Alan,
Alan W. Irwin wrote:
> On 2009-03-12 13:47-0000 Alban Rochel wrote:
>
> Hi Alban:
>
> I am answering you on the PLplot development list because I think others
> there will be interested in how you have improved qt so much.
>
> I have committed your work as revision 9729.
Thanks
> I suspect you should just revert your text placement changes for svgqt. We
> have found previously that any Linux SVG viewer/converter having to do with
> GNOME (e.g., eog) or ImageMagick (convert or display) relies on the librsvg
> library which has some severe text placement bugs. I have put in several
> bug reports about this, but the official GNOME "maintainer" of that library
> seems proud of the fact that GNOME does not use SVG that much and that
> therefore librsvg does not have to be maintained (!) Hopefully, GNOME will
> eventually replace that guy with someone who is competent, but for now you
> have to be careful of anything that depends on librsvg.
>
> I recommend using konqueror for getting the best text-placement results for
> SVG files. Scribus-ng (but not Scribus) was good, and there were some other
> reliable apps as well whose names I have forgotten. Right now, konqueror
> shows some text placement issues with the svg results, but I bet those go
> away once your SVG "correction" is removed.
Thanks for the info, but I use Inkscape for this, and I haven't found
that it was based on librsvg. Actually, the problem is not a placement
problem, it is that the bounding box returned by Qt is about 30% too
wide (but seems to have the good height) when I write the text, and I
rely on this bounding box to fit the element in the page. The problem
could be me forgetting to do something with resolutions for example
(however, the height would not be good). I'll try and see what happens
on a minimal example.
My fix is simply reducing this bounding box by an arbitrary factor, so
when switching to subscript for example, there may be a slight offset.
> I am not sure how to advise you with regards to example 23. My first guess
> was you were being too specific with the desired font names, but I don't
> spot anything in your code that is too specific that way. For example,
>
> f.setFamily("sans-serif")
>
> seems to use quite generic name which is exactly what you want. More
> likely, there is some way to control font lookup within Qt, and you have to
> set something or configure something to make it look for all "sans-serif"
> possibilities rather than taking, e.g., the first one (which is likely not
> to have all the glyphs you need).
There is a document on the font selection engine in the Qt
documentation, I'll have a look at it.
> I confirmed all the other issues you mentioned.
>
> One other issue I noticed was the glyphs are roughly a factor of two too
> small. (I would adjust the qt character size to mimic what you get for
> xcairo.)
Is this for every driver? For a specific example? I've just checked on
example 01: when pls->chrht=3.59mm (I assumed this was mm, as it is
returned by plgchr)
- png: 11 pix, 72dpi -> 3.89mm (not bad)
- eps: 3.8mm (ditto)
- svg: 2.71mm (ouch... Interesting however, the error factor is about
the same as for my bounding box...)
> I looked at example 24, and (aside from the size issue) looks good with the
> CTL (Complex Text Layout) languages being rendered correctly. So that is a
> huge improvement which shows the great potential there is with qt-related
> devices.
>
> Thanks very much for your continuing efforts with the qt device driver.
>
> Alan
>
Alban
------------------------------------------------------------------------------
Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are
powering Web 2.0 with engaging, cross-platform capabilities. Quickly and
easily build your RIAs with Flex Builder, the Eclipse(TM)based development
software that enables intelligent coding and step-through debugging.
Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com
_______________________________________________
Plplot-devel mailing list
Plplot-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/plplot-devel
