I'm looking at structr as a promising CMS, especially since my interest is in using Neo4j to show relationships in a graph. A project I currently manage achieves this through a wiki-like environment (with metadata/tags and faceted browsing) in which persons can be linked to persons, and persons can be linked to organizations; the backend is MySQL, so joins with lots of data can be rather slow--a CMS running atop Neo4j seems to be a smart way to achieve increased performance for creating graphs with several hundred nodes/edges.
In structr, how are large chunks of text stored? Can a few paragraphs of text be stored as a property? Or would some other kind of store be needed for relatively small chunks? I've seen some of the comments on the Neo4j discussion boards about blob/document storage, and I was curious what the structr team has learned.
Also, going forward, will structr remain an open source project? I saw that tweet about possible investors in structr, and that sort of talk always sends up a red flag, at least for me.
Keep up the great work, and I look forward to updates. I'll try to install structr soon so I can have some useful feedback.
> I'm looking at structr as a promising CMS, especially since my > interest is in using Neo4j to show relationships in a graph. A project > I currently manage achieves this through a wiki-like environment (with > metadata/tags and faceted browsing) in which persons can be linked to > persons, and persons can be linked to organizations; the backend is > MySQL, so joins with lots of data can be rather slow--a CMS running > atop Neo4j seems to be a smart way to achieve increased performance > for creating graphs with several hundred nodes/edges.
> In structr, how are large chunks of text stored?
Internally, they are stored as String properties on Neo4j nodes and relationships.
Binary files (which can hold text of course) are stored in the filesystem with just a pointer (file system path and meta-data) stored in Neo4j.
> Can a few paragraphs of text be stored as a property?
Yes, anything up to some 100k is not a problem. The limits are rather on the network side than in the storage.
> Or would some other kind of store be needed for relatively small > chunks? I've seen some of the comments on the Neo4j discussion boards > about blob/document storage, and I was curious what the structr team > has learned.
We don't store blobs in Neo4j. You could store them in byte[] properties, but we think that would bloat the db too much. A valid option for the future may be to split a file into chunks and store them in a graph structure, allowing parallel read/write.
> Also, going forward, will structr remain an open source project? I saw > that tweet about possible investors in structr, and that sort of talk > always sends up a red flag, at least for me.
structr will definitely remain open source, as we believe in the Open Source way of doing software projects. We're even thinking about a non-commercial foundation to own all the rights. But of course we're thinking and talking about building a company to back structr and build a business around it.
> Keep up the great work, and I look forward to updates. I'll try to > install structr soon so I can have some useful feedback.
Thanks a lot! We will release a 0.5.2 version in a few days, together with a downloadable and more easy-to-handle installation package.
Thanks for the response. I look forward to the next update.
On a purely practical level, is there an OS that is best suited for deploying structr? I know this is often a matter of personal choice, but I'm at something of a crossroads as far as choosing a development platform: I have a Windows 7 setup, but I'm thinking of getting OS X, or even running Ubuntu, if that would help me with development, deployment, testing, etc.
On Friday, August 17, 2012 4:20:37 AM UTC-4, Axel Morgner wrote:
> Jack,
> thanks for joining!
> Hello:
> I'm looking at structr as a promising CMS, especially since my interest > is in using Neo4j to show relationships in a graph. A project I currently > manage achieves this through a wiki-like environment (with metadata/tags > and faceted browsing) in which persons can be linked to persons, and > persons can be linked to organizations; the backend is MySQL, so joins with > lots of data can be rather slow--a CMS running atop Neo4j seems to be a > smart way to achieve increased performance for creating graphs with several > hundred nodes/edges.
> In structr, how are large chunks of text stored?
> Internally, they are stored as String properties on Neo4j nodes and > relationships.
> Binary files (which can hold text of course) are stored in the filesystem > with just a pointer (file system path and meta-data) stored in Neo4j.
> Can a few paragraphs of text be stored as a property?
> Yes, anything up to some 100k is not a problem. The limits are rather on > the network side than in the storage.
> Or would some other kind of store be needed for relatively small chunks? > I've seen some of the comments on the Neo4j discussion boards about > blob/document storage, and I was curious what the structr team has learned.
> We don't store blobs in Neo4j. You could store them in byte[] properties, > but we think that would bloat the db too much. A valid option for the > future may be to split a file into chunks and store them in a graph > structure, allowing parallel read/write.
> Also, going forward, will structr remain an open source project? I saw > that tweet about possible investors in structr, and that sort of talk > always sends up a red flag, at least for me.
> structr will definitely remain open source, as we believe in the Open > Source way of doing software projects. We're even thinking about a > non-commercial foundation to own all the rights. But of course we're > thinking and talking about building a company to back structr and build a > business around it.
> Keep up the great work, and I look forward to updates. I'll try to > install structr soon so I can have some useful feedback.
> Thanks a lot! We will release a 0.5.2 version in a few days, together with > a downloadable and more easy-to-handle installation package.
Linux (Ubuntu, Debian, but any distro should do) and OS X are preferred, but we're working on a setup for Windows, too (see [1] <https://github.com/structr/structr/issues/20>).
For testing and dev, I personally would use a virtual machine with Ubuntu 12.04.
> Thanks for the response. I look forward to the next update.
> On a purely practical level, is there an OS that is best suited for > deploying structr? I know this is often a matter of personal choice, > but I'm at something of a crossroads as far as choosing a development > platform: I have a Windows 7 setup, but I'm thinking of getting OS X, > or even running Ubuntu, if that would help me with development, > deployment, testing, etc.
> --Jack
> On Friday, August 17, 2012 4:20:37 AM UTC-4, Axel Morgner wrote:
> Jack,
> thanks for joining!
>> Hello:
>> I'm looking at structr as a promising CMS, especially since my
>> interest is in using Neo4j to show relationships in a graph. A
>> project I currently manage achieves this through a wiki-like
>> environment (with metadata/tags and faceted browsing) in which
>> persons can be linked to persons, and persons can be linked to
>> organizations; the backend is MySQL, so joins with lots of data
>> can be rather slow--a CMS running atop Neo4j seems to be a smart
>> way to achieve increased performance for creating graphs with
>> several hundred nodes/edges.
>> In structr, how are large chunks of text stored?
> Internally, they are stored as String properties on Neo4j nodes
> and relationships.
> Binary files (which can hold text of course) are stored in the
> filesystem with just a pointer (file system path and meta-data)
> stored in Neo4j.
>> Can a few paragraphs of text be stored as a property?
> Yes, anything up to some 100k is not a problem. The limits are
> rather on the network side than in the storage.
>> Or would some other kind of store be needed for relatively small
>> chunks? I've seen some of the comments on the Neo4j discussion
>> boards about blob/document storage, and I was curious what the
>> structr team has learned.
> We don't store blobs in Neo4j. You could store them in byte[]
> properties, but we think that would bloat the db too much. A valid
> option for the future may be to split a file into chunks and store
> them in a graph structure, allowing parallel read/write.
>> Also, going forward, will structr remain an open source project?
>> I saw that tweet about possible investors in structr, and that
>> sort of talk always sends up a red flag, at least for me.
> structr will definitely remain open source, as we believe in the
> Open Source way of doing software projects. We're even thinking
> about a non-commercial foundation to own all the rights. But of
> course we're thinking and talking about building a company to back
> structr and build a business around it.
>> Keep up the great work, and I look forward to updates. I'll try
>> to install structr soon so I can have some useful feedback.
> Thanks a lot! We will release a 0.5.2 version in a few days,
> together with a downloadable and more easy-to-handle installation
> package.
