On Thu, 18 Oct 2018 at 19:38, jpff <[hidden email]> wrote:
I looked at that PR about search paths but it was not my code.
Hope Victor can review it
Reply | Threaded
Open this post in threaded view
|

Re: [Csnd-dev] Release 6.12 is imminent

Mauro Giubileo
In reply to this post by jpff

It's always nice to see that a project you like is actively improved and developed!

Good work to everyone and thank you for your efforts!

Best Regards,
Mauro

 


Il 2018-10-18 15:18 John ha scritto:

All being well we intend releasing v6.12 next week, probably Monday.

Below are the draft release notes.  Any comments or corrections welcome

==John ffitch

DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT   DRAFT



# CSOUND VERSION 6.12 RELEASE NOTES

Many changes including the removal of vst2cs functionality due to a copyright issue.

The changes made in 6.11 to raw format reading have been modified so
gen1 and diskin ignore positive file formats and use the file header,
unless the format is negative when it uses the absolute value in a raw
audio file.  This should preserve most compatibility issues.

There are a number of new and improved opcodes, new facilities in
scores and many bug fixes.


-- The Developers

## USER-LEVEL CHANGES

### New opcodes

- fluidinfo retrieves program information from a currently loaded soundfont.

- New opcode ftaudio writes a ftable to an audio file; irate and triggered
    k-rate version exist and k-rate version supports sync or async writing.

- Version of OSClisten that writes the data to a k-rate array now exists.

- OSCcount returns the number of OSC incoming messages pending.

- faustplay and faustdsp are new opcodes splitting the faust DSP
  instantiation and performance.

- OSCbundle sends a collection of similar OSC messages as a single packet
  for efficiency.

- beosc and beadsyn are band enhanced oscillator and oscillator bank.
  bpfcos is for breakpoint files with interpolation.  Similarly lincos
  adds cosine interpolation to linlin.

- printarray and reshapearray act on arrays.

- trim and trim_i adjust the size of a 1 dimensional array, either
  bigger or smaller, preserving data and/or padding with zeros.

### New Gen and Macros

-

### Orchestra

- New preprocessor option #includestr.  This is like #include but has
  macro expansion in the double-quote delimited string.

- Use of tied notes in subinstr fixed.

- Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.

- runtime error message now (usually) include a line number and a file/macro trace.

- multiple assignments such as ka,kb=1,2 are again supported by the
  parser; it had inadvertently got lost.

- the problematic use of i() with an array element directly is now
  flagged as an error.

- If an included file name has a "/" in it then any nested include is
  relative to that directory. (Issue #973)

### Score

- New preprocessor option #includestr.  This is like #include but has
  macro expansion in the double-quote delimited string

- 'd' score opcode for real-time performance.  This was issue #966.

- bug in np operation fixed.

- Use of [] syntax in a score could lead to a loss of precision for
  numbers over about 1 million; older version restored.

- Nesting macro calls more than about 10 caused a crash; now unlimited nesting works.

- The forms for delayed ending of sections (e 5 or s 5) now work with
  fractional delays; previously only read the integer part.

- If an included file name has a "/" in it then any nested include is
  relative to that directory. (Issue #973)

### Options

-

### Modified Opcodes and Gens

- Add optional argument to ftom for rounding answer to integer.

- flooper2 and syncgrain etc  now allow resampling.

- chnclear can now take a list of channels to clear instead of just one.

- printf and printf_i now are like the manual: all arguments beyond the
  format and trigger are optional.

- prints and printks can take string arguments printed with %s.

- GEN2 can now take a size of zero, which is interpreted as size
  sufficient for the number of values provided.

- faustcompile now includes a new optional parameter to allow it to be
run in a blocking mode. Defaults (as before) to non-blocking.

- fillarray can be run at k-rate if any of the argument/values are k-rate.

- slicearray for other than i-rate arrays runs at k-rate only.

### Utilities

- A coding error in mixer was fixed.  It was very broken.

### Frontends

- Belacsound:

- CsoundQt:

### General Usage

## Bugs Fixed

- A typo in p5glove meant that the command to read the button status
  as a bitmap only returned state of button A.

- diskin to array fixed and also use with small ksmps.

- in loscil it sometimes failed to deal with the ibas argument; this
  has now been reworked to be correct.

- madsr could overflow an internal counter when given a negative p3.

- Fixed mapping from threads to lua_States (issue #959).

- the time calculation in flooper2, flooper, and syncgrain was corrected.

- resampling and pitch fixed in pvstanal.

- Rare buffer overflow case in faust opcodes fixed.

- pvsftw had an incorrect check for fft format which led to incorrect
  claim of bad format; fixed.

- if ksmps was 1 the opcode linenr at arate failed to work; fixed.

- window opcode fixed.

- The test for compatible subtypes of f-values in a number of pvs
  opcodes was wrong, causing spurious error messages.

- cosseg was broken for more than one segment; now OK.

- monitor opcode did not work correctly when using multiple processes; fixed.

- in the linenr opcode if the release stage was entered before the end
  of the rise the output value jumped to the end value and then
  decayed, causing a glitch.  This is fixed. (#1048)

# SYSTEM LEVEL CHANGES

- the various -zN options now reports the number of opcodes for the request, so
differs with respect to deprecated and polymorphic opcodes.

### System Changes

- Recompilation of named instruments totally reworked to avoid errors
  and memory leaks.

- The allocation of instrument names to internal numbers has been
  rewritten and should now be usable with replacements in live coding.

- Printing the number allocated to a named instrument now behaves the
  manual, not just for debug.

- if liblo version 0.29 is available csound can be built to use it (with
  a compiler flag LIBLO29) and this fixes some bugs related to
  heavy/complex use of OSClisten.  Unfortunately the older 0.28 version
  is being distributed by some Linux distros.

- The orchestra compiler has a number of new optimisations, avoiding
unnecessary assignments and doing some more expression optimisations.

### Translations

### API

- csound->ReadScore was changed so it behaves the same as a score in a
  csd or sco file.  This could incorrectly give an infinite score or not
  in unexpected cases.

### Platform Specific

- WebAudio:

- iOS

- Android

- Windows

- MacOS

- GNU/Linux

- Bela
 - allow analog in and out with different channel numbers.

==END==