Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-20 Thread Jonathan Wilkes


--- On Sat, 9/19/09, Mathieu Bouchard  wrote:

> From: Mathieu Bouchard 
> Subject: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch 
> Authors]
> To: pd-list@iem.at
> Date: Saturday, September 19, 2009, 9:43 PM
> 
> On Sat, 5 Sep 2009, Jonathan Wilkes wrote:
> 
> > I've been assuming that one of the aims of tags (i.e.,
> keywords) is that there would be a search window in the
> browser so you can search for relevant help patches/tutorial
> patches/ etc.  Or maybe an actual patch, since Hans has
> said that the keywords should be "parsable in pd." There are
> already some categories from which simple, standard keywords
> may be used, but for some terms like "frequency modulation"
> there's the aforementioned problem of the space between the
> words.  Given that, frequency_modulation is certainly
> one solution, and users can get used to using underscores
> when searching, but why not also include "fm" and
> "modulation" in case the user happens to type that (which
> would be completely reasonable)?
> 
> Typically, with tags, the user has access to the list of
> all currently existing tags, so that the user can choose
> from there. You could have a feature to search by regexp
> and/or thesaurus on the list of tags, but when you look for
> a tag in patches, the goal is sort of to have one concept
> per tag and one tag per concept, no synonyms, no homonyms.
> 
> > > What's a tag?
> > A keyword.
> 
> That's pretty terse. What are the expectations about tags?
> what about synonyms and homonyms? what do the tags mean,
> that the whole text of the patch doesn't?
> 
> Because, what's the advantage of searching in tags vs
> searching in the whole text?

I thought it was mostly a speed issue, in which searching for 3 or 4 
tags in a single file would be quicker than doing a full text search 
of every help patch.  But if the user has access to all the tags and 
goes from there, I see your point about naming conventions.

> 
> > I saw your nick in the irc discussion of the PDDP,
> didn't you actually
> > take part in some of these decisions?
> 
> What??... what does "take part" mean to you? I don't
> understand.

I remember reading something about the "desiredata people" wanting 
related objects in their own subpatch, and I think I read some comments 
you had about putting general object behavior in a centralized location 
like an abstraction, rather than repeating it for every help file.

> 
> I don't recall my opinions being represented in PDDP, but
> then, I don't recall trying to push them.

Why didn't you push (for) them?

-Jonathan

> 
>  _ _ __ ___ _  _
> _ ...
> | Mathieu Bouchard, Montréal, Québec. téléphone:
> +1.514.383.3801
> -Inline Attachment Follows-
> 
> ___
> Pd-list@iem.at
> mailing list
> UNSUBSCRIBE and account-management -> 
> http://lists.puredata.info/listinfo/pd-list
> 


  

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


[PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-19 Thread Mathieu Bouchard


On Sat, 5 Sep 2009, Jonathan Wilkes wrote:

I've been assuming that one of the aims of tags (i.e., keywords) is that 
there would be a search window in the browser so you can search for 
relevant help patches/tutorial patches/ etc.  Or maybe an actual patch, 
since Hans has said that the keywords should be "parsable in pd." There 
are already some categories from which simple, standard keywords may be 
used, but for some terms like "frequency modulation" there's the 
aforementioned problem of the space between the words.  Given that, 
frequency_modulation is certainly one solution, and users can get used 
to using underscores when searching, but why not also include "fm" and 
"modulation" in case the user happens to type that (which would be 
completely reasonable)?


Typically, with tags, the user has access to the list of all currently 
existing tags, so that the user can choose from there. You could have a 
feature to search by regexp and/or thesaurus on the list of tags, but when 
you look for a tag in patches, the goal is sort of to have one concept per 
tag and one tag per concept, no synonyms, no homonyms.



> What's a tag?
A keyword.


That's pretty terse. What are the expectations about tags? what about 
synonyms and homonyms? what do the tags mean, that the whole text of the 
patch doesn't?


Because, what's the advantage of searching in tags vs searching in the 
whole text?



I saw your nick in the irc discussion of the PDDP, didn't you actually
take part in some of these decisions?


What??... what does "take part" mean to you? I don't understand.

I don't recall my opinions being represented in PDDP, but then, I don't 
recall trying to push them.


 _ _ __ ___ _  _ _ ...
| Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-19 Thread Mathieu Bouchard

On Sat, 5 Sep 2009, Martin Peach wrote:

As I see it a tag like "fm" would be associated with synonyms like 
"frequency modulation", "frequency_modulation", "Frequency Modulation" 
"modulation de frequence", "DX7 synthesis" that could be added to it at 
any later date, so a search for any of the synonyms would ultimately 
return whatever was connected to the tag.


Yeah, either considering a tag to have several names (hardlinks), or a tag
to have a real name and several aliases (symlinks), or something like
that. This is good when there is no central authority that gathers all
tags together before they begin to be used, or when even the central
authority is not sufficient anymore because people can't figure out 
synonyms at the right time (too many tags, insufficient search engine, 
etc).


 _ _ __ ___ _  _ _ ...
| Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Martin Peach

Jonathan Wilkes wrote:


I've been assuming that one of the aims of tags (i.e., keywords) is that 
there would be a search window in the browser so you can search for 
relevant help patches/tutorial patches/ etc.  Or maybe an actual patch, 
since Hans has said that the keywords should be "parsable in pd."  
There are already some categories from which simple, standard keywords 
may be used, but for some terms like "frequency modulation" there's the 
aforementioned problem of the space between the words.  Given that, 
frequency_modulation is certainly one solution, and users can get 
used to using underscores when searching, but why not also include "fm" 
and "modulation" in case the user happens to type that (which would be 
completely reasonable)


As I see it a tag like "fm" would be associated with synonyms like 
"frequency modulation", "frequency_modulation", "Frequency Modulation" 
"modulation de frequence", "DX7 synthesis" that could be added to it at 
any later date, so a search for any of the synonyms would ultimately 
return whatever was connected to the tag.


Martin

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Jonathan Wilkes


--- On Sun, 9/6/09, Mathieu Bouchard  wrote:

> From: Mathieu Bouchard 
> Subject: Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch 
> Authors]
> To: "Jonathan Wilkes" 
> Cc: "Frank Barknecht" , pd-list@iem.at, "João Pais" 
> 
> Date: Sunday, September 6, 2009, 1:34 AM
> On Sat, 5 Sep 2009, Jonathan Wilkes
> wrote:
> 
> > I assume your talking about the HELP_PATCH_AUTHOR
> field, and not
> > AUTHOR.  (But maybe there are better names for
> these.)
> 
> Yes, I confused the two with my example, but let's say that
> the authorship of one object-class is "Johannes wrote the
> original version, Martin added a TCP transport module" or
> "Paquette wrote the whole abstraction, but we'd like to
> thank the XYZ institute for encumbering it with a patent".
> There are numerous things that one may want to put in an
> AUTHOR field in the real world. It's not like a university
> lab, where you just have to write the name of your advisor
> in that field and not think twice (except if you are unlucky
> enough to have several advisors at once).
> 
> > I don't understand your last point, as far as parsing
> in Pd is concerned. If you have the tag
> "frequency_modulation," how is it that the user can search
> for "fm" and get appropriate results?
> 
> They can't. Why should they?

I've been assuming that one of the aims of tags (i.e., keywords) is that 
there would be a search window in the browser so you can search for 
relevant help patches/tutorial patches/ etc.  Or maybe an actual patch, 
since Hans has said that the keywords should be "parsable in pd."  
There are already some categories from which simple, standard keywords 
may be used, but for some terms like "frequency modulation" there's the 
aforementioned problem of the space between the words.  Given that, 
frequency_modulation is certainly one solution, and users can get 
used to using underscores when searching, but why not also include "fm" 
and "modulation" in case the user happens to type that (which would be 
completely reasonable)?

> 
> What's a tag?

A keyword.

> 
> Tell me what's a tag and what's the point of having tags.

I saw your nick in the irc discussion of the PDDP, didn't you actually 
take part in some of these decisions?  Anyhow, here's a link to the 
specification:

http://puredata.info/dev/pddp/SpecificationForHelpPatchSearching

> 
>  _ _ __ ___ _  _
> _ ...
> | Mathieu Bouchard, Montréal, Québec. téléphone:
> +1.514.383.3801


  

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Jonathan Wilkes


--- On Sat, 9/5/09, Mathieu Bouchard  wrote:

> From: Mathieu Bouchard 
> Subject: Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch 
> Authors]
> To: "Jonathan Wilkes" 
> Cc: "Frank Barknecht" , pd-list@iem.at, "João Pais" 
> 
> Date: Saturday, September 5, 2009, 8:59 PM
> On Fri, 4 Sep 2009, Jonathan Wilkes
> wrote:
> 
> > But then what to do about multiple help patch
> authors?
> 
> I don't know, but if a template doesn't allow to say
> something complex like «Pierre wrote the reference section,
> Jean wrote the theoretical background section, Jacques
> provided the examples» then that template should be just
> shot dead.

You have included commas in your example.  But I thought you were 
arguing that commas will cause problems down the line.

-Jonathan

> 
>  _ _ __ ___ _  _
> _ ...
> | Mathieu Bouchard, Montréal, Québec. téléphone:
> +1.514.383.3801


  

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Mathieu Bouchard

On Sat, 5 Sep 2009, Jonathan Wilkes wrote:


I assume your talking about the HELP_PATCH_AUTHOR field, and not
AUTHOR.  (But maybe there are better names for these.)


Yes, I confused the two with my example, but let's say that the authorship 
of one object-class is "Johannes wrote the original version, Martin added 
a TCP transport module" or "Paquette wrote the whole abstraction, but we'd 
like to thank the XYZ institute for encumbering it with a patent". There 
are numerous things that one may want to put in an AUTHOR field in the 
real world. It's not like a university lab, where you just have to write 
the name of your advisor in that field and not think twice (except if you 
are unlucky enough to have several advisors at once).


I don't understand your last point, as far as parsing in Pd is 
concerned. If you have the tag "frequency_modulation," how is it that 
the user can search for "fm" and get appropriate results?


They can't. Why should they?

What's a tag?

Tell me what's a tag and what's the point of having tags.

 _ _ __ ___ _  _ _ ...
| Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Jonathan Wilkes


--- On Sat, 9/5/09, Mathieu Bouchard  wrote:

> From: Mathieu Bouchard 
> Subject: Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch 
> Authors]
> To: "Jonathan Wilkes" 
> Cc: "Frank Barknecht" , pd-list@iem.at, "João Pais" 
> 
> Date: Saturday, September 5, 2009, 11:39 PM
> On Sat, 5 Sep 2009, Jonathan Wilkes
> wrote:
> > --- On Sat, 9/5/09, Mathieu Bouchard 
> wrote:
> >> I don't know, but if a template doesn't allow to
> say
> >> something complex like «Pierre wrote the
> reference section,
> >> Jean wrote the theoretical background section,
> Jacques
> >> provided the examples» then that template should
> be just
> >> shot dead.
> > You have included commas in your example.  But I
> thought you were
> > arguing that commas will cause problems down the
> line.
> 
> If you are making a tag system and are going to search by
> full tagnames, than each tag has better be a single symbol
> and each request better not contain commas. But there are
> things for which you'd search by text content, and people
> are using commas all over english text, so naturally, if the
> author field needs to be in plain english, it will need to
> be searched like plain english is to be searched.

I assume your talking about the HELP_PATCH_AUTHOR field, and not 
AUTHOR.  (But maybe there are better names for these.)

> 
> I'm supposing that a tag system is something in which a tag
> is atomic (you don't search for substrings of tags when you
> search by tag) and in which tags are considered more special
> than the rest of the text, because else why bother calling
> it a tag system and why bother making it any different from
> a plain text search. furthermore, i suppose that there is an
> aim of semantic unambiguïty, that is, there is some kind of
> standard on tags such that either "fm" or
> "frequency_modulation" is used but not both because all
> synonymous words for 1 thing are represented by 1 tag.

I don't understand your last point, as far as parsing in Pd is concerned. 
If you have the tag "frequency_modulation," how is it that the user can 
search for "fm" and get appropriate results?

-Jonathan

> 
>  _ _ __ ___ _  _
> _ ...
> | Mathieu Bouchard, Montréal, Québec. téléphone:
> +1.514.383.3801


  

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Mathieu Bouchard

On Sat, 5 Sep 2009, Jonathan Wilkes wrote:

--- On Sat, 9/5/09, Mathieu Bouchard  wrote:

I don't know, but if a template doesn't allow to say
something complex like «Pierre wrote the reference section,
Jean wrote the theoretical background section, Jacques
provided the examples» then that template should be just
shot dead.

You have included commas in your example.  But I thought you were
arguing that commas will cause problems down the line.


If you are making a tag system and are going to search by full tagnames, 
than each tag has better be a single symbol and each request better not 
contain commas. But there are things for which you'd search by text 
content, and people are using commas all over english text, so naturally, 
if the author field needs to be in plain english, it will need to be 
searched like plain english is to be searched.


I'm supposing that a tag system is something in which a tag is atomic (you 
don't search for substrings of tags when you search by tag) and in which 
tags are considered more special than the rest of the text, because else 
why bother calling it a tag system and why bother making it any different 
from a plain text search. furthermore, i suppose that there is an aim of 
semantic unambiguïty, that is, there is some kind of standard on tags such 
that either "fm" or "frequency_modulation" is used but not both because 
all synonymous words for 1 thing are represented by 1 tag.


 _ _ __ ___ _  _ _ ...
| Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Jonathan Wilkes


--- On Sat, 9/5/09, Mathieu Bouchard  wrote:

> From: Mathieu Bouchard 
> Subject: Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch 
> Authors]
> To: "Jonathan Wilkes" 
> Cc: "Frank Barknecht" , pd-list@iem.at, "João Pais" 
> 
> Date: Saturday, September 5, 2009, 8:59 PM
> On Fri, 4 Sep 2009, Jonathan Wilkes
> wrote:
> 
> > But then what to do about multiple help patch
> authors?
> 
> I don't know, but if a template doesn't allow to say
> something complex like «Pierre wrote the reference section,
> Jean wrote the theoretical background section, Jacques
> provided the examples» then that template should be just
> shot dead.

You have included commas in your example.  But I thought you were 
arguing that commas will cause problems down the line.

-Jonathan

> 
>  _ _ __ ___ _  _
> _ ...
> | Mathieu Bouchard, Montréal, Québec. téléphone:
> +1.514.383.3801


  

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Mathieu Bouchard

On Fri, 4 Sep 2009, Jonathan Wilkes wrote:


But then what to do about multiple help patch authors?


I don't know, but if a template doesn't allow to say something complex 
like «Pierre wrote the reference section, Jean wrote the theoretical 
background section, Jacques provided the examples» then that template 
should be just shot dead.


 _ _ __ ___ _  _ _ ...
| Mathieu Bouchard, Montréal, Québec. téléphone: +1.514.383.3801___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-05 Thread Frank Barknecht
Hallo,
João Pais hat gesagt: // João Pais wrote:

> anyway, I downloaded a couple files, and saw no reference subatch. I  
> downloaded rj_accum-help.pd and rj_senergy~-help.pd, and they were normal 
> help patches. Which ones did you meant?

I was talking about the "rj" library of abstractions, which you can get here:
http://trac.rjdj.me/browser/trunk/rjlib/rj

or: 
  svn co http://svn.rjdj.me/scenes/trunk/rjlib/

I'm currenty converting the help files in alphabetical order, so things like
a_onset, e_alias or e_bitcrusher are already in the new format.

> And btw, one thing I would do is to download the rjdj library to study. 
> is there a place where the whole pack of files can be downloaded?

For that just grab the RjDj Composer pack: 
http://trac.rjdj.me/wiki/ComposersPack

Ciao
-- 
Frank

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-04 Thread Jonathan Wilkes
There isn't any explanation, so let's make one.  On pd-ext, have a 
look at the help file for [float], there should be a pd META subpatch 
there with all tags except the ones I proposed to add:
AUTHOR
HELP_PATCH_AUTHOR

Most tags are self-explanatory; the only one that isn't is GENRE.  I 
think we should keep it and link it up with the work being done 
on the "Properties" object-menu.  So any object can be part of several 
GENREs.  So the question is, what are the possible categories for GENRE?

The other unresolved issue is whether there can be commas separating 
the different keywords.  Since I'm not sure exactly how the META data 
will be used, I guess we should keep it as simple as possible and 
not have commas for now.

But then what to do about multiple help patch authors?

-Jonathan

--- On Fri, 9/4/09, João Pais  wrote:

> From: João Pais 
> Subject: Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch 
> Authors]
> To: "Frank Barknecht" , pd-list@iem.at
> Date: Friday, September 4, 2009, 2:05 PM
> >> Is there any place where
> it's explained what are the meta patches, and
> >> what should be in each patch? I never understood
> the meaning of them, and
> >> therefore, have never made any.
> > 
> > I could find any clear docs about that myself. I can
> however send you the
> 
> I could try to do it myself as well. but as I suspected
> that I would find only an outdated wiki page with some loose
> information on it, I found it better to ask who really knows
> about it.
> 
> 
> > specification for "pd REFERENCE" supbatches in RjDj
> help files. Note that this
> > is not dealing with META information like lices,
> author, etc. Think of it as an
> > alternative implementation of the main pddp help patch
> area into a parsable
> > format. The RjDj helpfiles can be got here: http://trac.rjdj.me ("Browse 
> > source")
> 
> thanks, but that doesn't really answer my question, it just
> hints at it.
> 
> anyway, I downloaded a couple files, and saw no reference
> subatch. I downloaded rj_accum-help.pd and
> rj_senergy~-help.pd, and they were normal help patches.
> Which ones did you meant?
> 
> And btw, one thing I would do is to download the rjdj
> library to study. is there a place where the whole pack of
> files can be downloaded?
> 
> ___
> Pd-list@iem.at
> mailing list
> UNSUBSCRIBE and account-management -> 
> http://lists.puredata.info/listinfo/pd-list
> 


  

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


Re: [PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-04 Thread João Pais

Is there any place where it's explained what are the meta patches, and
what should be in each patch? I never understood the meaning of them,  
and

therefore, have never made any.


I could find any clear docs about that myself. I can however send you the


I could try to do it myself as well. but as I suspected that I would find  
only an outdated wiki page with some loose information on it, I found it  
better to ask who really knows about it.



specification for "pd REFERENCE" supbatches in RjDj help files. Note  
that this
is not dealing with META information like lices, author, etc. Think of  
it as an
alternative implementation of the main pddp help patch area into a  
parsable
format. The RjDj helpfiles can be got here: http://trac.rjdj.me ("Browse  
source")


thanks, but that doesn't really answer my question, it just hints at it.

anyway, I downloaded a couple files, and saw no reference subatch. I  
downloaded rj_accum-help.pd and rj_senergy~-help.pd, and they were normal  
help patches. Which ones did you meant?


And btw, one thing I would do is to download the rjdj library to study. is  
there a place where the whole pack of files can be downloaded?


___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list


[PD] [pd REFERENCE] format [was: Re: Pd META: Author/Help Patch Authors]

2009-09-03 Thread Frank Barknecht
Hallo,
João Pais hat gesagt: // João Pais wrote:

> Is there any place where it's explained what are the meta patches, and  
> what should be in each patch? I never understood the meaning of them, and 
> therefore, have never made any.

I could find any clear docs about that myself. I can however send you the
specification for "pd REFERENCE" supbatches in RjDj help files. Note that this
is not dealing with META information like lices, author, etc. Think of it as an
alternative implementation of the main pddp help patch area into a parsable
format. The RjDj helpfiles can be got here: http://trac.rjdj.me ("Browse 
source")


Embedding parsable reference info into a subpatch
=

Each help file includes a subpatch called "REFERENCE" with graph on parent
(GOP) enabled.  The subpatch's GOP area is 450 px wide. The subpatch contains
only comments. 

The comments will generally consist of a colon-separated key:value pair. Also
allowed are "free" comments that don't use any of the following key names.
These will be appended to the "Description" field as separate paragraphs,
ordered by the value of their "y"-coordinate.  (We can think about adding them
to the previously encountered field instead.)

Required key names:

  Name: the name of the object
  Summary: short description of the object
  Description: Longer description of the object

Optional key names:

  Argument : argument x, where x is an integer index of the arg, zero-based.
  Inlet : message or audio inlet x, where x is an integer index of the 
inlet, zero-based.
  Outlet : message or audio outlet x, where x is an integer index of the 
outlet, zero-based.
  Tags: dash-separated list of key words (was "comma-separated" before)

Quick awk-oneliner to filter out the subpatch:

$ cat file-help.pd | awk 'BEGIN {RS=";"} /canvas .* REFERENCE/,/restore .* 
REFERENCE/  { for (i=4; i<=NF; i++) printf("%s ", $i); print ""}'  | grep -v 
REFE | sort

Ciao
-- 
Frank

___
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> 
http://lists.puredata.info/listinfo/pd-list