[great-users] runtime for crosslink creation increases exponentially

[Kevin Smyth]

Hi Tim,

It looks like creating a crosslink is a O(n) operation (in the number of
existing crosslinks). Looks like you want to make O(n^2) crosslinks, which
would make the transform O(n^3) (not exponential).

The slow runtime is a consequence of design decisions that aren't easily
fixed. However, I'd be willing to accept patches that don't break backwards
compatibility.

Sorry for the bad news.

Kevin

On Tue, Mar 28, 2017 at 2:44 PM, Belschner, Tim <
tim.belschner at ils.uni-stuttgart.de> wrote:

> Hello everybody,
>
>
>
> we are facing a runtime issue with our GReAT transformations, if
> crosslinks are used extensively. Due to the fact, that the runtime for our
> affected transformations are increasing exponentially if the number of
> objects within their input models is raised linear, we did some timing
> measurements. The measurements were conducted with our specific as well as
> an example transformation with different input model sizes. Depending on
> the size, a single crosslink creation takes up to 9 seconds.
>
> The problem’s source could be narrowed down to the crosslink creation
> within the effector ( e.g. “currMatch.matchContainer_4a.gl100dst()+=
> currMatch.element1_4c;“), but we struggled when examining it further on.
>
> This is why we would like to ask if someone could have a look at it?
>
>
>
> With the following link, we provide our example transformation and three
> input models. The transformation creates two kinds of crosslinks, between
> two objects of the same metamodel and between a metamodel and a temporary
> class defined in the transformation itself. The runtime issue occurs in
> both cases.
>
>
>
> https://belschner.org/owncloud/index.php/s/95rn9A1AhvOqU7S
>
> PW: GReAT
>
>
>
> Thanks in advance and best regards
>
>
>
> Tim
>
> _______________________________________________
> great-users mailing list
> great-users at list.isis.vanderbilt.edu
> http://list.isis.vanderbilt.edu/cgi-bin/mailman/listinfo/great-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.isis.vanderbilt.edu/pipermail/great-users/attachments/20170404/49bd9dc2/attachment.html>