#490 svg circle

[Adam Majewski]

Hi,

I would like to draw the family of circles using gnuplot svg terminal.

Svg has a command circle for this. All the images which I have found use svg path command not circle, but circle is better because it takes less size ( optimisation).

In the gnuplot source code I have not found any command for using svg circle (only Point types 5 and 6 but it is cumbersome in practice because the units of "pointsize" have no connection to the axis coordinates of the plot)

Is it posible to add explicit circle command for svg terminal ?

============

I use : G N U P L O T Version 5.3 patchlevel 0 last modified 2018-03-05 in Ubuntu 18.04 LTS

TIA

[Ethan Merritt]

Ticket moved from /p/gnuplot/bugs/2184/

Can't be converted:

• _milestone:
• _priority:

[Mingye Wang]

The current "terminal" interface allows mostly just line segments, so every curve from circle to bezier are, like you said, represented using line segments. It might be quite a bit of work to add such an interface and make the appropriate "pass-through" calls, so don't expect this anytime soon.

The official stance is that it's fine the way it is right now. A "circle" thing was mentioned to be in the interface, but it was not used because of a lack of full drawing options or something. (But what about fallbacks? Still feels like a "not important enough" issue mainly.)

Many other vector formats like PDF are affected by this issue too.

[Ethan Merritt]

As per discussion tracked by the original Bug #2369, resolution of svg coordinates was increased by a factor of 10 in the current release version (5.4.1) and in the development source. To the best of my knowledge this resolved the issue as reported for svg.

I do not see the issue in PostScript output.

Have you tried increasing GP_CAIRO_SCALE as suggested in the issue tracker? Does that resolve your concern with pdf output? If not, please provide a sample plot that illustrates why the existing and increased resolution both fail to provide what you need. I have no objection to increasing the cairo oversampling, but it only makes sense if the increase actually addresses a real problem.

"many other vector formats" is pretty vague. What else did you have in mind?