After a coffee break and a quick chat with Norrie and David from EMC, the technical committee gathered together and posed for a team photo taken by none other than Jay Brown (IBM). Looking forward to seeing the result…
David Churchland (HP) reopened the late morning discussions on Secondary Types by asking for the reason why they should be needed if they are simply just properties.
There is a need to clarify "required" property to secondary types. This is currently ambiguous in the proposal. Our Alex (WeWebU) stressed that it wasn't clear how the system could communicate the default data value for the "required" secondary type, when needed at creation time. It was explained that this should be set automatically by the repository if a system value or by the application if a context-specific property. Nevertheless it was agreed that it should be defined more clearly, since there are differences in the semantics of the use of the "required" properties in the case of Secondary Types.
Olivier Grisel (Nuxeo) asked whether it would be a consideration to be able to filter secondary type property lists or would this be delivered in one large bundle. Would this be handled by the application or filtered by the server?
Break for lunch: another wonderful gourmet meal from Microsoft catering courtesy of the Microsoft Marketing team. I know that our host is one of the richest companies in the world, but it is still extremely generous of them to provide all with free food and drinks. With all the spare dollars in my pocket, I'll try to buy something nice at the Microsoft shop later
I need to buy the kids some souvenirs… Perhaps I'll buy them some MS Project licenses - I'm sure they'll love me for that 
David Churchland (HP) reopened the late morning discussions on Secondary Types by asking for the reason why they should be needed if they are simply just properties.
There is a need to clarify "required" property to secondary types. This is currently ambiguous in the proposal. Our Alex (WeWebU) stressed that it wasn't clear how the system could communicate the default data value for the "required" secondary type, when needed at creation time. It was explained that this should be set automatically by the repository if a system value or by the application if a context-specific property. Nevertheless it was agreed that it should be defined more clearly, since there are differences in the semantics of the use of the "required" properties in the case of Secondary Types.
Olivier Grisel (Nuxeo) asked whether it would be a consideration to be able to filter secondary type property lists or would this be delivered in one large bundle. Would this be handled by the application or filtered by the server?
Break for lunch: another wonderful gourmet meal from Microsoft catering courtesy of the Microsoft Marketing team. I know that our host is one of the richest companies in the world, but it is still extremely generous of them to provide all with free food and drinks. With all the spare dollars in my pocket, I'll try to buy something nice at the Microsoft shop later
I need to buy the kids some souvenirs… Perhaps I'll buy them some MS Project licenses - I'm sure they'll love me for that
Lunch over. Now down to business…
We're trying out OpenWorkdesk on Alfresco 4.0 today and it works, although we will need to rework the Alfresco CMIS extensions since we are having some problems with Aspects
However this doesn't appear to be a big problem and this is the purpose of the Plugfest: to sniff out those goddam bugs!
We got OpenWorkdesk on a Stick working with the IBM P8 cloud and Jay Brown has just copied it into a virtual machine and is going to send it to his colleagues to start testing.
Just had a coffee with Norrie Quinn from EMC. I really like this guy. I wonder if we can get something going with Documentum. That would be cool!
We're trying out OpenWorkdesk on Alfresco 4.0 today and it works, although we will need to rework the Alfresco CMIS extensions since we are having some problems with Aspects
However this doesn't appear to be a big problem and this is the purpose of the Plugfest: to sniff out those goddam bugs!We got OpenWorkdesk on a Stick working with the IBM P8 cloud and Jay Brown has just copied it into a virtual machine and is going to send it to his colleagues to start testing.
Just had a coffee with Norrie Quinn from EMC. I really like this guy. I wonder if we can get something going with Documentum. That would be cool!
Tools down at 16:30 and off to do some shopping with the team.
One of the perks this week, which almost everyone was looking forward to, was a shopping spree at the Microsoft Shop. The Microsoft Shop is renowned for selling the latest Microsoft software at a fraction of the price. For example, Windows 7 Ultimate edition retails at $50 and MS Office 2010 Professional for $60. If you own an XBox, there's a paradise of games for you here. Apart from the software and games, there are some really cool T-shirts for kids with slogans like "My daddy can program better than your daddy" and other souvenirs ranging from sports gear to cuddly toys. My kids won't be disappointed when I get home
One of the perks this week, which almost everyone was looking forward to, was a shopping spree at the Microsoft Shop. The Microsoft Shop is renowned for selling the latest Microsoft software at a fraction of the price. For example, Windows 7 Ultimate edition retails at $50 and MS Office 2010 Professional for $60. If you own an XBox, there's a paradise of games for you here. Apart from the software and games, there are some really cool T-shirts for kids with slogans like "My daddy can program better than your daddy" and other souvenirs ranging from sports gear to cuddly toys. My kids won't be disappointed when I get home
Another excellent evening, which was this time organized by Ryan McVeigh (Zia Consulting) at Earl's Steak House. As a European, we always hear these stories about the U.S where they serve up these massive steaks on a plate the size of small pacific island and then expect you to eat it all to prove that you're worthy to be a man. Very fortunately this wasn't the case. The steaks were large but moderately proportioned even for us Europeans.
I had a nice chat with Jay Brown (IBM), Bram de Kruijff (GX Software) and David Churchland (HP), who were all situated at my end of the table. It was a good evening.
Some of the team went off to shoot some pool afterwards, while the rest of us "losers" went on back to our hotels to hit the hay. Needless to say that I feel absolutely fine this morning with a clear head and ready for a good hearty Microsoft breakfast
I had a nice chat with Jay Brown (IBM), Bram de Kruijff (GX Software) and David Churchland (HP), who were all situated at my end of the table. It was a good evening.
Some of the team went off to shoot some pool afterwards, while the rest of us "losers" went on back to our hotels to hit the hay. Needless to say that I feel absolutely fine this morning with a clear head and ready for a good hearty Microsoft breakfast
After a nice breakfast with Jay Brown (IBM) and Olivier Grisel (Nuxeo), we entered the conference room for a 3rd day of Face2Face discussions.
By the way, I was asked by a colleague before coming out here to count the number of Macbooks used by the CMIS Technical Committee. I can confirm that there are 7 Macbooks, 7 Dells, 3 IBM Thinkpads and 2 HP Notebooks. The Mac users are Alfresco, Adobe, Zia Consulting and SAP.
From the start, David Churchland (HP) started proceedings with a discussion on stored queries and the use of virtual folders. The question is whether this should be included in CMIS or not. There are some usecases that would benefit from stored queries in the repository. However should this be visible in CMIS? There were some interesting discussions here neither for nor against their inclusion from Adam Harmetz (Microsoft), David Choy (EMC), Jens Hübel (Open Text), Jay Brown (IBM) and Florian Müller (Alfresco). Jay reminded us that we couldn't just simply paste a file into a virtual folder and David Choy (EMC) also said that this would be a complex problem that we would need to solve, if we were to pursue this further.
At 9:00 we started the phone-in meeting with some general announcements from David Choy (chairman) regarding the upcoming schedule of CMIS TC meetings and other OASIS related business.
David gave us a stark reminder that all proposals are the team's responsibility. Once accepted, the team (TC) is ultimately responsible for this item. As a team, we should try to work out each proposal and reach resolution on its own merit. We should also try to visualize a roadmap for CMIS and define a goal for the next release and scope out its content.
There should be 2 proposal steps: an initiation/development phase to allow the ideas behind the proposal to grow and mature and the opportunity to say NO if it really is a bad idea; an acceptance phase to allow the proposal to be critically reviewed by the technical committee and the opportunity to say YES if a good idea. There should also be 2 decisions per each proposal: one for acceptance and the other for inclusion in a release.
Reviewing the current state of CMIS, David Choy remarked that we had had a great start with CMIS 1.0 and that CMIS had gained quite a bit of visibility and interest. So far there are a lot of implementations of CMIS and its success can be accredited to the TC. However, CMIS is still in its infancy and therefore is yet to truly establish itself. For CMIS to become a successful standard would depend on its adoption, since many standards in the past never got traction. The top priority would be to increase and promote its adoption. Adoption would be the key to the success of CMIS, since without it there would be no systems to test it against. Once a standard is widely accepted, it should be driven more by consumers and the demand should drive technology.
Norrie Quinn (EMC) appealed with the committee not to rush anything, but to deliver the same quality as v1.0 to satisfy the expectations of consumers. Ryan McVeigh (Zia Consulting) expressed concern that if CMIS did not deliver what it promised, it would be deemed to death similar to JCR. So therefore CMIS is being viewed by customers with high expectations and great anticipation.
VOTING:
Browser Binding: the proposal was officially accepted and the majority voted for its release in the v1.1 spec, although there were some concerns raised by Norrie Quinn (EMC) that it would confuse CMIS users with an "optional" binding API and that it would make more sense to make it mandatory, since there appeared to be a sizable demand for this feature.
Coffee Break: had a nice chat with the Stephan Klevenz (SAP), Martin Hermes (SAP), Jens Hübel (Open Text) and David Caruana (Alfresco). David tells me that there is now a CMIS-bridge for Activiti and they have now got Activiti integrated into Alfresco. Brilliant !!! Apparently it's possible to download Activiti, install it and have it up-and-running in less than a minute… Seeing is believing. I'll have to time it when I'm back in the office next week
The meeting began again at 10:50 with Jay Brown opening discussions with the Type Mutability proposal. Olivier asked whether it would be possible to change the attribute of a property. This has not been covered yet. Norrie (EMC) also commented on localization, which Jay (IBM) admitted was something that had not yet been covered by the proposal, but could easily be adapted to do so. Norrie also questioned whether Type Mutability had enough value to be included in the v1.1 specification, to which Jay responded that it was particularly for his team's needs. How would this increase adoption if it would be included in v1.1? It makes it easier for vertical applications to benefit from this feature and should help to solve several namespacing problems.
Decison: the proposal had been previously accepted, so the vote was only for the release version. The majority voted for its release in the v1.1 spec.
Also Legal Holds were also discussed by Martin Hermes (SAP) and the consensus was that we should also work towards including this v1.1, although this is not formally approved yet. Adam Harmetz (Microsoft) would like to have a few of the details investigated further.
Ryan McVeigh (Zia Consulting) asked if anyone would consider volunteering to help with preparing a proposal for workflows. Jay Brown (IBM) suggested that it might be helpful to be able to extend the existing "super" object with CMIS for other scenarios that would be useful for workflow. David Caruana (Alfresco) recommended that we should not go down this route, because in his view it would get complicated quickly and it felt that business workflows should not be a part of the CMIS. Norrie said that he agreed with David's comments and that it was as if business workflows should exist in a completely different layer. Florian Müller (Alfresco) suggested that there should be a formal statement from the CMIS TC why we don't currently support this in the standard. Ryan proposed starting a discussion group to determine and define what was needed to add some basic workflow features. The TC needs a statement on solving this problem.
Lunch and WeWebU's opportunity to take centre stage. Yippee !!! Yes, it's our turn to put forward our Annotations proposal. Admittedly we are in the early stages of this proposal and trying to find our way forward with the committee, but here's our main opportunity to lay on the table to gauge the perspective of the committee. So in an attempt to eat and speak at the same time (which I admit is not advisable) I opened the discussion on our needs for annotations and the need for these to be handled separately from the content and treating them as separate objects that might have policies. I stressed that we were looking for a lightweight CMIS solution that would add little complexity and that we shouldn't overdefine annotations, which might restrict innovation from repository vendors and viewer manufacturers.
Alex continued discussions explaining our 4 possible options currently available to us to implement this. If you're not already familiar with our proposal, you can find it here: http://www.oasis-open.org/committees/download.php/42214/CMIS%20Annotation%20Proposal_v0.2.docx. Anyhow we proposed either to implement this as an ObjectService extension, a relationship extension, an integration into the secondary types proposal or add a new base type object cmis:annotation (Jay's idea).
There was quite a lot of discussion on this topic and there appeared to be a definite need to address annotations in the CMIS spec, but the question really was which way forward could we go with this. There was a lot of support from both Norrie (EMC) and Jay (IBM). Both IBM P8 and Documentum repositories support annotations and it was opening several questions with their companies how this could be supported by CMIS. David Caruana (Alfresco) wasn't in favor of any of the 4 options and proposed that we considered using renditions as the mechanism to relay annotations to the client. However Alex pointed out that this would not be feasible since renditions are read only and since it should be possible to create, edit and delete annotations at runtime. David Choy interjected and suggested that we should consider adding a 5th option and to spend some time looking at extending CMIS Renditions to incorporate write operations such as create, edit and delete. He also suggested that we should proceed by surveying as many repositories that supported annotations and extracting common semantics from all of these existing systems and modeling a system based on these findings. So there we have it… That's our homework cut out for us for the next few weeks.
By the way, I was asked by a colleague before coming out here to count the number of Macbooks used by the CMIS Technical Committee. I can confirm that there are 7 Macbooks, 7 Dells, 3 IBM Thinkpads and 2 HP Notebooks. The Mac users are Alfresco, Adobe, Zia Consulting and SAP.
From the start, David Churchland (HP) started proceedings with a discussion on stored queries and the use of virtual folders. The question is whether this should be included in CMIS or not. There are some usecases that would benefit from stored queries in the repository. However should this be visible in CMIS? There were some interesting discussions here neither for nor against their inclusion from Adam Harmetz (Microsoft), David Choy (EMC), Jens Hübel (Open Text), Jay Brown (IBM) and Florian Müller (Alfresco). Jay reminded us that we couldn't just simply paste a file into a virtual folder and David Choy (EMC) also said that this would be a complex problem that we would need to solve, if we were to pursue this further.
At 9:00 we started the phone-in meeting with some general announcements from David Choy (chairman) regarding the upcoming schedule of CMIS TC meetings and other OASIS related business.
David gave us a stark reminder that all proposals are the team's responsibility. Once accepted, the team (TC) is ultimately responsible for this item. As a team, we should try to work out each proposal and reach resolution on its own merit. We should also try to visualize a roadmap for CMIS and define a goal for the next release and scope out its content.
There should be 2 proposal steps: an initiation/development phase to allow the ideas behind the proposal to grow and mature and the opportunity to say NO if it really is a bad idea; an acceptance phase to allow the proposal to be critically reviewed by the technical committee and the opportunity to say YES if a good idea. There should also be 2 decisions per each proposal: one for acceptance and the other for inclusion in a release.
Reviewing the current state of CMIS, David Choy remarked that we had had a great start with CMIS 1.0 and that CMIS had gained quite a bit of visibility and interest. So far there are a lot of implementations of CMIS and its success can be accredited to the TC. However, CMIS is still in its infancy and therefore is yet to truly establish itself. For CMIS to become a successful standard would depend on its adoption, since many standards in the past never got traction. The top priority would be to increase and promote its adoption. Adoption would be the key to the success of CMIS, since without it there would be no systems to test it against. Once a standard is widely accepted, it should be driven more by consumers and the demand should drive technology.
Norrie Quinn (EMC) appealed with the committee not to rush anything, but to deliver the same quality as v1.0 to satisfy the expectations of consumers. Ryan McVeigh (Zia Consulting) expressed concern that if CMIS did not deliver what it promised, it would be deemed to death similar to JCR. So therefore CMIS is being viewed by customers with high expectations and great anticipation.
VOTING:
Browser Binding: the proposal was officially accepted and the majority voted for its release in the v1.1 spec, although there were some concerns raised by Norrie Quinn (EMC) that it would confuse CMIS users with an "optional" binding API and that it would make more sense to make it mandatory, since there appeared to be a sizable demand for this feature.
Coffee Break: had a nice chat with the Stephan Klevenz (SAP), Martin Hermes (SAP), Jens Hübel (Open Text) and David Caruana (Alfresco). David tells me that there is now a CMIS-bridge for Activiti and they have now got Activiti integrated into Alfresco. Brilliant !!! Apparently it's possible to download Activiti, install it and have it up-and-running in less than a minute… Seeing is believing. I'll have to time it when I'm back in the office next week
The meeting began again at 10:50 with Jay Brown opening discussions with the Type Mutability proposal. Olivier asked whether it would be possible to change the attribute of a property. This has not been covered yet. Norrie (EMC) also commented on localization, which Jay (IBM) admitted was something that had not yet been covered by the proposal, but could easily be adapted to do so. Norrie also questioned whether Type Mutability had enough value to be included in the v1.1 specification, to which Jay responded that it was particularly for his team's needs. How would this increase adoption if it would be included in v1.1? It makes it easier for vertical applications to benefit from this feature and should help to solve several namespacing problems.
Decison: the proposal had been previously accepted, so the vote was only for the release version. The majority voted for its release in the v1.1 spec.
Also Legal Holds were also discussed by Martin Hermes (SAP) and the consensus was that we should also work towards including this v1.1, although this is not formally approved yet. Adam Harmetz (Microsoft) would like to have a few of the details investigated further.
Ryan McVeigh (Zia Consulting) asked if anyone would consider volunteering to help with preparing a proposal for workflows. Jay Brown (IBM) suggested that it might be helpful to be able to extend the existing "super" object with CMIS for other scenarios that would be useful for workflow. David Caruana (Alfresco) recommended that we should not go down this route, because in his view it would get complicated quickly and it felt that business workflows should not be a part of the CMIS. Norrie said that he agreed with David's comments and that it was as if business workflows should exist in a completely different layer. Florian Müller (Alfresco) suggested that there should be a formal statement from the CMIS TC why we don't currently support this in the standard. Ryan proposed starting a discussion group to determine and define what was needed to add some basic workflow features. The TC needs a statement on solving this problem.
Lunch and WeWebU's opportunity to take centre stage. Yippee !!! Yes, it's our turn to put forward our Annotations proposal. Admittedly we are in the early stages of this proposal and trying to find our way forward with the committee, but here's our main opportunity to lay on the table to gauge the perspective of the committee. So in an attempt to eat and speak at the same time (which I admit is not advisable) I opened the discussion on our needs for annotations and the need for these to be handled separately from the content and treating them as separate objects that might have policies. I stressed that we were looking for a lightweight CMIS solution that would add little complexity and that we shouldn't overdefine annotations, which might restrict innovation from repository vendors and viewer manufacturers.
Alex continued discussions explaining our 4 possible options currently available to us to implement this. If you're not already familiar with our proposal, you can find it here: http://www.oasis-open.org/committees/download.php/42214/CMIS%20Annotation%20Proposal_v0.2.docx. Anyhow we proposed either to implement this as an ObjectService extension, a relationship extension, an integration into the secondary types proposal or add a new base type object cmis:annotation (Jay's idea).
There was quite a lot of discussion on this topic and there appeared to be a definite need to address annotations in the CMIS spec, but the question really was which way forward could we go with this. There was a lot of support from both Norrie (EMC) and Jay (IBM). Both IBM P8 and Documentum repositories support annotations and it was opening several questions with their companies how this could be supported by CMIS. David Caruana (Alfresco) wasn't in favor of any of the 4 options and proposed that we considered using renditions as the mechanism to relay annotations to the client. However Alex pointed out that this would not be feasible since renditions are read only and since it should be possible to create, edit and delete annotations at runtime. David Choy interjected and suggested that we should consider adding a 5th option and to spend some time looking at extending CMIS Renditions to incorporate write operations such as create, edit and delete. He also suggested that we should proceed by surveying as many repositories that supported annotations and extracting common semantics from all of these existing systems and modeling a system based on these findings. So there we have it… That's our homework cut out for us for the next few weeks.
