IDE and subprocess defs

[John Mettraux]

Hi Lukas,

I released a post showing how to use a [sub]process definition :
http://jmettraux.wordpress.com/2006/05/17/activity-based-costing-a-hint/

(I also posted about your work :
http://jmettraux.wordpress.com/2006/05/16/lukas-openwfe-ide/ )

How could the IDE manage those subprocesses [calls] ? If you turn off
XSD validation, you can directly use the name of a process-definition
as an expression (an  XML element).
How could your IDE cope with that ?


Best regards,

--
John Mettraux   -///-   http://jmettraux.openwfe.org


-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid0709&bid&3057&dat1642
_______________________________________________
OpenWFE - Open source WorkFlow Engine
OpenWFE-devel mailing list
OpenWFE-devel@...
https://lists.sourceforge.net/lists/listinfo/openwfe-devel

[Lukas Eder]

Hi John,

Thanks for documenting my work on your blog!

> I released a post showing how to use a [sub]process definition :
> http://jmettraux.wordpress.com/2006/05/17/activity-based-costing-a-hint/
>
> [...]
>
> How could the IDE manage those subprocesses [calls] ? If you
> turn off XSD validation, you can directly use the name of a
> process-definition as an expression (an  XML element).
> How could your IDE cope with that ?

As far as that syntax and semantics is concerned, the IDE could cope with
that if I hard-coded such behaviour in the code. I.e. discover all
process-definitions and accept their "name" attribute values as legal XML
element tags. But the same semantics is already covered by the <subprocess
/> tag
(http://www.openwfe.org/docbook/build/ch06s02.html#expression_subprocess),
in a much more expressive way:

instead of
<activity aname="activity-c" acost="100" />

you would then write
<subprocess ref="activity" aname="activity-c" acost="100" />

if I'm not all mistaken... Are you sure it is necessary to shortcut that
syntax? Is that already implemented?

Lukas



-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
OpenWFE - Open source WorkFlow Engine
OpenWFE-devel mailing list
OpenWFE-devel@...
https://lists.sourceforge.net/lists/listinfo/openwfe-devel

[John Mettraux]

On 5/18/06, Lukas Eder <lukas.eder@...> wrote:
> Hi John,
>
> Thanks for documenting my work on your blog!

That's part of the communication tasks :-)
Sorry for your personal google ranking...
http://google.com/trends?q=eder&ctab=0&geo=all&date=all

> > I released a post showing how to use a [sub]process definition :
> > http://jmettraux.wordpress.com/2006/05/17/activity-based-costing-a-hint/
> >
> > [...]
> >
> > How could the IDE manage those subprocesses [calls] ? If you
> > turn off XSD validation, you can directly use the name of a
> > process-definition as an expression (an  XML element).
> > How could your IDE cope with that ?
>
> As far as that syntax and semantics is concerned, the IDE could cope with
> that if I hard-coded such behaviour in the code. I.e. discover all
> process-definitions and accept their "name" attribute values as legal XML
> element tags. But the same semantics is already covered by the <subprocess
> /> tag
> (http://www.openwfe.org/docbook/build/ch06s02.html#expression_subprocess),
> in a much more expressive way:
>
> instead of
> <activity aname="activity-c" acost="100" />
>
> you would then write
> <subprocess ref="activity" aname="activity-c" acost="100" />
>
> if I'm not all mistaken... Are you sure it is necessary to shortcut that
> syntax? Is that already implemented?

Yes, this syntax is very important, as it allows to have lighter and
more readable processes, it's somehow like :

"aadsfads".toUpperCase();
vs
invoke("addsfads", "toUpperCase");


    <cursor>
        <activity aname="activity-a" acost="100" />
        <activity aname="activity-b" acost="200" />
        <activity aname="activity-c" acost="50" />
    </cursor>
vs
    <cursor>
        <subprocess ref="activity" aname="activity-a" acost="100" />
        <subprocess ref="activity" aname="activity-b" acost="200" />
        <subprocess ref="activity" aname="activity-c" acost="50" />
    </cursor>

IMO, it helps making the language feel more natural, especially if IT
people made available such subprocesses to the 'business designer'
folks so that they can express their processes in something a [little]
bit closer to natural language.

Of course, your IDE somehow hides this XML stuff...


There's no problem with me if your IDE only outputs 'subprocess' expressions.

But on the other hand could we have something like suggestion, when
the designer (the person) types in 'act...', the IDE proposes him with
'activity' (just asking, it's similar to a method call).

Concerning the output, we could maybe have a switch in the
preferences, enabling one or the other variant.


What do you think ?

--
John Mettraux   -///-   http://jmettraux.openwfe.org


-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid0709&bid&3057&dat1642
_______________________________________________
OpenWFE - Open source WorkFlow Engine
OpenWFE-devel mailing list
OpenWFE-devel@...
https://lists.sourceforge.net/lists/listinfo/openwfe-devel

[Lukas Eder]

John,

> IMO, it helps making the language feel more natural,
> especially if IT people made available such subprocesses to
> the 'business designer'
> folks so that they can express their processes in something a
> [little] bit closer to natural language.

Of course, I can see that. One problem is that there are quite a few
reserved words and IT people as well as 'business designers' need to be
aware of that. Furthermore, new expressions may break old process
definitions. That may lead to problems that are hard to track. But I'm sure
this is documented and users are in fact aware of the situation.

I was just worrying about that "synonymous syntax" for the same semantics,
because they do these things a lot in PHP to "facilitate" coding. But in the
end, it becomes a great source of bugs. One example are the PHP magic
quotes. You can configure your server to automatically escape all parameters
("'" becomes "\'" etc...) passed by GET or POST in order to prevent SQL
injection and the likes. Now, this being a server configuration you never
know whether or not those quotes were added... You get the point...

> But on the other hand could we have something like
> suggestion, when the designer (the person) types in 'act...',
> the IDE proposes him with 'activity' (just asking, it's
> similar to a method call).

As I said, I can hard-code all kinds of additional behaviour on top of the
XML schema definition, as the IDE does not access the schema directly but
through a sort of proxy. There is a very simple model acting as that proxy,
defining which kinds of child nodes are acceptable for each node. It would
be easy to enhance that model. That way, the difference between a reserved
word like <sequence /> or a user-defined word like <activity /> would be
completely transparent to the user.

However, the attributes in
<activity aname="activity-c" acost="50" />

cannot be discovered automatically as they may have arbitrary names. This is
true for the <activity /> syntax as well as the <subprocess ref="activity"
/> one.

> Concerning the output, we could maybe have a switch in the
> preferences, enabling one or the other variant.
> What do you think ?

I think so too.

Lukas



-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
OpenWFE - Open source WorkFlow Engine
OpenWFE-devel mailing list
OpenWFE-devel@...
https://lists.sourceforge.net/lists/listinfo/openwfe-devel

[John Mettraux]

Hi Lukas,

On 5/18/06, Lukas Eder <lukas.eder@...> wrote:

> John,
>
> > IMO, it helps making the language feel more natural,
> > especially if IT people made available such subprocesses to
> > the 'business designer'
> > folks so that they can express their processes in something a
> > [little] bit closer to natural language.
>
> Of course, I can see that. One problem is that there are quite a few
> reserved words and IT people as well as 'business designers' need to be
> aware of that. Furthermore, new expressions may break old process
> definitions. That may lead to problems that are hard to track. But I'm sure
> this is documented and users are in fact aware of the situation.
>
> I was just worrying about that "synonymous syntax" for the same semantics,
> because they do these things a lot in PHP to "facilitate" coding. But in the
> end, it becomes a great source of bugs. One example are the PHP magic
> quotes. You can configure your server to automatically escape all parameters
> ("'" becomes "\'" etc...) passed by GET or POST in order to prevent SQL
> injection and the likes. Now, this being a server configuration you never
> know whether or not those quotes were added... You get the point...

Maybe we could act on the 'input side' and ban certain names for
[sub]process-definitions.
Let's try not to enforce too much stuff (especially if we have to
hardcode that enforcement).

> > But on the other hand could we have something like
> > suggestion, when the designer (the person) types in 'act...',
> > the IDE proposes him with 'activity' (just asking, it's
> > similar to a method call).
>
> As I said, I can hard-code all kinds of additional behaviour on top of the
> XML schema definition, as the IDE does not access the schema directly but
> through a sort of proxy. There is a very simple model acting as that proxy,
> defining which kinds of child nodes are acceptable for each node. It would
> be easy to enhance that model. That way, the difference between a reserved
> word like <sequence /> or a user-defined word like <activity /> would be
> completely transparent to the user.
>
> However, the attributes in
> <activity aname="activity-c" acost="50" />
>
> cannot be discovered automatically as they may have arbitrary names. This is
> true for the <activity /> syntax as well as the <subprocess ref="activity"
> /> one.

Indeed, and it makes it even trickier.

I have to say, I trust you, you'll find an elegant solution that won't
spoil what you currently have.


Cheers,

--
John Mettraux   -///-   http://jmettraux.openwfe.org


-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid0709&bid&3057&dat1642
_______________________________________________
OpenWFE - Open source WorkFlow Engine
OpenWFE-devel mailing list
OpenWFE-devel@...
https://lists.sourceforge.net/lists/listinfo/openwfe-devel