From: Egon W. <ego...@gm...> - 2011-07-13 20:13:53
|
Hi all, originally the nonotify classes were faster, because they did not care about sending around update change events... However, there was a bug somewhere deep that made it possible to still get data classes, because the nonotify instances inherited from the data classes. Moreover, it turns out that the implementation is not as fast as it used to be, and even seems slower than the data classes (though I do not entirely understant why): http://chem-bla-ics.blogspot.com/2011/07/data-nonotify-or-silent.html I created a new module called 'silent', which does effectively the same as the nonotify module, but does not inherit from the data module. Both are still a drop in replacement, but the new 'silent' module even more than the nonotify module. Now, given the performance regression of nonotify and the bug, I would like to replace nonotify completely by 'silent' in the CDK 1.4.x series. But, since that branch is basically in freeze, we do not take API changes lightly. Particularly not removing a whole API (though replaced by another). Therefore, I am consulting the user community about who is using the nonotify module, and who uses the API beyond the NoNotificationChemObject builder. If you do, please shout out if the nonotify module is critical to your environment. In the next days (maybe even tonight) I hope to have some more elaborate performance statistics, in addition to that in my log already. Egon -- Dr E.L. Willighagen Postdoctoral Researcher Institutet för miljömedicin Karolinska Institutet (http://ki.se/imm) Homepage: http://egonw.github.com/ LinkedIn: http://se.linkedin.com/in/egonw Blog: http://chem-bla-ics.blogspot.com/ PubList: http://www.citeulike.org/user/egonw/tag/papers |