|
Re: [EMF Compare] Support for dynamic EMF models: Indigo vs Helios [message #755350 is a reply to message #755118] |
Tue, 08 November 2011 08:42 |
|
Hi Tatiana,
What we dramatically improved is the comparison process as a whole; I can't really mention all bugs we treated that pertain to these aspects, some general information can be found on my blog for the merge improvements and for the scalability.
Other than that, we improved the support for proxies (both unresolved or "out of scope" of the comparison), the comparison algorithms for reference differences have been rethought, attribute differences are more accurately detected ...
All of these improvements have been made through Test-driven development ... and among the 4000 and more tests that have been added, and the number of failures we detected through this, most of the actual bugs haven't been tracked on bugzilla, rather fixed directly in the code. (We'd have needed to fill some dozen new bugs... and I was a little lazy ).
Laurent Goubet
Obeo
|
|
|
Re: [EMF Compare] Support for dynamic EMF models: Indigo vs Helios [message #755659 is a reply to message #755350] |
Wed, 09 November 2011 08:53 |
Tatiana Fesenko Messages: 62 Registered: July 2009 |
Member |
|
|
Hi Laurent,
Thank you for the quick response and the interesting blog posts.
I see, even though you didn't aim dynamic models support it was improved
as overall comparison algorithm was improved. Thank you for the work
you've done in this direction.
4000 automatic tests is a good result! I like TDD too, even though it's
not always easy to find time for writing tests prior to writing code. I
think writing tests for compare and merger deserves its own post too ;)
Thanks and regards,
On 11/8/11 9:42 AM, Laurent Goubet wrote:
> Hi Tatiana,
>
> What we dramatically improved is the comparison process as a whole; I
> can't really mention all bugs we treated that pertain to these aspects,
> some general information can be found on my blog for the
> http://eclipsemde.blogspot.com/2011/08/emf-compare-giant-steps-towards-working.html
> and for the
> http://eclipsemde.blogspot.com/2011/10/emf-compare-scalability.html.
>
> Other than that, we improved the support for proxies (both unresolved or
> "out of scope" of the comparison), the comparison algorithms for
> reference differences have been rethought, attribute differences are
> more accurately detected ...
>
> All of these improvements have been made through Test-driven development
> ... and among the 4000 and more tests that have been added, and the
> number of failures we detected through this, most of the actual bugs
> haven't been tracked on bugzilla, rather fixed directly in the code.
> (We'd have needed to fill some dozen new bugs... and I was a little lazy
> :p).
>
> Laurent Goubet
> Obeo
--
Tatiana Fesenko
ModelSolv, Inc.
Model-Oriented Solutions for Service Integration
|
|
|
Powered by
FUDForum. Page generated in 0.03208 seconds