Re: [graphviz-interest] table formatting

John Ellson <[email protected]>
Newsgroups gmane.comp.video.graphviz
Message-ID <[email protected]>
George,

I'm sorry that graphviz isn't working for you in its current state.

I'm sure that its possible to write a PDF renderer, using poppler,
and/or a PS renderer using an alternate API to gs, but this would take
me a week or two of clear time, each, which I just don't have right
now.   If anyone else would like to take these on I'd be happy mentor.

John


On 02/08/2013 04:53 AM, George Georgalis wrote:
> /Hi John,/
> /Hope you didn't take this negatively. I could see how you would
> especially if graphviz support is your job. I thought you've been
> really helpful identifying the problem. But given the situation of Mac
> and Ubuntu; X11 may well be a dying protocol, not to mention gs, and
> Cairo seems to be coming apart too./
> /For what it's worth, giving up Latex and Graphviz for PowerPoint and
> Visio is pretty sorry, but I got to get the docs out and the state of
> unix text processing api is not a battle I'm ready for./
> /Do you have any ideas for the future of unix graphics api or pipeline?/
> /-George/
>
> On Thu, Feb 7, 2013 at 4:32 PM, George Georgalis <[email protected]
> <mailto:[email protected]>> wrote:
>
>     Hi John,
>     Unfortunately this is way outside of the domain I can help :\
>     Have you considered using inkscape libs? Seems well developed.
>     Today I tried installing a bunch of stuff to make svg files but I
>     hit walls.
>     What is the status of graphviz support at ATT?
>     Is it EOL or something resources can be devoted to?
>     -George
>
>     On Sun, Jan 20, 2013 at 10:01 AM, John Ellson
>     <[email protected] <mailto:[email protected]>> wrote:
>
>         George,
>
>         Support for devcairo  in ghostscript was removed in r11582
>
>         Here is my last note on the topic:
>
>         -------- Original Message --------
>         Subject: 	Re: [graphviz-interest] Including LaTeXed labels
>         without rasterisation
>         Date: 	Fri, 14 Sep 2012 19:25:53 -0400
>         From: 	John Ellson <[email protected]>
>         <mailto:[email protected]>
>         To: 	[email protected]
>         <mailto:[email protected]>
>
>
>
>         Just to follow up on this...
>
>         The ghostscript folks have pointed me to the OpenPrinter device inteface 
>         (opvp)
>         from which it looks like a cairo surface could be attached by writing some
>         interface code for ~20 primitive operations.     I've asked for a copy 
>         of the old
>         cairo device code which must contain code that mostly implements this.
>
>         The cairo mailing list has pointed me to poppler-glib, which is a PDF 
>         interpreter based on
>         xpdf, which contains support for cairo surfaces.
>
>         So we have probably two solutions for graphviz: a fix for PS 
>         imageloading, and an approach for PDF imageloading.
>
>         Neither is trivial, and I'm not sure how much time I can spend on this 
>         right now ... but someday ;-)
>
>         John
>
>
>
>
>
>         On 09/14/2012 05:47 PM, John Ellson wrote:
>         > Oh well, it was nice while it worked.
>         >
>         > According to:
>         >     http://bugs.ghostscript.com/show_bug.cgi?id=690009
>         >
>         >     "As of r11582 the devcairo device is removed."
>         >
>         > I guess they didn't realize that graphviz was successfully using it. :-(
>         >
>         >
>         > I'll try to reopen the bug to see if it can be added back another way...
>         >
>         >
>         > John
>         >
>         >
>         >
>         > On 09/14/2012 05:29 PM, John Ellson wrote:
>         >> Mohan,
>         >>
>         >> -Tps output of graphs with ps images,  seems to be working.
>         >>
>         >>     $ echo 'digraph { A [shape=none] }' | dot -Tps >a.ps <http://a.ps>
>         >>     $ echo 'digraph { B [label="" image="a.ps <http://a.ps>"] }' | dot -Tps | gv -
>         >>
>         >> This doesn't use libgs or cairo .... which will be needed to produce 
>         >> any other format,
>         >> but perhaps it is useful to you?
>         >>
>         >> John
>         >>
>         >>
>         >> On 09/14/2012 05:10 PM, John Ellson wrote:
>         >>> Mohan,
>         >>>
>         >>> So, in theory, what you should be able to do is render your labels 
>         >>> to postscript, and then use the postscript image loader from the 
>         >>> graphviz-gs rpm.
>         >>>
>         >>> This takes a postscript image and renderers it through libgs and 
>         >>> libcairo to whatever output format you are using, only converting to 
>         >>> bitmap if the final format is bitmap.
>         >>>
>         >>> I say "in theory" because last week someone asked about SVG image 
>         >>> loading and I found that the code had suffered some bit rot..... 
>         >>> thats now fixed in the latest development snapshots but
>         >>> it seems that we have more bit rot in the postscript image loader....
>         >>>
>         >>>     $ echo 'digraph { A [shape=none] }' | dot -Tps >a.ps <http://a.ps>
>         >>>     $ echo 'digraph { B [label="" image="a.ps <http://a.ps>"] }' | dot -Tpng | xv -
>         >>>     Error: a.ps <http://a.ps>: gsapi_init_with_args() returned: -100 "(null)" 
>         >>> (Ghostscript internal error)
>         >>>
>         >>> I'll investigate further....
>         >>>
>         >>>
>         >>> John
>         >>>
>         >>>
>         >>>
>         >>>
>         >>>
>         >>> On 09/14/2012 03:31 PM, Mohan Ganesalingam wrote:
>         >>>> I'm using dot to produce a large graph with (very) complicated 
>         >>>> labels. There's an example at 
>         >>>> http://people.ds.cam.ac.uk/mg262/derivation404.pdf . At the moment 
>         >>>> I am
>         >>>>
>         >>>> 1. Using Latex to generate all the labels in a PDF, one per page 
>         >>>> (e.g. http://people.ds.cam.ac.uk/mg262/derivation_images.pdf )
>         >>>> 2. Using ImageMagick's 'convert' to turn this into a sequence of 
>         >>>> (appropriately cropped) PNGs
>         >>>> 3. Using GraphViz to generate the actual graph as a PDF
>         >>>>
>         >>>> I'm very happy with the general results: GraphViz is, as ever, 
>         >>>> beautiful. The snag is that step 2. rasterises the labels, meaning 
>         >>>> that I either have to produce an enormous PDF or accept some 
>         >>>> blurring. Is there any way in which I can preserve the vector 
>         >>>> format? (Generating SVGs in step 2 doesn't work -- it rasterises 
>         >>>> things. Also, dvisvgm produces nonsense.)
>         >>>>
>         >>>> I should also say that I don't have to generate PDFs in step 1... 
>         >>>> anything that can be produced from LaTeX is fine.
>         >>>>
>         >>>> Thanks and best wishes,
>         >>>> Mohan
>         >>>> _______________________________________________
>         >>>> [email protected] <mailto:[email protected]>
>         >>>> https://mailman.research.att.com/mailman/listinfo/graphviz-interest
>         >>>
>         >>
>         >
>         > _______________________________________________
>         > [email protected] <mailto:[email protected]>
>         > https://mailman.research.att.com/mailman/listinfo/graphviz-interest
>
>         _______________________________________________
>         [email protected] <mailto:[email protected]>
>         https://mailman.research.att.com/mailman/listinfo/graphviz-interest
>
>
>
>
>
>
>         On 01/20/2013 01:24 AM, George Georgalis wrote:
>>
>>         Do you recall what version of ps has a supported api ? Maybe
>>         I can use that one?
>>         -g
>>
>>         On Jan 18, 2013 11:14 PM, "John Ellson"
>>         <[email protected] <mailto:[email protected]>> wrote:
>>
>>             On 01/18/2013 11:09 PM, Ryan Schmidt wrote:
>>
>>                 On Jan 18, 2013, at 21:27, George Georgalis wrote:
>>
>>                     As soon as I resolve some other issues, I am
>>                     planning to render some
>>                     table information in a separate program (Latex)
>>                     then figure out how to
>>                     embed them as nodes in my graph. I've never used
>>                     anything but default
>>                     (built-in) shapes for the nodes. Is it possible
>>                     to reference pdf node
>>                     tables (or something ray-traced) when we build a
>>                     graph? Best solution
>>                     I could imagine would include query of the pdf
>>                     tables for their
>>                     dimensions as the graph is made.
>>
>>                 Yes, you should be able to include PostScript (or PNG
>>                 or SVG or ...) files in graphs, anywhere you can put
>>                 an image. Attached is an example using PNGs (where
>>                 the included PNG was itself coincidentally generated
>>                 by Graphviz.
>>
>>
>>                 Postscript files should work too but I get this error
>>                 when I try:
>>
>>                 Error: 1.ps <http://1.ps>: gsapi_init_with_args()
>>                 returned: -100 "(null)" (Ghostscript internal error)
>>
>>                 gs is Ghostscript 9.06 on my system.
>>
>>
>>
>>             I think postscript embedding has broken on most
>>             platforms, since ghostscript developers withdrew support
>>             for the api that I used.
>>
>>             I was hoping to write a PDF embedder using libpoppler,
>>             but I haven't made any progress on it yet.
>>
>>             John
>>
>
>
>
>
>     -- 
>     George Georgalis, (415) 894-2710 <tel:%28415%29%20894-2710>,
>     http://www.galis.org/
>
>
>
>
> -- 
> George Georgalis, (415) 894-2710, http://www.galis.org/

_______________________________________________
[email protected]
http://lists.research.att.com/mailman/listinfo/graphviz-interest
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.