Comparing Models

[floerio]

Hi!

I have an issue with the compare tool of the EA (7.1):

1) I copied an EAP file ("source").
2) I changed _only_ the notes field of an object in the copy
3) I exported the model to XMI
4) I started the source model
5) I compared the model with the XMI export of the copy

I expected only to see the change I made to the notes filed. And in fact, I saw it, but also the tool marked serveral constraints field.

It looked like that in one model CR/LF was interpreted as one character where as in the other model as two.

This only happens with constraint fields. No other text fields are marked falsely.

Does any one knows this behaviour?

Cheers,
Oliver

[Oliver F.]

I expected only to see the change I made to the notes filed. And in fact, I saw it, but also the tool marked serveral constraints field.

Indeed I have also noticed several other things showing up in the model when comparing to a previous version.
Especially the Audit view tends to be rather sensitive. When eg. changing a description of an element and then instantly using the conmpare function I can also see connectors and a few other items being marker as changed though I did not touch them at all.

I am not 100% sure whether I can treat it as a bug or a misusage from my side.

Oliver

[floerio]

Hi Oliver,

interessting to know that one is not allone with his problems :-)

Do you use the compate function for real or only for testing purpose? We plan to use it for out project business (creating EA projects from a "base line" and merging it back at the end of the project) so I am a little bit scared that, when the compare is not working properly, we have here an issue with our general process.

Cheers,
    Oliver

[Oliver F.]

Hi Oliver,

interessting to know that one is not allone with his problems :-)

Do you use the compate function for real or only for testing purpose? We plan to use it for out project business (creating EA projects from a "base line" and merging it back at the end of the project) so I am a little bit scared that, when the compare is not working properly, we have here an issue with our general process.

Actually I have tested it before deploying it to the public. And it scared me, too, because you need to be precise which item in the project browser you check before doing the compare. It reduces the number of changes a lot.
Besides that performance is getting extremely low the larger the number of elements to compare is. I did a compare over the whole model baseline and it brought the system into an unusable state for hours.

Oliver