28 Apr
2011
28 Apr
'11
12:34 p.m.
Hi all
I am a student at the University of Bern and for my Master's Thesis I am
using the FAMIX metamodel and I also intend to use Fame.
One thing that is most valuable for me is the possibility to generate
metamodel stub code from an fm3 mse file, in my case Java classes.
While filling in the stub classes for FAMIX core, I realized that much
of that work could also be done by the generator, if only it had some
additional information. So I suggest adding that information as pragmas.
Let me explain that on behalf of an example.
The message FAMIXNamedEntity>>belongsTo should be overridden in all of
FAMIXType
FAMIXScopingEntity
FAMIXMethod
FAMIXFunction
FAMIXLocalVariable
FAMIXUnknownVariable
FAMIXGlobalVariable
FAMIXParameter
FAMIXAttribute
FAMIXImplicitVariable
(and more outside FAMIX core)
but a code generator does not know that, so I have to implement the
corresponding messages/methods by hand. That is a bit annoying since
this really is very simple code that could easily be generated.
Example for one of the methods that have to be written by hand:
FAMIXAttribute>>belongsTo
^ self parentType
Every implementation of 'belongsTo' looks like that, i.e. is a very
simple delegate, or let us call it an "alias".
I suggest to add the information about such "aliases" to the pragmas:
1) add <alias> to NamedEntity>>belongsTo
2) add <aliasedBy #belongsTo> to all subclass properties, e.g. to
FAMIXAttribute>>parentType
Then, with that kind of information in an mse file, a code generator
could do more work for me and I had time to drink an additional coffee
or two :)
And maybe metamodel browsers can profit from that kind of information, too.
Cheers,
Raffael Krebs
28 Apr
28 Apr
1:59 p.m.
Hi Rafael,
Nice to see you again :).
The nice feature of FM3 is that it is small. Your change would imply extending FM3 with
alias information, and then getting the code generators to understand this part.
The problem is that the use case we are talking about is so small, that I am not sure it
warrants the trouble of changing FM3.
But, could you tell us more about the use case you are in? Is it generating Famix an
action that you have to do all the time?
Cheers,
Doru
On 28 Apr 2011, at 12:34, Raffael Krebs wrote:
Hi all
I am a student at the University of Bern and for my Master's Thesis I am using the
FAMIX metamodel and I also intend to use Fame.
One thing that is most valuable for me is the possibility to generate metamodel stub code
from an fm3 mse file, in my case Java classes.
While filling in the stub classes for FAMIX core, I realized that much of that work could
also be done by the generator, if only it had some additional information. So I suggest
adding that information as pragmas.
Let me explain that on behalf of an example.
The message FAMIXNamedEntity>>belongsTo should be overridden in all of
FAMIXType
FAMIXScopingEntity
FAMIXMethod
FAMIXFunction
FAMIXLocalVariable
FAMIXUnknownVariable
FAMIXGlobalVariable
FAMIXParameter
FAMIXAttribute
FAMIXImplicitVariable
(and more outside FAMIX core)
but a code generator does not know that, so I have to implement the corresponding
messages/methods by hand. That is a bit annoying since this really is very simple code
that could easily be generated.
Example for one of the methods that have to be written by hand:
FAMIXAttribute>>belongsTo
^ self parentType
Every implementation of 'belongsTo' looks like that, i.e. is a very simple
delegate, or let us call it an "alias".
I suggest to add the information about such "aliases" to the pragmas:
1) add <alias> to NamedEntity>>belongsTo
2) add <aliasedBy #belongsTo> to all subclass properties, e.g. to
FAMIXAttribute>>parentType
Then, with that kind of information in an mse file, a code generator could do more work
for me and I had time to drink an additional coffee or two :)
And maybe metamodel browsers can profit from that kind of information, too.
Cheers,
Raffael Krebs
_______________________________________________
Moose-dev mailing list
Moose-dev(a)iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
--
www.tudorgirba.com
"To lead is not to demand things, it is to make them happen."
29 Apr
29 Apr
3 p.m.
Hi Tudor,
I agree that the use case is small but i don't see how adding some pragmas would
change FM3. I mean the idea is just to have more static information stored into pragmas
that on the one hand can improve the comprehension of the meta-model itself (e.g. knowing
which methods are just aliases for something else and so have the same semantic), on the
other hand can be use in other contexts like, for instance, the automatically generation
of code to make it more efficient.
I agree that the code generation is not something that you do all the time but if you want
to have something that can go back a forth from a platform to Moose and the other way
around it is necessary to make this passage as easy and automatic as possible.
Cheers,
Fabrizio
On 28 Apr 2011, at 13:59, Tudor Girba wrote:
Hi Rafael,
Nice to see you again :).
The nice feature of FM3 is that it is small. Your change would imply extending FM3 with
alias information, and then getting the code generators to understand this part.
The problem is that the use case we are talking about is so small, that I am not sure it
warrants the trouble of changing FM3.
But, could you tell us more about the use case you are in? Is it generating Famix an
action that you have to do all the time?
Cheers,
Doru
On 28 Apr 2011, at 12:34, Raffael Krebs wrote:
Hi all
I am a student at the University of Bern and for my Master's Thesis I am using the
FAMIX metamodel and I also intend to use Fame.
One thing that is most valuable for me is the possibility to generate metamodel stub code
from an fm3 mse file, in my case Java classes.
While filling in the stub classes for FAMIX core, I realized that much of that work could
also be done by the generator, if only it had some additional information. So I suggest
adding that information as pragmas.
Let me explain that on behalf of an example.
The message FAMIXNamedEntity>>belongsTo should be overridden in all of
FAMIXType
FAMIXScopingEntity
FAMIXMethod
FAMIXFunction
FAMIXLocalVariable
FAMIXUnknownVariable
FAMIXGlobalVariable
FAMIXParameter
FAMIXAttribute
FAMIXImplicitVariable
(and more outside FAMIX core)
but a code generator does not know that, so I have to implement the corresponding
messages/methods by hand. That is a bit annoying since this really is very simple code
that could easily be generated.
Example for one of the methods that have to be written by hand:
FAMIXAttribute>>belongsTo
^ self parentType
Every implementation of 'belongsTo' looks like that, i.e. is a very simple
delegate, or let us call it an "alias".
I suggest to add the information about such "aliases" to the pragmas:
1) add <alias> to NamedEntity>>belongsTo
2) add <aliasedBy #belongsTo> to all subclass properties, e.g. to
FAMIXAttribute>>parentType
Then, with that kind of information in an mse file, a code generator could do more work
for me and I had time to drink an additional coffee or two :)
And maybe metamodel browsers can profit from that kind of information, too.
Cheers,
Raffael Krebs
_______________________________________________
Moose-dev mailing list
Moose-dev(a)iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
--
www.tudorgirba.com
"To lead is not to demand things, it is to make them happen."
_______________________________________________
Moose-dev mailing list
Moose-dev(a)iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
3:22 p.m.
Hi Fabrizio,
On 29 Apr 2011, at 15:00, Fabrizio Perin wrote:
Hi Tudor,
I agree that the use case is small but i don't see how adding some pragmas would
change FM3. I mean the idea is just to have more static information stored into pragmas
that on the one hand can improve the comprehension of the meta-model itself (e.g. knowing
which methods are just aliases for something else and so have the same semantic), on the
other hand can be use in other contexts like, for instance, the automatically generation
of code to make it more efficient.
I agree that the code generation is not something that you do all the time but if you
want to have something that can go back a forth from a platform to Moose and the other way
around it is necessary to make this passage as easy and automatic as possible.
This back-and-forth between languages is achieved through FM3. Pragmas are only the means
we link the meta-descriptions to the Smalltalk implementation, but in the end, these
pragmas must be reflected in FM3 if you want the Java generator to take them into
account.
If you want to add a new pragma with a new meaning you will have to change FM3 with the
new information and then you will have to change the generators to take this new
information into account.
Now, I did not say that it is not interesting. It is. Of course, everyone is welcome to
give such an extension a try :). Will you guys try it?
But, basically the only place where this is needed right now is in belongsTo. So, the use
case is really small.
Cheers,
Doru
Cheers,
Fabrizio
On 28 Apr 2011, at 13:59, Tudor Girba wrote:
--
www.tudorgirba.com
"Don't give to get. Just give."
Hi Rafael,
Nice to see you again :).
The nice feature of FM3 is that it is small. Your change would imply extending FM3 with
alias information, and then getting the code generators to understand this part.
The problem is that the use case we are talking about is so small, that I am not sure it
warrants the trouble of changing FM3.
But, could you tell us more about the use case you are in? Is it generating Famix an
action that you have to do all the time?
Cheers,
Doru
On 28 Apr 2011, at 12:34, Raffael Krebs wrote:
_______________________________________________
Moose-dev mailing list
Moose-dev(a)iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev Hi all
I am a student at the University of Bern and for my Master's Thesis I am using the
FAMIX metamodel and I also intend to use Fame.
One thing that is most valuable for me is the possibility to generate metamodel stub code
from an fm3 mse file, in my case Java classes.
While filling in the stub classes for FAMIX core, I realized that much of that work could
also be done by the generator, if only it had some additional information. So I suggest
adding that information as pragmas.
Let me explain that on behalf of an example.
The message FAMIXNamedEntity>>belongsTo should be overridden in all of
FAMIXType
FAMIXScopingEntity
FAMIXMethod
FAMIXFunction
FAMIXLocalVariable
FAMIXUnknownVariable
FAMIXGlobalVariable
FAMIXParameter
FAMIXAttribute
FAMIXImplicitVariable
(and more outside FAMIX core)
but a code generator does not know that, so I have to implement the corresponding
messages/methods by hand. That is a bit annoying since this really is very simple code
that could easily be generated.
Example for one of the methods that have to be written by hand:
FAMIXAttribute>>belongsTo
^ self parentType
Every implementation of 'belongsTo' looks like that, i.e. is a very simple
delegate, or let us call it an "alias".
I suggest to add the information about such "aliases" to the pragmas:
1) add <alias> to NamedEntity>>belongsTo
2) add <aliasedBy #belongsTo> to all subclass properties, e.g. to
FAMIXAttribute>>parentType
Then, with that kind of information in an mse file, a code generator could do more work
for me and I had time to drink an additional coffee or two :)
And maybe metamodel browsers can profit from that kind of information, too.
Cheers,
Raffael Krebs
_______________________________________________
Moose-dev mailing list
Moose-dev(a)iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
--
www.tudorgirba.com
"To lead is not to demand things, it is to make them happen."
_______________________________________________
Moose-dev mailing list
Moose-dev(a)iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
4986
days inactive
4987
days old
3 comments
3 participants
participants (3)
-
Fabrizio Perin -
Raffael Krebs -
Tudor Girba