Customised Bloom via HTTP

[tard_gabriel]

Hello there

Is it possible to share a customised Bloom with HTTP(s) with clients with a simple browser link?
And thus allowing the client to use and interact with it as a web app?

If so, is it possible to add an home made or customised statistics panel on the side?
Disable or enable certain options or panel to focus on what's important for a client.

That's about it, the rest Bloom deal with it really well.
I saw some personalised things on the website using Bloom in the background and I was wondering how these are done.

Thanks

[marius_conjeaud]

Hello,

You should have a look at Bloom deep links here : Bloom features in detail - Neo4j Bloom

This allows you to generate and share perspectives via a link.

As for your second question, this points towards an iframe to include inside your own webpage.

Marius

[abk]

None of this is production, but what would you think about...

There has been some preliminary work on a thing called @relate/web which is essentially Neo4j Desktop as a web server relate/packages/web at master · neo4j-devtools/relate · GitHub

With Bloom shifting from local storage to file/API storage, that could make sharing Bloom perspectives and scenes a reasonable activity, similar to Tableau's server.

Bloom scenes (a perspective + particular DB) could be embeddable in a google maps sort of way, with enough of an API to coordinate state+actions with other stuff on the web page.

Would that be interesting? If so, how interesting would it be?

-ABK

[tard_gabriel]

Thank you @marius.conjeaud and @abk for you answers.
They both provide information I needed, and indeed I will need to go further about this.

I'm working to build a real time visualiser for a client and only Bloom is able to archive that for now.

I tried the Javascript Keylines toolkit, it works really great with good customisation for quick and sweet website integration. But as it's using JavaScript there is this 1 CPU interpreted processing limitation. So loading the data into the keylines chart when there is a lot of them there is always a gap and it's not as instant as Bloom. I guess Bloom can work directly with Neo4j data with a binary stream without any data conversion between Neo4j and Bloom or if conversion there is it's probably using all processors with a more efficient language. And for the rendering part Bloom is using real GPU computing which make a huge difference. Keylines rendering is ok but based on webGL I was not expecting a miracle with big data.

[abk · Accepted solution]

Bloom does a good job of utilizing both CPU and GPU processing.

Other options to look at:

• Ogma from Linkurious
• Graphistry

Cheers,
ABK