sbml-required — The SBML L3 Required Elements package discussion list

Re: [sbml-required] API change question

[Lucian S. <luc...@gm...>]

The original (short!) proposal is at:

http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Required_Elements_Proposal

And indeed, those two attributes cover the same semantic area, and in fact
'true' and 'false' still mean the same thing:  'core has alternate math'
meant that if you ignored the package math, you still got a workable
mathematic solution.  'viable without change' means the same thing, but no
longer implies that core is the only place the alternate might live, simply
that if the element in question was removed, what was left would be viable
(though perhaps different).

-Lucian


On Fri, Aug 15, 2014 at 11:20 AM, Weatherby,Gerard <gwe...@uc...>
wrote:

> What's the semantic meaning of conversion of the bools?
> Is old setCoreHasAlternateMath(true) = setViableWithoutChange(false) -- or
> are the two bools modeling different things.? (I've found the documentation
> for the current draft at
> http://sbml.org/Documents/Specifications#SBML_Level_3_Packages, but can't
> seem to find a description of the previous one).
>
> > -----Original Message-----
> > From: Sarah Keating [mailto:ske...@ca...]
> > Sent: Friday, August 15, 2014 2:12 PM
> > To: The SBML L3 Required Elements package discussion list
> > Subject: Re: [sbml-required] API change question
> >
> > Hi Gerard
> >
> > The function you mention is not (or indeed was not) part of the libsbml
> API so
> > I'm not sure how to answer your question.
> >
> >
> >
> > If you are working with the required elements package code you will have
> > noticed it changed dramatically. Apologies if you have worked out the
> > following but just for anyone else who sees this in the list.
> >
> > Originally any SBase object could have a plugin that gave you functions
> >
> > get/setCoreHasAlternateMath - boolean indicating alternate math is
> present
> >
> > get/setMathOverridden - name of the package that provides alternate math
> >
> > Now any SBase object can have a child element ListOfChangedMaths which
> > contains children of a new class ChangedMath.
> >
> > The changedMath objects now have functions
> >
> > get/setViableWithoutChange boolean indicating that the alternate math
> > can be ignored and a model will still have some meaning
> > This is essentially the same as setCoreHasAlternateMath.
> >
> > get/setChangedBy - uri of the package that provides alternate math
> > Again essentially the same as setMathOverridden.
> >
> >
> > Hopefully this helps. Let me know if I can help further.
> >
> > Sarah
> >
> >
> >
> >
> ------------------------------------------------------------------------------
> > _______________________________________________
> > sbml-required mailing list
> > sbm...@li...
> > https://lists.sourceforge.net/lists/listinfo/sbml-required
>
>
> ------------------------------------------------------------------------------
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required
>

Re: [sbml-required] API change question

[Weatherby,Gerard <gwe...@uc...>]

What's the semantic meaning of conversion of the bools?
Is old setCoreHasAlternateMath(true) = setViableWithoutChange(false) -- or are the two bools modeling different things.? (I've found the documentation for the current draft at http://sbml.org/Documents/Specifications#SBML_Level_3_Packages, but can't seem to find a description of the previous one).

> -----Original Message-----
> From: Sarah Keating [mailto:ske...@ca...]
> Sent: Friday, August 15, 2014 2:12 PM
> To: The SBML L3 Required Elements package discussion list
> Subject: Re: [sbml-required] API change question
> 
> Hi Gerard
> 
> The function you mention is not (or indeed was not) part of the libsbml API so
> I'm not sure how to answer your question.
> 
> 
> 
> If you are working with the required elements package code you will have
> noticed it changed dramatically. Apologies if you have worked out the
> following but just for anyone else who sees this in the list.
> 
> Originally any SBase object could have a plugin that gave you functions
> 
> get/setCoreHasAlternateMath - boolean indicating alternate math is present
> 
> get/setMathOverridden - name of the package that provides alternate math
> 
> Now any SBase object can have a child element ListOfChangedMaths which
> contains children of a new class ChangedMath.
> 
> The changedMath objects now have functions
> 
> get/setViableWithoutChange boolean indicating that the alternate math
> can be ignored and a model will still have some meaning
> This is essentially the same as setCoreHasAlternateMath.
> 
> get/setChangedBy - uri of the package that provides alternate math
> Again essentially the same as setMathOverridden.
> 
> 
> Hopefully this helps. Let me know if I can help further.
> 
> Sarah
> 
> 
> 
> ------------------------------------------------------------------------------
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required

Re: [sbml-required] API change question

[Sarah K. <ske...@ca...>]

Hi Gerard

The function you mention is not (or indeed was not) part of the libsbml 
API so I'm not sure how to answer your question.



If you are working with the required elements package code you will have 
noticed it changed dramatically. Apologies if you have worked out the 
following but just for anyone else who sees this in the list.

Originally any SBase object could have a plugin that gave you functions

get/setCoreHasAlternateMath - boolean indicating alternate math is present

get/setMathOverridden - name of the package that provides alternate math

Now any SBase object can have a child element ListOfChangedMaths which 
contains children of a new class ChangedMath.

The changedMath objects now have functions

get/setViableWithoutChange boolean indicating that the alternate math 
can be ignored and a model will still have some meaning
This is essentially the same as setCoreHasAlternateMath.

get/setChangedBy - uri of the package that provides alternate math
Again essentially the same as setMathOverridden.


Hopefully this helps. Let me know if I can help further.

Sarah

[sbml-required] API change question

[Weatherby,Gerard <gwe...@uc...>]

I'm upgrading Virtual Cell to use the current (5.10.2) version of SBML.

I'm inferring / guessing that the prior method setAlternateCoreHasAlternateFront is replaced with setViableWithoutChange?

Gerard Weatherby
Appllication Architect
Virtual Cell, Center for Cell Analysis and Modeling (CCAM)
University of Connecticut Health Center
860-679-2065

[sbml-required] New required elements package code

[Sarah K. <ske...@ca...>]

Hi People who might be using the requiredElements package

As you will be aware the specification for this package has changed 
dramatically since it's first inception. Up until now libSBML has 
implemented support for the old specification but the next release of 
libSBML will update to support the new version.

** This will break people's code **

Whilst looking at the libsbml code I noticed that there were some file 
name inconsistencies; this presents me with a dilemma. I have two options.

Option 1: I change the filenames to be consistent. This will be a very 
obvious break in existing code; and so easy for developers to notice !

Option 2: I leave the names as-is. It would then be seamless for a 
developer to extract the req code archive for either the old 
specification (libsbml-5.10.0) or that latest specification (will be 
libsbml-5.10.2) and create a libsbml build with whichever specification 
they are currently working with. With Option 1 this would involve 
possibly needing to remove some files.

Whilst it is best that developers update to the latest versions of 
specifications, we acknowledge that this can take time. So there are 
advantages/disadvantages to either option.

I'd be grateful for any opinions.

Thanks

Sarah

[sbml-required] libSBML update to spatial/req code

[Sarah K. <ske...@ca...>]

Dear members of the 'spatial' and 'req' mailing lists,

Earlier this year we said that we would update the spatial/req code in 
libSBML to be in line with the latest version of the specifications, and 
provide the update in our April release of libSBML.

However, Frank, (while looking at implementing spatial simulations that 
use the 'spatial' specification) has raised some issues with the latest 
version of the specification draft version 0.84. His posting about these 
issues to the mailing list in January 30 remain unresolved (and indeed, 
Frank received no feedback at all). This makes it difficult to gauge the 
status of people's implementations of software with respect to that 
specification.

We do not wish to hamper potential implementation work and exchange of 
models by updating libSBML to use a version of the specification that no 
one is yet using, or indeed one that may change further. Consequently, 
we have decided that we should wait the resolution of the issues before 
continuing further with implementation.  Given the dependency of the 
existing 'spatial' code on the Required Elements ('req') package, this 
means we will also not update the 'req' code on this occasion.

It would be really useful for our planning if people could tell us:

  1. Are you actively developing support for spatial and/or req using 
libSBML?

and if so,

  2. Are you waiting for libSBML to be updated in line with the last 
version (0.84, Nov. 2013) of the specification?

We apologise for the change in plans, and hope we have not 
inconvenienced anyone.

The libSBML Team

[sbml-required] Updates to spatial and req

[Sarah K. <ske...@ca...>]

There has been significant work on the specifications of both the
Required Elements (req) and the Spatial Processes (spatial) packages
which are not yet reflected in libSBML.

We have not updated code as there are changes that may have a major
impact on implementations already using the code for these packages.
Significant changes are

1. 'spatial' no longer depends on the 'req' package

2. the 'req' package has undergone major revision

3. the latest 'spatial' proposal contains some structural changes and is 
currently under discussion; thus it may change again further

However, libSBML does need to update to reflect the latest versions of
the specifications. Referring to the proposed release schedule for
libSBML in 2014 (http://sbml.org/SBML_Projects/libSBML/Release_schedule)
   we would anticipate updating the code in the next libSBML full release
(April 2014). At this point we will update 'req'  (which appears to be
stable) and will update 'spatial' with what has been agreed upon at that
point.

We are aware that this change may break user's code and may also hinder
the ex-changeability of models between tools currently using spatial
and/or req; as models now considered valid may no longer be considered
valid when the new specifications are applied.

Obviously this presents us with a chicken and egg problem. The changing
specifications need to be reflected in libSBML and we hope that users of
the spatial and/or req specifications will update their software to the
latest versions of the specification. However we do acknowledge that
this takes time and resources.

Hopefully this can be considered enough notice that developers can be
aware that libsbml-5.10 will make these changes and we will not
seriously hinder progress for people working in these areas. If this is
not the case, please let us know so we can discuss a schedule that meets
everyone's requirements.

Thanks for all your support.

The libSBML Team

Re: [sbml-required] Release 0.5

[Michael H. <mh...@ca...>]

Lucian,

Excellent; thanks for producing this.

MH

Re: [sbml-required] Release 0.5

[Chris J. M. <my...@ec...>]

Ok, I will look for it and implement support when released with libsbml.

Chris

On May 31, 2013, at 11:43 AM, Lucian Smith <lp...@sp...> wrote:

> Basic libsbml support should not be a problem, with Sarah's code 
> generator; it will be a little more work to generate something that runs 
> through a model and automatically adds req elements to package-modified 
> elements, but even that won't be *too* hard, and until then, you could 
> add them yourself.
> 
> I bet a function to gather a list of namespaces from req elements also 
> would not be difficult to implement.
> 
> -Lucian
> 
> * Chris J. Myers <my...@ec...> [2013-05-31 18:40] writes:
>> We could implement if there is libsbml support.  Basically, it would be nice to have a validation function that takes a list of the namespaces that ones tool supports and returns the objects that are changed by other names spaces as warnings when viable is true and errors otherwise.  As for creation, is there support in libsbml to create these elements yet, or is that also still to come?
>> 
>> Chris
>> 
>> On May 30, 2013, at 3:40 PM, Lucian Smith <lp...@sp...> wrote:
>> 
>>> Following some discussions at HARMONY, here's a new version of the 
>>> 'Required Elements' package:
>>> 
>>> https://docs.google.com/file/d/0B71r5kLs3FliWGxFN3o0dnNrZE0/edit
>>> 
>>> The main change here is the ability to reference custom annotations with 
>>> the 'ChangedMath' elements as well as packages, and the addition of some 
>>> validation rules.
>>> 
>>> At this point, I think we can probably muster up enough tools to 
>>> *produce* these files; the trick is going to be finding a client that 
>>> wants to *use* the package.  In other words, finding someone for whom 
>>> this package is actually useful, that wants to be able to be more 
>>> informed about packages and annotations they don't know about.
>>> 
>>> Any takers?  Should I troll for responses on sbml-discuss?
>>> 
>>> (Any comments on this version of the spec are also welcome, of course.)
>>> 
>>> -Lucian
>>> 
>>> ------------------------------------------------------------------------------
>>> Introducing AppDynamics Lite, a free troubleshooting tool for Java/.NET
>>> Get 100% visibility into your production application - at no cost.
>>> Code-level diagnostics for performance bottlenecks with <2% overhead
>>> Download for free and get started troubleshooting in minutes.
>>> http://p.sf.net/sfu/appdyn_d2d_ap1
>>> _______________________________________________
>>> sbml-required mailing list
>>> sbm...@li...
>>> https://lists.sourceforge.net/lists/listinfo/sbml-required
>> 
>> 
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite
>> It's a free troubleshooting tool designed for production
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://p.sf.net/sfu/appdyn_d2d_ap2
>> _______________________________________________
>> sbml-required mailing list
>> sbm...@li...
>> https://lists.sourceforge.net/lists/listinfo/sbml-required
> 
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite
> It's a free troubleshooting tool designed for production
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://p.sf.net/sfu/appdyn_d2d_ap2
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required

Re: [sbml-required] Release 0.5

[Lucian S. <lp...@sp...>]

Basic libsbml support should not be a problem, with Sarah's code 
generator; it will be a little more work to generate something that runs 
through a model and automatically adds req elements to package-modified 
elements, but even that won't be *too* hard, and until then, you could 
add them yourself.

I bet a function to gather a list of namespaces from req elements also 
would not be difficult to implement.

-Lucian

* Chris J. Myers <my...@ec...> [2013-05-31 18:40] writes:
> We could implement if there is libsbml support.  Basically, it would be nice to have a validation function that takes a list of the namespaces that ones tool supports and returns the objects that are changed by other names spaces as warnings when viable is true and errors otherwise.  As for creation, is there support in libsbml to create these elements yet, or is that also still to come?
> 
> Chris
> 
> On May 30, 2013, at 3:40 PM, Lucian Smith <lp...@sp...> wrote:
> 
> > Following some discussions at HARMONY, here's a new version of the 
> > 'Required Elements' package:
> > 
> > https://docs.google.com/file/d/0B71r5kLs3FliWGxFN3o0dnNrZE0/edit
> > 
> > The main change here is the ability to reference custom annotations with 
> > the 'ChangedMath' elements as well as packages, and the addition of some 
> > validation rules.
> > 
> > At this point, I think we can probably muster up enough tools to 
> > *produce* these files; the trick is going to be finding a client that 
> > wants to *use* the package.  In other words, finding someone for whom 
> > this package is actually useful, that wants to be able to be more 
> > informed about packages and annotations they don't know about.
> > 
> > Any takers?  Should I troll for responses on sbml-discuss?
> > 
> > (Any comments on this version of the spec are also welcome, of course.)
> > 
> > -Lucian
> > 
> > ------------------------------------------------------------------------------
> > Introducing AppDynamics Lite, a free troubleshooting tool for Java/.NET
> > Get 100% visibility into your production application - at no cost.
> > Code-level diagnostics for performance bottlenecks with <2% overhead
> > Download for free and get started troubleshooting in minutes.
> > http://p.sf.net/sfu/appdyn_d2d_ap1
> > _______________________________________________
> > sbml-required mailing list
> > sbm...@li...
> > https://lists.sourceforge.net/lists/listinfo/sbml-required
> 
> 
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite
> It's a free troubleshooting tool designed for production
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://p.sf.net/sfu/appdyn_d2d_ap2
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required

Re: [sbml-required] Release 0.5

[Chris J. M. <my...@ec...>]

We could implement if there is libsbml support.  Basically, it would be nice to have a validation function that takes a list of the namespaces that ones tool supports and returns the objects that are changed by other names spaces as warnings when viable is true and errors otherwise.  As for creation, is there support in libsbml to create these elements yet, or is that also still to come?

Chris

On May 30, 2013, at 3:40 PM, Lucian Smith <lp...@sp...> wrote:

> Following some discussions at HARMONY, here's a new version of the 
> 'Required Elements' package:
> 
> https://docs.google.com/file/d/0B71r5kLs3FliWGxFN3o0dnNrZE0/edit
> 
> The main change here is the ability to reference custom annotations with 
> the 'ChangedMath' elements as well as packages, and the addition of some 
> validation rules.
> 
> At this point, I think we can probably muster up enough tools to 
> *produce* these files; the trick is going to be finding a client that 
> wants to *use* the package.  In other words, finding someone for whom 
> this package is actually useful, that wants to be able to be more 
> informed about packages and annotations they don't know about.
> 
> Any takers?  Should I troll for responses on sbml-discuss?
> 
> (Any comments on this version of the spec are also welcome, of course.)
> 
> -Lucian
> 
> ------------------------------------------------------------------------------
> Introducing AppDynamics Lite, a free troubleshooting tool for Java/.NET
> Get 100% visibility into your production application - at no cost.
> Code-level diagnostics for performance bottlenecks with <2% overhead
> Download for free and get started troubleshooting in minutes.
> http://p.sf.net/sfu/appdyn_d2d_ap1
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required

[sbml-required] Release 0.5

[Lucian S. <lp...@sp...>]

Following some discussions at HARMONY, here's a new version of the 
'Required Elements' package:

https://docs.google.com/file/d/0B71r5kLs3FliWGxFN3o0dnNrZE0/edit

The main change here is the ability to reference custom annotations with 
the 'ChangedMath' elements as well as packages, and the addition of some 
validation rules.

At this point, I think we can probably muster up enough tools to 
*produce* these files; the trick is going to be finding a client that 
wants to *use* the package.  In other words, finding someone for whom 
this package is actually useful, that wants to be able to be more 
informed about packages and annotations they don't know about.

Any takers?  Should I troll for responses on sbml-discuss?

(Any comments on this version of the spec are also welcome, of course.)

-Lucian

Re: [sbml-required] Latest draft

[Michael H. <mh...@ca...>]

On Fri, 17 May 2013 18:15:47 +0100, Lucian Smith wrote:
> 
> https://docs.google.com/file/d/0B71r5kLs3FliNkJaNDdXVkNfeGs/edit
> 
> I have yet again renamed 'MathChanged'; this time to 'ChangedMath', with 
> the vague hope that someone else will think that this is OK, and not 
> think that I am kidding.

:-).  This one seems okay IMHO.

MH

[sbml-required] Latest draft

[Lucian S. <lp...@sp...>]

https://docs.google.com/file/d/0B71r5kLs3FliNkJaNDdXVkNfeGs/edit

I have yet again renamed 'MathChanged'; this time to 'ChangedMath', with 
the vague hope that someone else will think that this is OK, and not 
think that I am kidding.

Also, other bits have been reworded, particularly to address some of 
Sarah's questions/concerns about the previous draft.

See you at HARMONY!

-Lucian

Re: [sbml-required] New version

[Sarah K. <ske...@ca...>]

I have added some comments - although I could not work out how to make 
the comment relative to the text it related to :-)

Sarah

On 26/04/2013 22:52, Lucian Smith wrote:
> No new validation rules yet, but everything else has been overhauled to
> use the new system, and there is extensive discussion about how
> different packages may interact with this package:
>
> https://docs.google.com/file/d/0B71r5kLs3FliVDZDR2xlLUJOWUk/edit
>
> -Lucian
>
> ------------------------------------------------------------------------------
> Try New Relic Now&  We'll Send You this Cool Shirt
> New Relic is the only SaaS-based application performance monitoring service
> that delivers powerful full stack analytics. Optimize and monitor your
> browser, app,&  servers with just a few lines of code. Try New Relic
> and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required
>
>

[sbml-required] New version

[Lucian S. <lp...@sp...>]

No new validation rules yet, but everything else has been overhauled to 
use the new system, and there is extensive discussion about how 
different packages may interact with this package:

https://docs.google.com/file/d/0B71r5kLs3FliVDZDR2xlLUJOWUk/edit

-Lucian

Re: [sbml-required] Updates

[Michael H. <mh...@ca...>]

I'm so used to assuming that these things are listOf's that I didn't notice it wasn't so :-).  Yes, I agree this should be a listOf, for consistency.

MH

On Fri, 26 Apr 2013 14:58:34 +0100, Sarah Keating wrote:
> Firstly - yes this is a great way to go :-)
> 
> BUT
> 
> You can guess what I am going to say -
> 
> where is my listOfMathChanged element :-)
> 
> Up to this point SBML has very clearly always had a containing 
> listOfXYZ element that acts as a parent to the children in the list. 
> Here (and also in current groups discussion) we are starting to 
> change this policy. Do we need to actually review that as an overall 
> strategy ?
> 
> Programmatically it is extremely useful having a base ListOf class 
> from which these listOf elements are derived. Obviously the class 
> derived from a ListOf does not *have* to be called listOfXYZ.  
> However the situation here is going to be slightly more difficult - 
> since there is actually no intervening container between the SBase 
> object and the children (that might or might not be in a listOf).
> 
> Yes I know there can be a dummy parent object but users of SBML in OO 
> situations (yes libSBML but also JSBML or any other implementation) 
> are used to having a correspondence between classes in code and 
> elements in SBML. I have to say I find the comp libsbml code a little 
> confusing because it does introduce some classes that are not 
> reflected in the spec !
> 
> Also it does not hurt (it just makes the xml bigger!) to have a 
> listOfMathChanged and it clearly indicates to people that there can 
> be more than one. We have no other situation in SBML when an element 
> can contain more than one child of the same type; where there is not 
> a parent listOfXYZ element that contains these.
> 
> Sarah
> 
> PS Yes I know "listOfMathChanged" does not quite work as a name :-)

Re: [sbml-required] Updates

[Lucian S. <lp...@sp...>]

It's certainly possible to introduce a ListOfMathChangeds, obviously the 
progression went 'hmm, maybe we should use more than one?  Yeah, that 
makes sense, I'll edit the UML,' and I just changed the '0,1' to '0..*'.

However, if we're going to say that this is The Way SBML Does Things, we 
should really say that in the package development instructions.  Do we 
want to say that?

I guess this is is general editors question.

(I also just created a ticket for creating a ListOf class in l3v2, too.)

-Lucian

* Sarah Keating <ske...@ca...> [2013-04-26 15:00] writes:
> Firstly - yes this is a great way to go :-)
> 
> BUT
> 
> You can guess what I am going to say -
> 
> where is my listOfMathChanged element :-)
> 
> Up to this point SBML has very clearly always had a containing listOfXYZ 
> element that acts as a parent to the children in the list. Here (and 
> also in current groups discussion) we are starting to change this 
> policy. Do we need to actually review that as an overall strategy ?
> 
> Programmatically it is extremely useful having a base ListOf class from 
> which these listOf elements are derived. Obviously the class derived 
> from a ListOf does not *have* to be called listOfXYZ.  However the 
> situation here is going to be slightly more difficult - since there is 
> actually no intervening container between the SBase object and the 
> children (that might or might not be in a listOf).
> 
> Yes I know there can be a dummy parent object but users of SBML in OO 
> situations (yes libSBML but also JSBML or any other implementation) are 
> used to having a correspondence between classes in code and elements in 
> SBML. I have to say I find the comp libsbml code a little confusing 
> because it does introduce some classes that are not reflected in the spec !
> 
> Also it does not hurt (it just makes the xml bigger!) to have a 
> listOfMathChanged and it clearly indicates to people that there can be 
> more than one. We have no other situation in SBML when an element can 
> contain more than one child of the same type; where there is not a 
> parent listOfXYZ element that contains these.
> 
> Sarah
> 
> PS Yes I know "listOfMathChanged" does not quite work as a name :-)
> 
> ------------------------------------------------------------------------------
> Try New Relic Now & We'll Send You this Cool Shirt
> New Relic is the only SaaS-based application performance monitoring service 
> that delivers powerful full stack analytics. Optimize and monitor your
> browser, app, & servers with just a few lines of code. Try New Relic
> and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required

Re: [sbml-required] Updates

[Sarah K. <ske...@ca...>]

Firstly - yes this is a great way to go :-)

BUT

You can guess what I am going to say -

where is my listOfMathChanged element :-)

Up to this point SBML has very clearly always had a containing listOfXYZ 
element that acts as a parent to the children in the list. Here (and 
also in current groups discussion) we are starting to change this 
policy. Do we need to actually review that as an overall strategy ?

Programmatically it is extremely useful having a base ListOf class from 
which these listOf elements are derived. Obviously the class derived 
from a ListOf does not *have* to be called listOfXYZ.  However the 
situation here is going to be slightly more difficult - since there is 
actually no intervening container between the SBase object and the 
children (that might or might not be in a listOf).

Yes I know there can be a dummy parent object but users of SBML in OO 
situations (yes libSBML but also JSBML or any other implementation) are 
used to having a correspondence between classes in code and elements in 
SBML. I have to say I find the comp libsbml code a little confusing 
because it does introduce some classes that are not reflected in the spec !

Also it does not hurt (it just makes the xml bigger!) to have a 
listOfMathChanged and it clearly indicates to people that there can be 
more than one. We have no other situation in SBML when an element can 
contain more than one child of the same type; where there is not a 
parent listOfXYZ element that contains these.

Sarah

PS Yes I know "listOfMathChanged" does not quite work as a name :-)

Re: [sbml-required] Updates

[Michael H. <mh...@ca...>]

I like the changes!  The list idea in particular addresses a limitation of the previous scheme.

MH

On Thu, 25 Apr 2013 22:10:45 +0100, Lucian Smith wrote:
> I'm working on updating 'req' to use the new 'mathChanged' child, and I 
> have a few other changes to propose, all visible at
> 
> https://docs.google.com/drawings/d/1tmE9ZorGlMlqnAkoln65ilIHcAkDLCKv5pfYl8TwUJ4
> 
> * That we allow more than one 'MathChanged' child, for situations where 
> the math of an element is affected by more than one package (i.e. 
> Spatial and Comp; or Distrib and Arrays)
> 
> * A change in the attribute names:
> 
> 'mathOverridden' -> 'changedBy' (the fact that the math has changed is 
> now encapsulated in the object name)
> 
> 'coreHasAlternativeMath' -> 'viableWithoutChange'  (when multiple 
> packages affect the same element, it's not core that needs to be 
> checked; it's 'whatever is there minus this package' that needs to be 
> checked.)
> 
> Any opinions?
> 
> -Lucian
> 
> ------------------------------------------------------------------------------
> Try New Relic Now & We'll Send You this Cool Shirt
> New Relic is the only SaaS-based application performance monitoring service 
> that delivers powerful full stack analytics. Optimize and monitor your
> browser, app, & servers with just a few lines of code. Try New Relic
> and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required

[sbml-required] Updates

[Lucian S. <lp...@sp...>]

I'm working on updating 'req' to use the new 'mathChanged' child, and I 
have a few other changes to propose, all visible at

https://docs.google.com/drawings/d/1tmE9ZorGlMlqnAkoln65ilIHcAkDLCKv5pfYl8TwUJ4

* That we allow more than one 'MathChanged' child, for situations where 
the math of an element is affected by more than one package (i.e. 
Spatial and Comp; or Distrib and Arrays)

* A change in the attribute names:

'mathOverridden' -> 'changedBy' (the fact that the math has changed is 
now encapsulated in the object name)

'coreHasAlternativeMath' -> 'viableWithoutChange'  (when multiple 
packages affect the same element, it's not core that needs to be 
checked; it's 'whatever is there minus this package' that needs to be 
checked.)

Any opinions?

-Lucian

Re: [sbml-required] The 'Required Elements' package

[Michael H. <mh...@ca...>]

Hi Anu,

Thanks for the note. This is encouraging.

MH

On Fri, 19 Apr 2013 11:15:22 -0400, Lakshminarayana,Anuradha wrote:
> 
> Mike,
> 
> Speaking for the spatial package that uses the requiredElements 
> package, I don't believe there are a lot of models out there. 
> Presently I'm aware of only those that are exported from/imported 
> into Virtual Cell. I'm not saying that there aren't any others, I'm 
> just not aware of them. Nobody has brought up issues with 
> spatial/requiredElements, perhaps not many people have tried it out? 
> I'm not sure if any other package uses requiredElements currently.
> 
> As far as implementation goes, if the change is made to the 
> requiredElements package (subElement instead of 2 attributes on 
> SBase), modifying the software to adjust to the new way should not be 
> difficult.
> 
> Hope this helps,
> ~Anuradha.
> 
> 
> -----Original Message-----
> From: Michael Hucka [mailto:mh...@ca...] 
> Sent: Thursday, April 18, 2013 8:32 PM
> To: The SBML L3 Required Elements package discussion list; 
> Moraru,Ion; Lakshminarayana,Anuradha
> Subject: Re: [sbml-required] The 'Required Elements' package
> 
> Also, it's hoped that the modified scheme (the one with the element) 
> would be sufficiently similar to the previous 2-attribute thing that 
> it wouldn't be too hard to modify software to write it in the new way.
> 
> (However, it's true that if there are a lot of models out there using 
> the previous form, then it becomes more work.  Are there a lot of 
> models that use the required elements attributes?)
> 
> MH

Re: [sbml-required] The 'Required Elements' package

[Lakshminarayana,Anuradha <an...@uc...>]

Mike,

Speaking for the spatial package that uses the requiredElements package, I don't believe there are a lot of models out there. Presently I'm aware of only those that are exported from/imported into Virtual Cell. I'm not saying that there aren't any others, I'm just not aware of them. Nobody has brought up issues with spatial/requiredElements, perhaps not many people have tried it out? I'm not sure if any other package uses requiredElements currently.

As far as implementation goes, if the change is made to the requiredElements package (subElement instead of 2 attributes on SBase), modifying the software to adjust to the new way should not be difficult.

Hope this helps,
~Anuradha.


-----Original Message-----
From: Michael Hucka [mailto:mh...@ca...] 
Sent: Thursday, April 18, 2013 8:32 PM
To: The SBML L3 Required Elements package discussion list; Moraru,Ion; Lakshminarayana,Anuradha
Subject: Re: [sbml-required] The 'Required Elements' package

Also, it's hoped that the modified scheme (the one with the element) would be sufficiently similar to the previous 2-attribute thing that it wouldn't be too hard to modify software to write it in the new way.

(However, it's true that if there are a lot of models out there using the previous form, then it becomes more work.  Are there a lot of models that use the required elements attributes?)

MH

Re: [sbml-required] The 'Required Elements' package

[Michael H. <mh...@ca...>]

Also, it's hoped that the modified scheme (the one with the element) would be sufficiently similar to the previous 2-attribute thing that it wouldn't be too hard to modify software to write it in the new way.

(However, it's true that if there are a lot of models out there using the previous form, then it becomes more work.  Are there a lot of models that use the required elements attributes?)

MH

Re: [sbml-required] The 'Required Elements' package

[Lucian S. <lp...@sp...>]

The motivation for moving it to an element was not to make it easier to 
ignore, but to resolve an issue with defaults.  SBML Level 3 tried very 
hard to make sure that no attribute had a default, and took particular 
care with Booleans to make sure no Boolean attribute was optional at all, 
because people often assume that the absence of a bookean means 'false' 
instead of 'undefined'.

This conflicts with the 'coreHasAlternativeMath' attribute, since it was 
boolean and optional, but to make it *not* optional, you'd have to put it 
on every single SBML element in the model, many of which don't have 
mathematical meaning in the first place!  One option is to say that 
both 'req' attributes are optional, but if one is defined, they both have 
to be defined.  And another option is to put both attributes on a new 
child object, and make them both required on the object.  This 
accomplishes the goal of not having defaults, but still having those 
attributes present everywhere that you need them.  As Mike points out, it 
would also allow annotation of the object itself, providing another 
semantic hook to people who were interested.

However, it's probably not a huge deal to not do it this way if it would 
represent a significant cost to you all to change it at this point--if it 
was going to be difficult to convert a ton of your models, or the like.  
Do you have a strong preference either way, or is either version workable?

As far as 'what do we have in core', all package have the ability to 
change the mathematical meaning of SBML symbols.  All that is required is 
to put 'requried=true' in the header.  What 'required elements' does is 
let you annotate "This *particular* symbol and/or its math was changed by 
this package".  If you don't care about that, you don't need to worry 
about the package.

-Lucian

* Schaff,Jim <Sc...@NE...> [2013-04-18 17:48] writes:
> Hi All,
> 
> Of course I'm not up to speed on this discussion, but I find that the Required Elements package delivered what it was designed for.  In this case, I'm coming at this from a group who "writes" these attributes rather than interpret them for possibly unknown packages (as of today).
> 
> It seems that moving these attributes to a separate element makes them slightly easier to ignore ... what was the motivation for this?
> 
> If we remove the required elements altogether, is there something proposed in "core" to provide for declaring package-defined mathematical symbols and package-defined semantics changes on existing SBML Core elements?
> 
> Thanks,
> 
> Jim.
> 
> 
> 
> -----Original Message-----
> From: Michael Hucka [mailto:mh...@ca...] 
> Sent: Wednesday, April 17, 2013 2:12 PM
> To: The SBML L3 Required Elements package discussion list
> Cc: Schaff,Jim; Moraru,Ion; Lakshminarayana,Anuradha
> Subject: Re: [sbml-required] The 'Required Elements' package
> 
> I like this idea.  It would also allow annotations to be put on the new element, possibly helping to convey additional information about the nature of the overriding.
> 
> MH
> 
> On Thu, 11 Apr 2013 20:10:37 +0100, Lucian Smith wrote:
> > So, last December, we finally had a vote on the 'Required Elements' 
> > package, and the response was mixed and low, but ultimately positive, 
> > officially greenlighting the package for full development:
> > 
> > http://sbml.org/Community/Wiki/SBML_Level_3_Proposals/Surveys/Level_3_
> > Package_Proposal_Voting_Results:_req
> > 
> > Before the list was created, we had a discussion on the sbml-editors 
> > list (which turns out to have all four people who were interested in 
> > joining the PWG:  Mike, Chris, Sarah, and I) about the package, and we 
> > had some ideas about changing it somewhat, but before we implement 
> > those changes, I really want to know how it would affect the Spatial package.
> > 
> > I also want to know how integral the required elements package is to 
> > the spatial package software--have you found it to be helpful?  Extremely so?
> > Moderately so?
> > 
> > The proposed change would be to move the two attributes from being 
> > direct members of SBase to being members of an optional new child of 
> > SBase.  So, changed from:
> > 
> > https://docs.google.com/drawings/d/181TH9YquQpSWvxEUtPQZshUqKE_kZ5dh26
> > JtQgYW9EA
> > 
> > to:
> > 
> > https://docs.google.com/drawings/d/1tmE9ZorGlMlqnAkoln65ilIHcAkDLCKv5p
> > fYl8TwUJ4
> > 
> > This would allow the package elements to still be optional, but not 
> > have an optional boolean attribute.
> > 
> > If the UI from SBML was set up to smooth the transition, would the new 
> > scheme work for you?
> > 
> > In general, if you have found the package helpful in the past and 
> > anticipate that it'll be helpful in the future, I am more than happy 
> > to help support getting this package out the door.  If at this point 
> > you
> > *don't* feel it's helpful, however, there's enough ambivalence from 
> > the rest of the community to put this on the shelf again until it's needed.
> > 
> > -Lucian
> > 
> > ----------------------------------------------------------------------
> > -------- Precog is a next-generation analytics platform capable of 
> > advanced analytics on semi-structured data. The platform includes APIs 
> > for building apps and a phenomenal toolset for data science. 
> > Developers can use our toolset for easy data analysis & visualization. 
> > Get a free account!
> > http://www2.precog.com/precogplatform/slashdotnewsletter
> > _______________________________________________
> > sbml-required mailing list
> > sbm...@li...
> > https://lists.sourceforge.net/lists/listinfo/sbml-required
> 
> ------------------------------------------------------------------------------
> Precog is a next-generation analytics platform capable of advanced
> analytics on semi-structured data. The platform includes APIs for building
> apps and a phenomenal toolset for data science. Developers can use
> our toolset for easy data analysis & visualization. Get a free account!
> http://www2.precog.com/precogplatform/slashdotnewsletter
> _______________________________________________
> sbml-required mailing list
> sbm...@li...
> https://lists.sourceforge.net/lists/listinfo/sbml-required