You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think (at least I like to think) that there's a way to achieve this using our attribute declaration.
Some questions that we need to answer:
Do we need all restyle docalc, doplot, ... distinctions ? Can you merge some of them?
Most docalc updates relate to our data_array and our arrayOk attributes. Meaning that we could simply use the attribute declaration to figure our what update path to take. Are there any exceptions to that rule?
anything else ?
And most importantly, much of restyle (and relayout) is un-tested. We should be very careful when making this change.
Carpet is going to add a huge number of attrs to recalcAttrs. (up to 287 total or so!?) The array of attributes is recalculated on every plot and it looks them up with indexOf. It should place them in a hash (e.g. {mode: true, visible: true, ...}) and should be defined in a plot-api-global context since it never changes.
@alexcjohnson notes that a larger refactoring is desired, but this might be a start since, at least once carpet is merged, it shouldn't take more than five minutes and may very well be at the point of having a non-negligible impact on performance.
Activity
etpinard commentedon Jun 15, 2016
I think (at least I like to think) that there's a way to achieve this using our attribute declaration.
Some questions that we need to answer:
docalc
,doplot
, ... distinctions ? Can you merge some of them?docalc
updates relate to ourdata_array
and ourarrayOk
attributes. Meaning that we could simply use the attribute declaration to figure our what update path to take. Are there any exceptions to that rule?And most importantly, much of
restyle
(andrelayout
) is un-tested. We should be very careful when making this change.scatter3d
snail trail lines [WIP] #617etpinard commentedon Apr 10, 2017
From @rreusser 's #1578
etpinard commentedon May 9, 2017
A pretty valuable first step was made in https://github.com/plotly/plotly.js/pull/1653/files/413e5459960c4b4c2c4c6212919e3a5c3183f981#diff-2941ab69a12080c0633ff4ac8ea3aa83
etpinard commentedon Sep 13, 2017
#1999 will probably close this.
etpinard commentedon Oct 5, 2017
Superseded by #2025