Re: [Blueobelisk-discuss] http://wave.google.com/
1 view
Skip to first unread message
Cameron Neylon
Aug 16, 2009, 6:49:01 AM8/16/09
to David Wild, Egon Willighagen, knowled...@googlegroups.com, Antony Williams, blueobeli...@lists.sourceforge.net
Dear All
This is in response to a query on the Blue Obelisk mailing list (which supports an open source/standards development community for Chemistry) about using Wave. The original query arose from the idea of having a “News and Views” or response type article in a journal that could evolve over time. Other ideas involved manipulating 3d structures, searching PubChem, converting names to standard forms etc.
To the Blue Obelisk people, I have copied in a Google Group which we set up to try and coordinate activities to develop scientific tooling for Wave. Please feel free to join at http://groups.google.com/group/knowledge-waves or cross post. A small group of us are doing a demo at the Science Online meeting in London this Saturday which will be broadcast in Second Life and I will try to make sure a video or screencast makes it up ASAP.
Broadly speaking my feeling is that Wave will be a (or will inspire a set of) very effective collaborative authoring tools. The fundamental of Wave is that it facilitates communication between people and between people and services. A wave can be embedded within a Web page, which in turn enables people with Wave accounts to edit or comment, but my feeling is that in terms of publicly editable documents Wave is not a good fit. As a way of preparing rich, possibly semantic, documents that can be published and remain editable by a small group of people it could be very powerful.
The Wave protocol and API have been open sourced. It is not clear whether the server architecture (which is built on XMPP) will be as yet but I imagine Google will have to. The API enables Robots and Gadgets (check the documentation at http://code.google.com/apis/wave/guide.html) which can do interesting things. For instance I have hacked together a Robot which scans a Wave for text of the form ?chem(**chemicalname**) grabs the name, searches ChemSpider for an ID, and then inserts the text **chemicalname** (http://linktochemspiderentry) into the Wave. Code is at github.com/cameronneylon/ChemSpidey
It is crude, but I knocked it up in a matter of hours, most of which was ironing out undocumented incompatibilies between versions of the Wave API and AppEngine and realising that all my internal tests to make sure I was pushing around strings were choking because AppEngine was pushing unicode. My next aim is to spruce this up so that it also checks for text of the form e.g. chem:glucouse 5g and replaces it with the link plus a calculation of the number of moles. Inserting an image should also be straightforward but haven’t got that far yet. But marking up text, given a suitable web service to do the parsing is very very easy.
In terms of more sophisticated manipulations, e.g. Of 3d or 2 structures, or embedding spectral viewers there is an underlying problem. A Gadget could probably be relatively easily built that converted an underlying object or link within the wave in an embedded viewer. However this would not be a first class Wave object. That is other viewers of the same Wave would not see any manipulation of the e.g. 3d view. To make this work there needs to be a server side representation of the state of the object and it needs to send either 3d view or images over the wire. From talking to Peter it seems likely that this might involve significant re-engineering of existing code of e.g. Jmol to enable the split between server side and client side.
In terms of access I may be able to get a few people quickly into the developer Sandbox. The public beta is scheduled to open on September 30 and there is still a lot of work going on just to make things stable. There is a lot of hype and some perfectly reasonable concerns about tie-ins and whether there will be widespread adoption. My personal view is that Wave could make an extremely good interface layer for the average scientist to interact with sophisticated webservices. It may be possible to also make it a semantic authoring tool, either directly inside the wave by wrapping up the semantics into the XML of the wave itself or by pushing it out to e.g. External triple stores via a Robot. Whether or not it takes off and how widely applicable it is remains to be seen but I think if it doesn’t end up being the big thing, it is a view of where things are going on the web.
If it is possible I would very much work with BO people to coordinate with the others who are interested so as to try and keep things coordinated and prevent re-inventing of the wheel. There are some very obvious places (spotting text and replacing it, inserting images) where having a standard library would be helpful. One of the places where I think Wave could be very powerful would be as a Lab notebook. Hooking in the services the BO community have built would be a really powerful demonstrator of what could be done in my view.
Cheers
Cameron
This is in response to a query on the Blue Obelisk mailing list (which supports an open source/standards development community for Chemistry) about using Wave. The original query arose from the idea of having a “News and Views” or response type article in a journal that could evolve over time. Other ideas involved manipulating 3d structures, searching PubChem, converting names to standard forms etc.
To the Blue Obelisk people, I have copied in a Google Group which we set up to try and coordinate activities to develop scientific tooling for Wave. Please feel free to join at http://groups.google.com/group/knowledge-waves or cross post. A small group of us are doing a demo at the Science Online meeting in London this Saturday which will be broadcast in Second Life and I will try to make sure a video or screencast makes it up ASAP.
Broadly speaking my feeling is that Wave will be a (or will inspire a set of) very effective collaborative authoring tools. The fundamental of Wave is that it facilitates communication between people and between people and services. A wave can be embedded within a Web page, which in turn enables people with Wave accounts to edit or comment, but my feeling is that in terms of publicly editable documents Wave is not a good fit. As a way of preparing rich, possibly semantic, documents that can be published and remain editable by a small group of people it could be very powerful.
The Wave protocol and API have been open sourced. It is not clear whether the server architecture (which is built on XMPP) will be as yet but I imagine Google will have to. The API enables Robots and Gadgets (check the documentation at http://code.google.com/apis/wave/guide.html) which can do interesting things. For instance I have hacked together a Robot which scans a Wave for text of the form ?chem(**chemicalname**) grabs the name, searches ChemSpider for an ID, and then inserts the text **chemicalname** (http://linktochemspiderentry) into the Wave. Code is at github.com/cameronneylon/ChemSpidey
It is crude, but I knocked it up in a matter of hours, most of which was ironing out undocumented incompatibilies between versions of the Wave API and AppEngine and realising that all my internal tests to make sure I was pushing around strings were choking because AppEngine was pushing unicode. My next aim is to spruce this up so that it also checks for text of the form e.g. chem:glucouse 5g and replaces it with the link plus a calculation of the number of moles. Inserting an image should also be straightforward but haven’t got that far yet. But marking up text, given a suitable web service to do the parsing is very very easy.
In terms of more sophisticated manipulations, e.g. Of 3d or 2 structures, or embedding spectral viewers there is an underlying problem. A Gadget could probably be relatively easily built that converted an underlying object or link within the wave in an embedded viewer. However this would not be a first class Wave object. That is other viewers of the same Wave would not see any manipulation of the e.g. 3d view. To make this work there needs to be a server side representation of the state of the object and it needs to send either 3d view or images over the wire. From talking to Peter it seems likely that this might involve significant re-engineering of existing code of e.g. Jmol to enable the split between server side and client side.
In terms of access I may be able to get a few people quickly into the developer Sandbox. The public beta is scheduled to open on September 30 and there is still a lot of work going on just to make things stable. There is a lot of hype and some perfectly reasonable concerns about tie-ins and whether there will be widespread adoption. My personal view is that Wave could make an extremely good interface layer for the average scientist to interact with sophisticated webservices. It may be possible to also make it a semantic authoring tool, either directly inside the wave by wrapping up the semantics into the XML of the wave itself or by pushing it out to e.g. External triple stores via a Robot. Whether or not it takes off and how widely applicable it is remains to be seen but I think if it doesn’t end up being the big thing, it is a view of where things are going on the web.
If it is possible I would very much work with BO people to coordinate with the others who are interested so as to try and keep things coordinated and prevent re-inventing of the wheel. There are some very obvious places (spotting text and replacing it, inserting images) where having a standard library would be helpful. One of the places where I think Wave could be very powerful would be as a Lab notebook. Hooking in the services the BO community have built would be a really powerful demonstrator of what could be done in my view.
Cheers
Cameron
--
Scanned by iCritical.
Reply all
Reply to author
Forward
0 new messages