Not sure if it's WIP or not..

@Ankit Meel (ankitm) Thank you. I implemented your suggestions.

@Pablo Vazquez (pablovazquez) You might want to have a look here.

Visually this is a nice improvement, the current socket looks pretty bad.

However I'm not a fan of the way this is implemented, sticking a string comparison in low level code like this is probably not the way to go.
I wonder if adding a new socket shape and using it here would make sense.

Great improvement! Visually it's a +1 from me, code-wise I leave it to you @Hans Goudey (HooglyBoogly)

Thanks for working on it Fabian!

In D10080#255572, @Hans Goudey (HooglyBoogly) wrote:

I wonder if adding a new socket shape and using it here would make sense.

@Hans Goudey (HooglyBoogly) Just let me add some context:
The problem space here is how we get a different outline color for an individual socket. The fill color is returned by it draw_color function of each socket. That function samples the color from an Array based on the socket->type. The color for the outline is fixed at the moment and just uses the themes font color.
My first Idea was to add an draw_outline_color function to bNodeSocket simillar to the draw_color function. That would in my opinion be the best way to do it, but I thought maybe thats a bit overdone for this patch.
Second "problem" is that the socket type of "Virtual Socket" is -1 so we can't use it as index for a color in an array.

But.. What I don't understand myself is, why I haven't just compared the socket->type == -1 instead of comparing the id_name string. Would that be a better solution for you?

Sorry. I seems that i have accidently mixed up some branches. Have to disentangle them before uploading a new diff.

However I'm not a fan of the way this is implemented, sticking a string comparison in low level code like this is probably not the way to go.

@Hans Goudey (HooglyBoogly) I changed socket comparison from STREQLEN() to (socket_type == -1). Please tell me if you like that better.

Seems more reasonable now. Still too bad about the extra special case but I don't see a good way to avoid it.

Could it be that this patch makes sockets lose anti-alising? I'm testing the zip bundle patch from D10067

Maybe it's a different patch from the bundle. I'm a bit lost which patch does what.

At the bottom is how they look in the temp-nodes-redesign branch, which in terms of sockets it's basically same as master but with a thicker outline and fully opaque color (instead of the current half-transparent).

node_draw.c

• 32 - your changes do not require adding BKE_curve.h
• 819 - socket_type comparison to "-1" could use enum SOCK_CUSTOM
• 886 - avoid adding new lines that are not associated with your changes

In D10080#258451, @Pablo Vazquez (pablovazquez) wrote:

Could it be that this patch makes sockets lose anti-alising? I'm testing the zip bundle patch from D10067

Interresting! Good you found this. Haven't noticed yet. On my hi-dpi monitor its not really visible. This is for sure a problem with another patch => D10181 because of replacing the shader with a new one to be able to draw the new shape for Multi Input Sockets. I will fix this soon in https://developer.blender.org/D10181