Metamodel changes and Moose versionning
Locked
Metamodel changes and Moose versionning
 
|
Hi all,
We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. The situation is not really cool, and the message sent by this problem also. To solve this problem, I would propose to replace the version 4.3 by a version 5.0. Like that we can build tools to move from version 4.x to 5.x. And the message is better. What is your opinion about these changes and what is the impact on the community ? Thanks Jannik _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
|
|
+1
Alexandre On 15 Mar 2011, at 08:52, jannik.laval wrote: > Hi all, > > We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. > For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. > The situation is not really cool, and the message sent by this problem also. > > To solve this problem, I would propose to replace the version 4.3 by a version 5.0. > Like that we can build tools to move from version 4.x to 5.x. > And the message is better. > > What is your opinion about these changes and what is the impact on the community ? > Thanks > > Jannik > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
In reply to this post by jannik laval
I do not know exactly but it seems like a good idea.
Stef On Mar 15, 2011, at 12:52 PM, jannik.laval wrote: > Hi all, > > We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. > For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. > The situation is not really cool, and the message sent by this problem also. > > To solve this problem, I would propose to replace the version 4.3 by a version 5.0. > Like that we can build tools to move from version 4.x to 5.x. > And the message is better. > > What is your opinion about these changes and what is the impact on the community ? > Thanks > > Jannik > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
In reply to this post by jannik laval
What are the problems exactly?
One thing would be to add a Mse version metadata in each file. I think Cyrille has started something similar. Simon On 15 mars 2011, at 12:52, "jannik.laval" <[hidden email]> wrote: > Hi all, > > We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. > For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. > The situation is not really cool, and the message sent by this problem also. > > To solve this problem, I would propose to replace the version 4.3 by a version 5.0. > Like that we can build tools to move from version 4.x to 5.x. > And the message is better. > > What is your opinion about these changes and what is the impact on the community ? > Thanks > > Jannik > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
On Mar 15, 2011, at 13:28 , Simon Denier wrote: > What are the problems exactly? > > One thing would be to add a Mse version metadata in each file. I think Cyrille has started something similar. Yes, we should do that. But, if we want to keep versions of mse files, we should keep also the possibility to load them. Now, it is not possible. The idea of changing major version is to say: pay attention, changes are important, and we are making tools to move in the new version. This is what we did with Cyrille from Famix2 to Famix3. Jannik > > Simon > > On 15 mars 2011, at 12:52, "jannik.laval" <[hidden email]> wrote: > >> Hi all, >> >> We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. >> For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. >> The situation is not really cool, and the message sent by this problem also. >> >> To solve this problem, I would propose to replace the version 4.3 by a version 5.0. >> Like that we can build tools to move from version 4.x to 5.x. >> And the message is better. >> >> What is your opinion about these changes and what is the impact on the community ? >> Thanks >> >> Jannik >> _______________________________________________ >> Moose-dev mailing list >> [hidden email] >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
In reply to this post by simondenier
Hi,
I am on the run now, but what MSE files do not work? The changes to the FAMIX meta-model should not have broken anything. I have files created in 2009 that load just fine. So, before jumping to conclusions, please let's take a look at the actual cases. Could you make available such an MSE file? Cheers, Doru On 15 Mar 2011, at 13:28, Simon Denier wrote: > What are the problems exactly? > > One thing would be to add a Mse version metadata in each file. I think Cyrille has started something similar. > > Simon > > On 15 mars 2011, at 12:52, "jannik.laval" <[hidden email]> wrote: > >> Hi all, >> >> We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. >> For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. >> The situation is not really cool, and the message sent by this problem also. >> >> To solve this problem, I would propose to replace the version 4.3 by a version 5.0. >> Like that we can build tools to move from version 4.x to 5.x. >> And the message is better. >> >> What is your opinion about these changes and what is the impact on the community ? >> Thanks >> >> Jannik >> _______________________________________________ >> Moose-dev mailing list >> [hidden email] >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev -- www.tudorgirba.com "No matter how many recipes we know, we still value a chef." _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
Hi,
I am not jumping to conclusion, I am beginning discussion :) With a small time to check, it seems that it is MooseCook that is broken. Here is the example: http://dl.dropbox.com/u/7739334/seaside30-28062010.mse.zip I put it in Dropbox because of 3Mb. You can load it in Moose without problems. Now, if you browse "All Model Classes" and you select one class, there is an error. because a FamixClass does not now a parentType. It seems that there is a change in scope. Cheers, Jannik On Mar 15, 2011, at 13:50 , Tudor Girba wrote: > Hi, > > I am on the run now, but what MSE files do not work? The changes to the FAMIX meta-model should not have broken anything. > > I have files created in 2009 that load just fine. So, before jumping to conclusions, please let's take a look at the actual cases. Could you make available such an MSE file? > > Cheers, > Doru > > > On 15 Mar 2011, at 13:28, Simon Denier wrote: > >> What are the problems exactly? >> >> One thing would be to add a Mse version metadata in each file. I think Cyrille has started something similar. >> >> Simon >> >> On 15 mars 2011, at 12:52, "jannik.laval" <[hidden email]> wrote: >> >>> Hi all, >>> >>> We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. >>> For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. >>> The situation is not really cool, and the message sent by this problem also. >>> >>> To solve this problem, I would propose to replace the version 4.3 by a version 5.0. >>> Like that we can build tools to move from version 4.x to 5.x. >>> And the message is better. >>> >>> What is your opinion about these changes and what is the impact on the community ? >>> Thanks >>> >>> Jannik >>> _______________________________________________ >>> Moose-dev mailing list >>> [hidden email] >>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >> >> _______________________________________________ >> Moose-dev mailing list >> [hidden email] >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > -- > www.tudorgirba.com > > "No matter how many recipes we know, we still value a chef." > > > > > > > > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
Hi Jannik,
On 15 Mar 2011, at 14:49, jannik.laval wrote: > Hi, > > I am not jumping to conclusion, I am beginning discussion :) Ok, then I misunderstood your statement :) > With a small time to check, it seems that it is MooseCook that is broken. > > Here is the example: http://dl.dropbox.com/u/7739334/seaside30-28062010.mse.zip > I put it in Dropbox because of 3Mb. > > You can load it in Moose without problems. > Now, if you browse "All Model Classes" and you select one class, there is an error. because a FamixClass does not now a parentType. > It seems that there is a change in scope. I took a look at your case study. The problem stems from a Smalltalk importer bug that was reported in Aug 2008 (!), and that is now finally fixed: the implicit variable used to be attached to the class and now it is attached to the method: http://code.google.com/p/moose-technology/issues/detail?id=16 Cheers, Doru > Cheers, > Jannik > > On Mar 15, 2011, at 13:50 , Tudor Girba wrote: > >> Hi, >> >> I am on the run now, but what MSE files do not work? The changes to the FAMIX meta-model should not have broken anything. >> >> I have files created in 2009 that load just fine. So, before jumping to conclusions, please let's take a look at the actual cases. Could you make available such an MSE file? >> >> Cheers, >> Doru >> >> >> On 15 Mar 2011, at 13:28, Simon Denier wrote: >> >>> What are the problems exactly? >>> >>> One thing would be to add a Mse version metadata in each file. I think Cyrille has started something similar. >>> >>> Simon >>> >>> On 15 mars 2011, at 12:52, "jannik.laval" <[hidden email]> wrote: >>> >>>> Hi all, >>>> >>>> We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. >>>> For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. >>>> The situation is not really cool, and the message sent by this problem also. >>>> >>>> To solve this problem, I would propose to replace the version 4.3 by a version 5.0. >>>> Like that we can build tools to move from version 4.x to 5.x. >>>> And the message is better. >>>> >>>> What is your opinion about these changes and what is the impact on the community ? >>>> Thanks >>>> >>>> Jannik >>>> _______________________________________________ >>>> Moose-dev mailing list >>>> [hidden email] >>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>> >>> _______________________________________________ >>> Moose-dev mailing list >>> [hidden email] >>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >> >> -- >> www.tudorgirba.com >> >> "No matter how many recipes we know, we still value a chef." >> >> >> >> >> >> >> >> _______________________________________________ >> Moose-dev mailing list >> [hidden email] >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev -- www.tudorgirba.com "Problem solving efficiency grows with the abstractness level of problem understanding." _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
On Mar 15, 2011, at 22:30 , Tudor Girba wrote: > Hi Jannik, > > On 15 Mar 2011, at 14:49, jannik.laval wrote: > >> Hi, >> >> I am not jumping to conclusion, I am beginning discussion :) > > Ok, then I misunderstood your statement :) no problem :) > >> With a small time to check, it seems that it is MooseCook that is broken. >> >> Here is the example: http://dl.dropbox.com/u/7739334/seaside30-28062010.mse.zip >> I put it in Dropbox because of 3Mb. >> >> You can load it in Moose without problems. >> Now, if you browse "All Model Classes" and you select one class, there is an error. because a FamixClass does not now a parentType. >> It seems that there is a change in scope. > > I took a look at your case study. The problem stems from a Smalltalk importer bug that was reported in Aug 2008 (!), and that is now finally fixed: the implicit variable used to be attached to the class and now it is attached to the method: > http://code.google.com/p/moose-technology/issues/detail?id=16 You are right, I remember we discussed with cyrille about this issue, but we did not see the impact on mse files. The problem is already here: we cannot import and use an mse from 4.1 So, I have no ideal solution. Maybe we should go back for this bug and integrate it in 5.0. Or we should do a hack in the importer that solve this problem. Have you some better solutions ? Another point that I would like to discuss: About the changes in Famix, there are only adding, ok. Now, it makes all importers evolve. >From a file 4.1 to 4.3, there is no problem and the new information fields are initialized at nil (or something else), there are not present in mse files. But from 4.3 to 4.1, there are multiple fields that are not taken into account in 4.1. What is the behavior ? Are these fields ignored (this is what we expect) ? or an error is raised ? My pont is: from 4.0 to 4.1, there are changes but the ecosystem had not to change. from 4.1 to 4.3, it seems (I say it seems, because I do not know about the real impact) that the changes impact the metamodel and a part of ecosystem: verveineJ, infusion, Cook (... maybe others ?) This is why I discuss to go to 5.0. Now, I am sorry to wake up too late, this is a discussion that we should had during your venue to Lille. Cheers, Jannik > > Cheers, > Doru > > > >> Cheers, >> Jannik >> >> On Mar 15, 2011, at 13:50 , Tudor Girba wrote: >> >>> Hi, >>> >>> I am on the run now, but what MSE files do not work? The changes to the FAMIX meta-model should not have broken anything. >>> >>> I have files created in 2009 that load just fine. So, before jumping to conclusions, please let's take a look at the actual cases. Could you make available such an MSE file? >>> >>> Cheers, >>> Doru >>> >>> >>> On 15 Mar 2011, at 13:28, Simon Denier wrote: >>> >>>> What are the problems exactly? >>>> >>>> One thing would be to add a Mse version metadata in each file. I think Cyrille has started something similar. >>>> >>>> Simon >>>> >>>> On 15 mars 2011, at 12:52, "jannik.laval" <[hidden email]> wrote: >>>> >>>>> Hi all, >>>>> >>>>> We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. >>>>> For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. >>>>> The situation is not really cool, and the message sent by this problem also. >>>>> >>>>> To solve this problem, I would propose to replace the version 4.3 by a version 5.0. >>>>> Like that we can build tools to move from version 4.x to 5.x. >>>>> And the message is better. >>>>> >>>>> What is your opinion about these changes and what is the impact on the community ? >>>>> Thanks >>>>> >>>>> Jannik >>>>> _______________________________________________ >>>>> Moose-dev mailing list >>>>> [hidden email] >>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>> >>>> _______________________________________________ >>>> Moose-dev mailing list >>>> [hidden email] >>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>> >>> -- >>> www.tudorgirba.com >>> >>> "No matter how many recipes we know, we still value a chef." >>> >>> >>> >>> >>> >>> >>> >>> _______________________________________________ >>> Moose-dev mailing list >>> [hidden email] >>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >> >> >> _______________________________________________ >> Moose-dev mailing list >> [hidden email] >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > -- > www.tudorgirba.com > > "Problem solving efficiency grows with the abstractness level of problem understanding." > > > > > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
Hi Jannik,
On 16 Mar 2011, at 09:01, jannik.laval wrote: > On Mar 15, 2011, at 22:30 , Tudor Girba wrote: ... >> >>> With a small time to check, it seems that it is MooseCook that is broken. >>> >>> Here is the example: http://dl.dropbox.com/u/7739334/seaside30-28062010.mse.zip >>> I put it in Dropbox because of 3Mb. >>> >>> You can load it in Moose without problems. >>> Now, if you browse "All Model Classes" and you select one class, there is an error. because a FamixClass does not now a parentType. >>> It seems that there is a change in scope. >> >> I took a look at your case study. The problem stems from a Smalltalk importer bug that was reported in Aug 2008 (!), and that is now finally fixed: the implicit variable used to be attached to the class and now it is attached to the method: >> http://code.google.com/p/moose-technology/issues/detail?id=16 > > You are right, I remember we discussed with cyrille about this issue, but we did not see the impact on mse files. > The problem is already here: we cannot import and use an mse from 4.1 > So, I have no ideal solution. Maybe we should go back for this bug and integrate it in 5.0. I do not understand what you mean. The fix is already integrated in 4.3. > Or we should do a hack in the importer that solve this problem. What importer? The MSE importer? > Have you some better solutions ? Yes. Remove all implicit variables that were produced with the old importer because they are simply wrong. > Another point that I would like to discuss: > About the changes in Famix, there are only adding, ok. Now, it makes all importers evolve. >> From a file 4.1 to 4.3, there is no problem and the new information fields are initialized at nil (or something else), there are not present in mse files. > But from 4.3 to 4.1, there are multiple fields that are not taken into account in 4.1. What is the behavior ? Are these fields ignored (this is what we expect) ? or an error is raised ? To be able to discuss this issue, we need to be specific. What fields are ignored? > My pont is: > from 4.0 to 4.1, there are changes but the ecosystem had not to change. > from 4.1 to 4.3, it seems (I say it seems, because I do not know about the real impact) that the changes impact the metamodel and a part of ecosystem: verveineJ, infusion, Cook (... maybe others ?) As I said, the inFusion that is available for download now is perfectly compatible. VerveineJ is still a work in progress, but even now it works with 4.3. By Cook, I think you mean Chef :). Chef relies on the proper FAMIX meta-model. As far as I see, this whole discussion seems to come from one single issues which was a bug in the Smalltalk importer in 4.1 that produced wrong information. There is not much we can do about that one except for ignoring the information that was produced. > This is why I discuss to go to 5.0. The meta-model changes in 4.3 should not crash anything major, it should only fix/improve things. So, this is why it is a small release, not a large one. When we will have something that either breaks things considerably or comes with something significantly different, we will change the prefix. Btw, just out of curiosity, how many files are we talking about that you have and cannot use? Cheers, Doru > Now, I am sorry to wake up too late, this is a discussion that we should had during your venue to Lille. > > Cheers, > Jannik > >> >> Cheers, >> Doru >> >> >> >>> Cheers, >>> Jannik >>> >>> On Mar 15, 2011, at 13:50 , Tudor Girba wrote: >>> >>>> Hi, >>>> >>>> I am on the run now, but what MSE files do not work? The changes to the FAMIX meta-model should not have broken anything. >>>> >>>> I have files created in 2009 that load just fine. So, before jumping to conclusions, please let's take a look at the actual cases. Could you make available such an MSE file? >>>> >>>> Cheers, >>>> Doru >>>> >>>> >>>> On 15 Mar 2011, at 13:28, Simon Denier wrote: >>>> >>>>> What are the problems exactly? >>>>> >>>>> One thing would be to add a Mse version metadata in each file. I think Cyrille has started something similar. >>>>> >>>>> Simon >>>>> >>>>> On 15 mars 2011, at 12:52, "jannik.laval" <[hidden email]> wrote: >>>>> >>>>>> Hi all, >>>>>> >>>>>> We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. >>>>>> For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. >>>>>> The situation is not really cool, and the message sent by this problem also. >>>>>> >>>>>> To solve this problem, I would propose to replace the version 4.3 by a version 5.0. >>>>>> Like that we can build tools to move from version 4.x to 5.x. >>>>>> And the message is better. >>>>>> >>>>>> What is your opinion about these changes and what is the impact on the community ? >>>>>> Thanks >>>>>> >>>>>> Jannik >>>>>> _______________________________________________ >>>>>> Moose-dev mailing list >>>>>> [hidden email] >>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>> >>>>> _______________________________________________ >>>>> Moose-dev mailing list >>>>> [hidden email] >>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>> >>>> -- >>>> www.tudorgirba.com >>>> >>>> "No matter how many recipes we know, we still value a chef." >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> _______________________________________________ >>>> Moose-dev mailing list >>>> [hidden email] >>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>> >>> >>> _______________________________________________ >>> Moose-dev mailing list >>> [hidden email] >>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >> >> -- >> www.tudorgirba.com >> >> "Problem solving efficiency grows with the abstractness level of problem understanding." >> >> >> >> >> _______________________________________________ >> Moose-dev mailing list >> [hidden email] >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev -- www.tudorgirba.com "Some battles are better lost than fought." _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
Hi Doru,
On Mar 16, 2011, at 11:21 , Tudor Girba wrote: > Hi Jannik, > > > On 16 Mar 2011, at 09:01, jannik.laval wrote: > >> On Mar 15, 2011, at 22:30 , Tudor Girba wrote: > > ... > >>> >>>> With a small time to check, it seems that it is MooseCook that is broken. >>>> >>>> Here is the example: http://dl.dropbox.com/u/7739334/seaside30-28062010.mse.zip >>>> I put it in Dropbox because of 3Mb. >>>> >>>> You can load it in Moose without problems. >>>> Now, if you browse "All Model Classes" and you select one class, there is an error. because a FamixClass does not now a parentType. >>>> It seems that there is a change in scope. >>> >>> I took a look at your case study. The problem stems from a Smalltalk importer bug that was reported in Aug 2008 (!), and that is now finally fixed: the implicit variable used to be attached to the class and now it is attached to the method: >>> http://code.google.com/p/moose-technology/issues/detail?id=16 >> >> You are right, I remember we discussed with cyrille about this issue, but we did not see the impact on mse files. >> The problem is already here: we cannot import and use an mse from 4.1 >> So, I have no ideal solution. Maybe we should go back for this bug and integrate it in 5.0. > > I do not understand what you mean. The fix is already integrated in 4.3. Yes, my idea is to rename 4.3 in 5.0 :) > >> Or we should do a hack in the importer that solve this problem. > > What importer? The MSE importer? yes, but this is really a crappy solution... > >> Have you some better solutions ? > > Yes. Remove all implicit variables that were produced with the old importer because they are simply wrong. ahah, you are right ! I will do that. Like that, my mse files will be valid for 4.1 and 4.3 Thank you Doru, Cheers, Jannik > >> Another point that I would like to discuss: >> About the changes in Famix, there are only adding, ok. Now, it makes all importers evolve. >>> From a file 4.1 to 4.3, there is no problem and the new information fields are initialized at nil (or something else), there are not present in mse files. >> But from 4.3 to 4.1, there are multiple fields that are not taken into account in 4.1. What is the behavior ? Are these fields ignored (this is what we expect) ? or an error is raised ? > > To be able to discuss this issue, we need to be specific. What fields are ignored? > >> My pont is: >> from 4.0 to 4.1, there are changes but the ecosystem had not to change. >> from 4.1 to 4.3, it seems (I say it seems, because I do not know about the real impact) that the changes impact the metamodel and a part of ecosystem: verveineJ, infusion, Cook (... maybe others ?) > > As I said, the inFusion that is available for download now is perfectly compatible. VerveineJ is still a work in progress, but even now it works with 4.3. > > By Cook, I think you mean Chef :). Chef relies on the proper FAMIX meta-model. As far as I see, this whole discussion seems to come from one single issues which was a bug in the Smalltalk importer in 4.1 that produced wrong information. There is not much we can do about that one except for ignoring the information that was produced. > >> This is why I discuss to go to 5.0. > > The meta-model changes in 4.3 should not crash anything major, it should only fix/improve things. So, this is why it is a small release, not a large one. When we will have something that either breaks things considerably or comes with something significantly different, we will change the prefix. > > Btw, just out of curiosity, how many files are we talking about that you have and cannot use? > > Cheers, > Doru > > >> Now, I am sorry to wake up too late, this is a discussion that we should had during your venue to Lille. >> >> Cheers, >> Jannik >> >>> >>> Cheers, >>> Doru >>> >>> >>> >>>> Cheers, >>>> Jannik >>>> >>>> On Mar 15, 2011, at 13:50 , Tudor Girba wrote: >>>> >>>>> Hi, >>>>> >>>>> I am on the run now, but what MSE files do not work? The changes to the FAMIX meta-model should not have broken anything. >>>>> >>>>> I have files created in 2009 that load just fine. So, before jumping to conclusions, please let's take a look at the actual cases. Could you make available such an MSE file? >>>>> >>>>> Cheers, >>>>> Doru >>>>> >>>>> >>>>> On 15 Mar 2011, at 13:28, Simon Denier wrote: >>>>> >>>>>> What are the problems exactly? >>>>>> >>>>>> One thing would be to add a Mse version metadata in each file. I think Cyrille has started something similar. >>>>>> >>>>>> Simon >>>>>> >>>>>> On 15 mars 2011, at 12:52, "jannik.laval" <[hidden email]> wrote: >>>>>> >>>>>>> Hi all, >>>>>>> >>>>>>> We are using Moose 4.3 and we have lots of problems due to the metamodel changes from 4.1 to 4.3. >>>>>>> For example all my mse files generated in 4.1 are obsolete in 4.3, but we develop tools in 4.3 not compatible with 4.1. >>>>>>> The situation is not really cool, and the message sent by this problem also. >>>>>>> >>>>>>> To solve this problem, I would propose to replace the version 4.3 by a version 5.0. >>>>>>> Like that we can build tools to move from version 4.x to 5.x. >>>>>>> And the message is better. >>>>>>> >>>>>>> What is your opinion about these changes and what is the impact on the community ? >>>>>>> Thanks >>>>>>> >>>>>>> Jannik >>>>>>> _______________________________________________ >>>>>>> Moose-dev mailing list >>>>>>> [hidden email] >>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>> >>>>>> _______________________________________________ >>>>>> Moose-dev mailing list >>>>>> [hidden email] >>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>> >>>>> -- >>>>> www.tudorgirba.com >>>>> >>>>> "No matter how many recipes we know, we still value a chef." >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> Moose-dev mailing list >>>>> [hidden email] >>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>> >>>> >>>> _______________________________________________ >>>> Moose-dev mailing list >>>> [hidden email] >>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>> >>> -- >>> www.tudorgirba.com >>> >>> "Problem solving efficiency grows with the abstractness level of problem understanding." >>> >>> >>> >>> >>> _______________________________________________ >>> Moose-dev mailing list >>> [hidden email] >>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >> >> >> _______________________________________________ >> Moose-dev mailing list >> [hidden email] >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > -- > www.tudorgirba.com > > "Some battles are better lost than fought." > > > > > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
In reply to this post by jannik laval
Hi all,
I have known and used moose for some years but now I am stepping into developing tools with it. So, I had a small experience with the changes in the model. We developed some code to calculate package metrics and the next week the code doesn't work. After spending a few hours on debugging and a helping hand from Jannik, I could resove the problem that came from the change in parentType. Although the problem was small, it took a newbie like quite some time to resolve. I knew all the methods regarding metrics calculation but when it came to meta-model I wasn't too sure what was going on.
Moreover, soon I am going to start work on Csharp MSE extractor. So, once we have the mse extractors for different langs, any changes to meta-model would require some work to align these extractors.
So, my suggestion is that any changes to the FAMIX meta-model should be considered major because they require quite a bit of effort behind the scene to align the tools using it.
_______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: Metamodel changes and Moose versionning
|
|
Hi Usman,
Great to see you getting active :) On 16 Mar 2011, at 15:53, Usman Bhatti wrote: > Hi all, > > I have known and used moose for some years but now I am stepping into developing tools with it. So, I had a small experience with the changes in the model. We developed some code to calculate package metrics and the next week the code doesn't work. After spending a few hours on debugging and a helping hand from Jannik, I could resove the problem that came from the change in parentType. Although the problem was small, it took a newbie like quite some time to resolve. I knew all the methods regarding metrics calculation but when it came to meta-model I wasn't too sure what was going on. Neither the definition nor the implementation of parentType changed in the meta-model. So what exact change are you talking about? :) > Moreover, soon I am going to start work on Csharp MSE extractor. So, once we have the mse extractors for different langs, any changes to meta-model would require some work to align these extractors. > > So, my suggestion is that any changes to the FAMIX meta-model should be considered major because they require quite a bit of effort behind the scene to align the tools using it. I agree, but until now there were no breaking changes in FAMIX. If you find cases where you suspect that a problem occurred, please raise it as a problem, not as a conclusion. It could very well be that the problem is in another place :) Cheers, Doru -- www.tudorgirba.com "Not knowing how to do something is not an argument for how it cannot be done." _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
«
Return to Moose
|
1 view|%1 views
| Free forum by Nabble | Edit this page |