Re: Chemistry / CMIS status
On 28 Apr 2009, at 11:31, Dominique Pfister wrote: Just a quick thing I noticed while working on a JCR implementation of your API, in the methods that return collections, e.g.: List getChildren(String folderId, BaseType type, String filter, boolean includeAllowableActions, boolean includeRelationships, int maxItems, int skipCount, String orderBy, boolean[] hasMoreItems); Instead of having this extra boolean-array parameter "hasMoreItems" in order to return more than one value from this method, don't you think it'd be more appropriate to create a new class for the return type, containing both the collection and this flag? Yes, it would certainly be more object oriented. Currently the call sequence is very C-like, because in the initial version I wanted to have something lean without drowning into additional DTO classes. But certainly, in a coming refactoring we'll provide a small class to wrap whatever results are needed. There's another place where I used a similar passing style, boolean contentCopied[] for the checkIn() method. Here too I'll put a proper class to be OO. I did not see where this argument was actually used, so would it even be possible to entirely omit it? It's not used yet but it's specified by the CMIS domain model definition and the protocol bindings, so it'll get used eventually. Keep in mind that this getChildren method is part of the low-level SPI (which you have to implement for a new backend of course), in actuality an application user will just see the Folder.getChildren() method whose API is much simpler. I'm finishing up a cleaner separation of API from SPI that I'll commit in our Mercurial this afternoon, and anyway we'll talk about this more in Basel tomorrow I hope. Regards, Florent -- Florent Guillaume, Head of R&D, Nuxeo Open Source, Java EE based, Enterprise Content Management (ECM) http://www.nuxeo.com http://www.nuxeo.org +33 1 40 33 79 87
Re: Chemistry / CMIS status
Hi Florent, Just a quick thing I noticed while working on a JCR implementation of your API, in the methods that return collections, e.g.: List getChildren(String folderId, BaseType type, String filter, boolean includeAllowableActions, boolean includeRelationships, int maxItems, int skipCount, String orderBy, boolean[] hasMoreItems); Instead of having this extra boolean-array parameter "hasMoreItems" in order to return more than one value from this method, don't you think it'd be more appropriate to create a new class for the return type, containing both the collection and this flag? I did not see where this argument was actually used, so would it even be possible to entirely omit it? Kind regards Dominique On Wed, Apr 8, 2009 at 12:45 AM, Florent Guillaume wrote: > On 7 Apr 2009, at 10:23, Dominique Pfister wrote: >> >> Hi Florent, >> >> I was very busy lately, but now I finally found some time to look at >> your sources. >> >> Kind regards >> Dominique > > Excellent thanks. Looking forward to your comments. > I've had feedback from Bogdan here at Nuxeo as well, and we'll be > refactoring some of it during the week. > Also I still have to check in the WS code that was used for the AIIM iECM > demo. > > Florent > > -- > Florent Guillaume, Head of R&D, Nuxeo > Open Source, Java EE based, Enterprise Content Management (ECM) > http://www.nuxeo.com http://www.nuxeo.org +33 1 40 33 79 87 > >
Re: Chemistry / CMIS status
On Wednesday, April 8, 2009, Florent Guillaume wrote: > On 7 Apr 2009, at 14:43, David Nuescheler wrote: > > Hi All, > > Given the fact that we are now looking into applying florents contributions, > I would like to propose that we rename the "jcr-cmis" folder in sandbox > to "chemistry". Thoughts? > > regards, > david > > > > I'd suggest just creating a "chemistry" > folder in the sandbox and keeping the > "jcr-cmis" one until all the functionality in > jcr-cmis has been subsumed by Chemistry. Sounds like am great plan to me. Regards, David -- Visit: http://dev.day.com/
Re: Chemistry / CMIS status
On 7 Apr 2009, at 14:43, David Nuescheler wrote: Hi All, Given the fact that we are now looking into applying florents contributions, I would like to propose that we rename the "jcr-cmis" folder in sandbox to "chemistry". Thoughts? regards, david I'd suggest just creating a "chemistry" folder in the sandbox and keeping the "jcr-cmis" one until all the functionality in jcr-cmis has been subsumed by Chemistry. Florent -- Florent Guillaume, Head of R&D, Nuxeo Open Source, Java EE based, Enterprise Content Management (ECM) http://www.nuxeo.com http://www.nuxeo.org +33 1 40 33 79 87
Re: Chemistry / CMIS status
On 7 Apr 2009, at 10:23, Dominique Pfister wrote: Hi Florent, I was very busy lately, but now I finally found some time to look at your sources. Kind regards Dominique Excellent thanks. Looking forward to your comments. I've had feedback from Bogdan here at Nuxeo as well, and we'll be refactoring some of it during the week. Also I still have to check in the WS code that was used for the AIIM iECM demo. Florent -- Florent Guillaume, Head of R&D, Nuxeo Open Source, Java EE based, Enterprise Content Management (ECM) http://www.nuxeo.com http://www.nuxeo.org +33 1 40 33 79 87
Re: Chemistry / CMIS status
Hi All, Given the fact that we are now looking into applying florents contributions, I would like to propose that we rename the "jcr-cmis" folder in sandbox to "chemistry". Thoughts? regards, david On Tue, Apr 7, 2009 at 10:23 AM, Dominique Pfister wrote: > Hi Florent, > > I was very busy lately, but now I finally found some time to look at > your sources. > > Kind regards > Dominique > > On Wed, Mar 11, 2009 at 1:58 AM, Florent Guillaume wrote: >> Hi all, >> >> Here's a status report on the progress of the Chemistry code. >> >> The API has been tweaked a bit to separate a programmer-usable API from a >> lower-level SPI that mirrors the CMIS spec. >> There is now an AtomPub server implementation (work in progress) that allows >> some read-only operations (for now): getting repo info, type info, children >> listing and document retrieval). >> >> You can check the latest sources by downloading >> http://hg.nuxeo.org/sandbox/chemistry/archive/tip.zip >> This is a maven-buildable project. >> >> The project includes a sample servlet so that folks can test their CMIS >> AtomPub clients against a simple in-memory implementation of the API (which >> is still very incomplete, but progressing well). Using: >> mvn -Dmaven.test.skip=true package >> you'll get a self contained JAR which you can run with: >> java -jar >> chemistry-tests/target/chemistry-tests-0.1-SNAPSHOT-jar-with-dependencies.jar >> Just hit ^C to stop the server. >> >> As it's now at a stage where I'd really like more people eyeballing it and >> contributing, I plan on submitting it tomorrow so that it could be checked >> in by a committer in the sanbox -- if that's ok with you. >> >> Thanks, >> Florent >> >> -- >> Florent Guillaume, Head of R&D, Nuxeo >> Open Source, Java EE based, Enterprise Content Management (ECM) >> http://www.nuxeo.com http://www.nuxeo.org +33 1 40 33 79 87 >> >> > -- Visit: http://dev.day.com/
Re: Chemistry / CMIS status
Hi Florent, I was very busy lately, but now I finally found some time to look at your sources. Kind regards Dominique On Wed, Mar 11, 2009 at 1:58 AM, Florent Guillaume wrote: > Hi all, > > Here's a status report on the progress of the Chemistry code. > > The API has been tweaked a bit to separate a programmer-usable API from a > lower-level SPI that mirrors the CMIS spec. > There is now an AtomPub server implementation (work in progress) that allows > some read-only operations (for now): getting repo info, type info, children > listing and document retrieval). > > You can check the latest sources by downloading > http://hg.nuxeo.org/sandbox/chemistry/archive/tip.zip > This is a maven-buildable project. > > The project includes a sample servlet so that folks can test their CMIS > AtomPub clients against a simple in-memory implementation of the API (which > is still very incomplete, but progressing well). Using: > mvn -Dmaven.test.skip=true package > you'll get a self contained JAR which you can run with: > java -jar > chemistry-tests/target/chemistry-tests-0.1-SNAPSHOT-jar-with-dependencies.jar > Just hit ^C to stop the server. > > As it's now at a stage where I'd really like more people eyeballing it and > contributing, I plan on submitting it tomorrow so that it could be checked > in by a committer in the sanbox -- if that's ok with you. > > Thanks, > Florent > > -- > Florent Guillaume, Head of R&D, Nuxeo > Open Source, Java EE based, Enterprise Content Management (ECM) > http://www.nuxeo.com http://www.nuxeo.org +33 1 40 33 79 87 > >
Re: Chemistry / CMIS status
Hi, It's come to my attention that the Sun Java 5 compiler has bugs (related to generics) which can prevent compilation. The symptoms are: [INFO] Compilation failure /tmp/chemistry/chemistry-atompub/src/main/java/org/apache/chemistry/ atompub/PropertiesElement.java:[130,27] org.apache.abdera.model.Element cannot be dereferenced At least the following JDK versions have this bug: Linux: Java HotSpot(TM) Client VM (build 1.5.0_10-b03, mixed mode) Linux: Java HotSpot(TM) Client VM (build 1.5.0_14-b03, mixed mode, sharing) Mac OS X: Java HotSpot(TM) Client VM (build 1.5.0_16-133, mixed mode, sharing) For now please try compiling with a Java 6 JDK. Florent On 11 Mar 2009, at 00:58, Florent Guillaume wrote: Here's a status report on the progress of the Chemistry code. The API has been tweaked a bit to separate a programmer-usable API from a lower-level SPI that mirrors the CMIS spec. There is now an AtomPub server implementation (work in progress) that allows some read-only operations (for now): getting repo info, type info, children listing and document retrieval). You can check the latest sources by downloading http://hg.nuxeo.org/sandbox/chemistry/archive/tip.zip This is a maven-buildable project. The project includes a sample servlet so that folks can test their CMIS AtomPub clients against a simple in-memory implementation of the API (which is still very incomplete, but progressing well). Using: mvn -Dmaven.test.skip=true package you'll get a self contained JAR which you can run with: java -jar chemistry-tests/target/chemistry-tests-0.1-SNAPSHOT-jar- with-dependencies.jar Just hit ^C to stop the server. As it's now at a stage where I'd really like more people eyeballing it and contributing, I plan on submitting it tomorrow so that it could be checked in by a committer in the sanbox -- if that's ok with you. Thanks, Florent -- Florent Guillaume, Head of R&D, Nuxeo Open Source, Java EE based, Enterprise Content Management (ECM) http://www.nuxeo.com http://www.nuxeo.org +33 1 40 33 79 87
Chemistry / CMIS status
Hi all, Here's a status report on the progress of the Chemistry code. The API has been tweaked a bit to separate a programmer-usable API from a lower-level SPI that mirrors the CMIS spec. There is now an AtomPub server implementation (work in progress) that allows some read-only operations (for now): getting repo info, type info, children listing and document retrieval). You can check the latest sources by downloading http://hg.nuxeo.org/sandbox/chemistry/archive/tip.zip This is a maven-buildable project. The project includes a sample servlet so that folks can test their CMIS AtomPub clients against a simple in-memory implementation of the API (which is still very incomplete, but progressing well). Using: mvn -Dmaven.test.skip=true package you'll get a self contained JAR which you can run with: java -jar chemistry-tests/target/chemistry-tests-0.1-SNAPSHOT-jar- with-dependencies.jar Just hit ^C to stop the server. As it's now at a stage where I'd really like more people eyeballing it and contributing, I plan on submitting it tomorrow so that it could be checked in by a committer in the sanbox -- if that's ok with you. Thanks, Florent -- Florent Guillaume, Head of R&D, Nuxeo Open Source, Java EE based, Enterprise Content Management (ECM) http://www.nuxeo.com http://www.nuxeo.org +33 1 40 33 79 87