Re: Revert general point reducing for charts, indroduced to solve Excel import problem
On Wed, Jul 31, 2013 at 7:25 AM, Regina Henschel wrote: > Hi all, > > an Excel chart with lot of points hanged on opening [1] and that was fixed > with r1388440. > > But it was done in a way, that it has undesirable effects. Now charts which > much data points, that show well in AOO3.4.1, are broken. [3][4] > > There exists already a data point reduction to eliminate "identical" points > for rendering, see class PlottingPositionHelper and its use. There the time > consuming rendering is reduced, but the data series itself are not changed. > > I think, that the way as it is done in [2], reducing the points in the data > series itself in general, is not the best approach, and it should be > reverted to solve the regression introduced by it. > +1 IMHO, better to be slow than wrong. There are other things we can do, beyond performance, if we want to deal better with very dense plots. For example, we could allow the user to modify alpha transparency. -Rob > Kind regards > Regina > > > [1] https://issues.apache.org/ooo/show_bug.cgi?id=121058 > [2] http://svn.apache.org/viewvc?view=revision&revision=1388440 > [3] https://issues.apache.org/ooo/show_bug.cgi?id=122822 > [4] https://issues.apache.org/ooo/show_bug.cgi?id=122907 > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: Revert general point reducing for charts, indroduced to solve Excel import problem
On 31.07.2013 13:25, Regina Henschel wrote: Hi all, an Excel chart with lot of points hanged on opening [1] and that was fixed with r1388440. But it was done in a way, that it has undesirable effects. Now charts which much data points, that show well in AOO3.4.1, are broken. [3][4] There exists already a data point reduction to eliminate "identical" points for rendering, see class PlottingPositionHelper and its use. There the time consuming rendering is reduced, but the data series itself are not changed. I think, that the way as it is done in [2], reducing the points in the data series itself in general, is not the best approach, Agreed, the data series itself should not be tampered with. Reducing the details only for visualization (i.e. in the chart module) is a much better approach. If reducing is desirable that is; but with Calc handling a million rows since OOo 3.3 its now too easy to make AOO's chart hang. Even millions of display items shouldn't be a problem on todays hardware. AFAIK the scalability problem comes notifications triggering other notification resulting in avalanche of them. Solving that would be best. and it should be reverted to solve the regression introduced by it. It is a bad regression indeed. The change that introduced it had bad problems because it only handled integers, some indices were not properly initialized and values other than minimum and maximum were ignored. We have solutions for the first two problems and another fix introducing some additional random sampling is easy. Until the chart-reduction or even better the notification-avalanche problems are solved maybe we should go with these suggested fixes as an intermediate step. Herbert - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org