1 |
>>>>> On Wed, 23 Feb 2011, James Cloos wrote: |
2 |
|
3 |
> Gnuplot’s cairopdf terminal generates vector output. One might argue |
4 |
> that it is not optimal, but only because it uses line segments |
5 |
> rather than cubic curves to approximate graphs. (Ie, it uses cairo’s |
6 |
> lineto rather than curveto functions.) |
7 |
|
8 |
> But so do gnuplot’s native terminals (I tested postscript and svg.) |
9 |
|
10 |
> That is just how gnuplot draws. |
11 |
|
12 |
Thank you for this nice analysis. |
13 |
|
14 |
> I’d still keep the pdf USE flag with its current meaning; users may |
15 |
> have existing gnuplot scripts which use gnuplot’s pdf terminal |
16 |
> and/or apps which generate such scripts. |
17 |
|
18 |
I don't think that keeping media-libs/pdflib in the tree is realistic. |
19 |
There is no Gentoo maintainer and it has open security issues. |
20 |
Besides, version 7 is discontinued, and version 8 is available only as |
21 |
a binary (static) library under commercial licensing. |
22 |
|
23 |
Of course, such decisions are sometimes painful. But will they be less |
24 |
painful if we postpone them by another year or two? |
25 |
|
26 |
> Unless, of course, a patch is added which makes «set term pdf» an |
27 |
> alias for «set term cairopdf». |
28 |
|
29 |
Upstream has wisely named the terminal pdfcairo (not cairopdf). |
30 |
Therefore, gnuplot's autocompletion will do its job, without any |
31 |
additional patch. |
32 |
|
33 |
Ulrich |