Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Magnus Jacobsson
    @magjac
    OK. Thanks.
    Magnus Jacobsson
    @magjac
    @innerop I've added https://github.com/magjac/d3-graphviz/blob/wip-port-to-hpcc-js-wasm4/CHANGELOG.md
    I will make the official release as soon as I get your feedback.
    @GordonSmith Thanks for the v0.3.6 release with the image support btw. Greatly appreciate it :pray:
    innerop
    @innerop
    @magjac my interpretation of this "However, there are two d3-graphviz functions, drawNode and drawEdge that calls the layout functions directly and if they are going to be used, the script type must be "application/javascript" or "text/javascript"."
    is that if we need to use drawNode or drawEdge then we can't use GraphViz in a Web Worker?
    N/m, I get what you mean now
    innerop
    @innerop
    I wonder why you can't call drawNode and drawEdge via postMessage (I've not used workers in a while but I recall that JS types serialize automatically via HTML5 structured cloning, done by the browser)
    at least in recent browsers...
    innerop
    @innerop
    @magjac only feedback so far is you need to mention how to include the script for d3-graphviz itself, e.g. <script src="https://unpkg.com/d3-graphviz@3.0.0-alpha.3/build/d3-graphviz.js"></script>
    everything seems to be working, at least in my demo
    i can't tell if it's snappier or not as I have a very small graph
    but so glad it's using the latest GraphViz and assuming @GordonSmith continues to keep up with GraphViz upstream or documents best practices (or tips) for building the WASM with Emscripten, we should be all good ...
    Magnus Jacobsson
    @magjac
    Thanks. Well, it's not using the latest Graphviz which is 2.42.3 which even that has a bug that is fixed in HEAD but not released. See our discussion above.
    Gordon Smith
    @GordonSmith
    FWIW graphviz/graphviz#1665 (I currently would need an official TAG to build from).
    Magnus Jacobsson
    @magjac
    d3-graphviz v3.0.0 has left the building! :tada:
    Thank you so much for your encouragement and support @GordonSmith @innerop
    Gordon Smith
    @GordonSmith
    Congrats and your welcome!
    (I guess I can bump mine to version 1 in the near future...)
    Magnus Jacobsson
    @magjac
    Hopefully you can include an upgrade to Graphviz 2.42.4 to get rid of graphviz/graphviz#1605
    Gordon Smith
    @GordonSmith
    Once its released let me know!
    innerop
    @innerop
    Your idea behind d3-graphviz is very unique. I've seen "graph stories" where they fade out one frame and fade in another (Dagre-D3 has such an example) but nothing like what you've done which involves node and edde tweening. The idea itself is very powerful. If combined with some markup for math equations it can be very useful for students of algebraic topology where animations would follow the flow of equations. It has many other use cases...
    Magnus Jacobsson
    @magjac
    @GordonSmith Will do!
    @innerop Thanks :slight_smile: that makes me glad to here. I'm always interested to see if someone has use for it.
    innerop
    @innerop
    http://rum-data.nature.com/neuro/journal/v20/n3/fig_tab/nn.4502_F3.html?foxtrotcallback=true
    I think scientists of all kinds who deal with graphs and topologies will find animation to be useful in illustrating concepts
    that's besides IP network topology monitoring, which is what I'm working on at the moment ...
    Magnus Jacobsson
    @magjac
    @GordonSmith Can you comment on https://github.com/magjac/d3-graphviz/issues/128#issuecomment-594690459 ?
    Gordon Smith
    @GordonSmith
    Hmmm - looks like I have gone out of my way to not include it...
    Rebuilding with now...
    innerop
    @innerop
    That’s interesting. In my demo I use the < <font ... > ... </font> > markup and that works ok (I think) but that may be unrelated
    Gordon Smith
    @GordonSmith
    That might get passed straight through? I did have it enabled in previous projects, just need to cmake FindEXPAT secret sauce...
    innerop
    @innerop
    Cool thanks! I didn’t even know that we can embed HTML structure inside nodes. I mean if the node border can be 0 and we embed a div then we can use CSS to create any custom shape. But I am not sure that GraphViz parses the CSS or if it does I’m not sure what subset of CSS it supports. But I can try and see ... interesting.
    Magnus Jacobsson
    @magjac
    @GordonSmith Thanks.
    Can you also comment on this issue: https://github.com/magjac/d3-graphviz/issues/127#issuecomment-594720193
    @innerop It's not actually HTML. It's HTML-like: https://www.graphviz.org/doc/info/shapes.html#html
    innerop
    @innerop
    Oh I see. Would have been great if it was HTML+CSS but I understand that the routing and layout algorithms would have to understand CSS and HTML too in that case in order to route/layout correctly ...
    innerop
    @innerop
    FWIW, people complaining about offline use cases should be more than ok running a Python or Node web server. It’s habit to kick off a web server process in each folder. D3 itself will have issues if there is no web server (when it needs to load data via .json and anything that employs XHR)
    Or they can do this: If using Chrome, the command line switch is --allow-file-access-from-files
    Not sure the flag would work in case of WASM but the underlying issue is XHR which is used to load WASM
    So it should work ...
    Magnus Jacobsson
    @magjac

    @innerop See above from Gordon:

    I do find launching chrome with the --disable-web-security helps in a lot of similar scenarios (only for debugging) and removes the need for a web server 90% of the time...
    If any of the web browsers supported a flag for similar behavior with web workers / wasm loading it would simplify the dev flow for sure...

    Hmm. Maybe not relevant...
    innerop
    @innerop
    Yeah the Chrome team came up with the file specific flag to solve the same problem but with much more narrowly defined security hole
    Imagine using disable-security fog for an offline app then using same browser instance on the web (bad)
    fog = flag
    Magnus Jacobsson
    @magjac
    @GordonSmith It seems easy enough to clone https://gitlab.com/graphviz/graphviz and build from HEAD which would eliminate the need for an official tag. Or do you not want to do that because of non-traceability?
    Gordon Smith
    @GordonSmith
    Correct the main reason to build from the tag is one of tractability - it would be easy enough to swap out the tag for a commit hash in the install script, but an official tag would be preferable...
    innerop
    @innerop
    I just realized @GordonSmith those Chrome flags work only for XHR but not for the Fetch API which is what I believe Emacripten generates in the JS to load the WASM.
    innerop
    @innerop
    I don’t know if you guys can answer this but if there is a glitch in a C++ app that makes it crash under some conditions, what happens when it’s ported you WASM? How does the browser handle its termination? I assume there is a global exception handler somewhere?
    Gordon Smith
    @GordonSmith
    Correct - but its not allways an exact match i.e. the c++ might work, but the wasm might crash (typically with memory allocation issues).
    innerop
    @innerop
    Yeah. Makes sense. They should really sort out dynamic memory growth so it is straight forward and default behavior.
    Gordon Smith
    @GordonSmith
    In wasm it is now a option (without the same overhead as ASM.js) and I have enabled FWIW
    Magnus Jacobsson
    @magjac
    @GordonSmith We have a new release: https://gitlab.com/graphviz/graphviz/-/archive/stable_release_2.42.4/graphviz-stable_release_2.42.4.tar.gz