Timestamps are in GMT/BST.
[6:49] -sendak.freenode.net- *** Looking up your hostname...
[6:49] -sendak.freenode.net- *** Checking Ident
[6:49] -sendak.freenode.net- *** Found your hostname
[6:49] -sendak.freenode.net- *** No Ident response
[6:49] * DuraLogBot (~PircBot@atlas.duraspace.org) has joined #duraspace
[6:49] * Topic is '[Welcome to DuraSpace - This channel is logged - http://irclogs.duraspace.org/]'
(manually inserted notes due to DuraLogBot failure...thanks ajs6f) [09:02] == ajs6f [89361d33@gateway/web/freenode/ip.184.108.40.206] has joined #duraspace [09:04] == eddies [~email@example.com] has joined #duraspace [09:04] == eddies [~firstname.lastname@example.org] has quit [Changing host] [09:04] == eddies [~eddies@unaffiliated/eddies] has joined #duraspace [09:04][6:49] * Set by cwilper!ad579d86@gateway/web/freenode/ip.220.127.116.11 on Fri Oct 22 01:19:41 UTC 2010
http://www.webexhibits.org/daylightsaving/franklin3.html [09:04] [ Full text - Benjamin Franklin - The Journal of Paris, 1784 ] - http://www.webexhibits.org/daylightsaving/franklin3.html [09:06] == ruckus [~ruckus@HSI-KBW-046-005-164-187.hsi8.kabel-badenwuerttemberg.de] has joined #duraspace [09:06] Adam reluctantly agrees to take notes. [09:06] == ruckus has changed nick to elschlomo [09:07] Possible special topic call for fcrepo-store. [09:08] Special topic all will be put off for a few weeks to ensure participation. [09:08] all -> call [09:11] Questions about JCR raised by Mark Diggory on mailing list. [09:11] Is Apache Jackrabbit scalable? [09:11] Does it support transactionality, and if so, how? [09:13] Frank: If we get into JCR, at what level of architecture would it be offered? [09:13] Chris: Could be several places. Northwestern implemented it at a high level. [09:14] Chris: Could be a JCR-specific implementation for storage. [09:14] Chris: Depends on what we want to commit to. [09:14] Dan: The JCR semantics are less capable than Fedora's. [09:15] Chris and Adam: They're different. "Disjoint" [09:15] Chris: Could be good to look at JCR's capabilities and what use it could be, but there's a cost of committing to those extra semantics. [09:16] Dan: JCR semantics are a great improvement over Java EE's API-centricity . [09:16] Dan: But Fedora is already even better. [09:17] Dan: Storage has two parts: authoring (where transactionality is very important) [09:17] and making it accisble to the world (where scalability is important). [09:17] accisible -> accessible [09:17] Dan: And now we have the cloud, which speaks to the latter, and JCR speaks to the former. [09:18] Dan: Tiered storage and asynchronous operations, grading out things for high scalability: these should also be on the table. [09:18] Frank: Ala EC2. [09:20] Chris: How does Jackrabbit scale? [09:21] Adam: Not totally sure-- lots of possibilities of different architectures. [09:21] Dan: We can narrow the search space. [09:21] Chris: Eddie and Chris have looked at Sling-- not sure where the architectural discussion would lead. [09:22] Eddie: Mea culpa! [09:22] Eddie: Didn't push the fcrepo-store-util work Eddie has done. [09:23] Eddie: Still work in progress-- should be out there to get people thinking. [09:24] fcrepo-store-util [09:24] is a framework and CLU to do migrations between stores. [09:24] Eddie: How to get this kind of CLU to chain? [09:25] Eddie: http://jewelcli.lexicalscope.com/ [09:26] Adam: Might take a look at Apache Karaf shell. [09:26] Eddie: Lots of use cases, including changing datastream type, Rebecca's checksum problem, lots of others. [09:27] Adam: Change the nature of the Fedora CLI? [09:27] Eddie: Maybe, not sure. [09:27] https://github.com/cwilper/fcrepo-store/blob/master/fcrepo-store-util/src/main/conf/commands/copy.xml [09:27] [ fcrepo-store/fcrepo-store-util/src/main/conf/commands/copy.xml at master · cwilper/fcrepo-store · GitHub ] - https://github.com/cwilper/fcrepo-store/blob/master/fcrepo-store-util/src/main/conf/commands/copy.xml [09:27] Eddie: Not enough motivation to do anything major. [09:28] Adam: (Shares Eddie's ennui.) [09:29] Chris: This is not fully tested, but it shows migration between Akubra stores. [09:29] Chris: The code uses Spring configuration, and this could be a general pattern. [09:30] Chris: This is also using one approach to pipelining, using filters that are tied together to process. [09:31] Chris: Filters can either transform or accept/reject objects. [09:31] http://cwilper.github.com/ttff/apidocs/com/github/cwilper/ttff/Filters.html [09:32] Chris: This is a filters utility class, which can AND/OR/NOT together filters. [09:32] Chris: Was originally written for CloudSync, but the challenge was configuring the workflow. [09:33] Dan: How about Apache Camel? [09:33] Chris: Have taken a look. That's one reason CHris didn't take the filters too far. [09:33] Chris: Lots of alternatives out there. [09:34] morning all. No mic, but on the call. [09:34] Hi, Ben! [09:34] Bye, Eddie! [09:34] Adam/Eddie: Lots of people have been using Camel in the larger community. [09:35] Bill Parod (NW), Chris Beer (WGBH), Adam, Scott PRater, others. [09:37] Chris: How about some Camel examples? [09:38] Adam: Can look at Camel website for good examples, and we can get some from the community (Bill P. and Adam, for example). [09:38] Adam: Camel is very strong, very widely used, great DSL, good place to build business logic for pipelining or message-passing workflows. [09:39] Chris: The fcrepo-store-util goes pretty much right against the filesystem. There are other ways to do it. [09:39] Chris: Please take a look at the utility-- let's get talking about 4.x. [09:40] Chris: Questions raised as to what is a core Fedora component. We've tried to keep that small, as an architectural principle. [09:40] Chris: So where does a utility like this live? In the same category as Fedora client? [09:41] Chris: Until we're ready to recommend and support something, it isn't a core component. [09:41] Adam: How about the GSearch model? [09:41] Chris: Prefers even more independence. [09:41] Chris: Good for these ancillary tools to grow and live outside the core committers group. [09:42] Chris: GSearch has a tighter relationship to Fedora's core. [09:42] Chris: Thinking of projects that are as "far way" as Blacklight. [09:43] Dan: Think of things like tool plugins, maintained by other folks. [09:43] Dan: things that have a cross-presence in DuraSpace/Fedora, ala Atlassian's Confluence/Jira plugin libraries. [09:44] Dan: They have a clear line as to what they support, but they still foster community and development outside that line. [09:44] Dan: Just drawing that line. [09:45] Adam: Not as easy for Fedora to offer plugin space. [09:45] Dan: Two good places to offer plugin-- at the code level, at the level of shareable objects... [09:45] and at the level of the framework. [09:45] Bye, Frank! [09:46] Chris: Attlassian's plugin ecosystem is quite rich. Just having a central registry would be a good deal. [09:46] Chris: How much are we (committers) maintaining collectively? [09:47] Chris: We should be empowered to support good things that don't come out of the core development effort. [09:47] Chris: E.g. MediaShelf-sourced client. [09:47] Chris: Is there an official line we can draw? [09:48] Adam: Doesn't think we have one now, but we could create one. [09:48] Dan: Atlassian is a good role model. [09:49] Dan: People do their own thing, and if it is badly needed, then you might want to take it on. [09:49] Adam: We want to be precise about "badly needed". [09:50] Chris: This utility is aimed at Rebecca's problem, but if it becomes the official upgrade path for legacy storage -> Akubra, then other committers should be involved. [09:50] Dan: Chris has laid out some of the guiding elements for the Fedora committers. [09:50] Dan: For 4.x and onwards, how do we make a vibrant community easy? [09:51] == elschlomo [~ruckus@HSI-KBW-046-005-164-187.hsi8.kabel-badenwuerttemberg.de] has quit [Quit: [BX] Automatically bored away] [09:51] Dan: Sometimes it's surprisingly easy to get mindshare by finishing out key extension points. [09:51] Dan: E.g. OSGi support, things that provide a framework for extensions. [09:52] Dan: Provide shareability for content modeling. [09:52] Adam: That's been on the horizon for a long time. [09:52] Dan: Hasn't been any motivation because the work has been mature enough. [09:52] has -> hasn't [09:53] Dan: This utility can be a point at which people can add on their own functionality. [09:53] Adam: So it's really important to have a good framework for utilities. [09:53] Chris: No urgent Jira issues. [09:54] Chris: Let's take a look at Jira. [09:54] https://jira.duraspace.org/browse/FCREPO-1057 [09:54] [ [#FCREPO-1057] ECM validation of MIMEType does not allow objects to specify charset if none is specified in the CModel - DuraSpace JIRA ] - https://jira.duraspace.org/browse/FCREPO-1057 [09:56] Chris: Some things need to be resolved, at least it should be specified down to inner syntax for MIMEtypes. [09:56] Chris: Let's open it. [09:57] Chris: Frank pointed out some cleanup in the use of collections that shold be done. [09:57] https://jira.duraspace.org/browse/FCREPO-1056 [09:57] [ [#FCREPO-1056] Unsafe usage of Lists in some classes - DuraSpace JIRA ] - https://jira.duraspace.org/browse/FCREPO-1056 [09:58] Adam: Just some pregenerics baggage. [09:58] Chris: Let's open it! [10:00] https://jira.duraspace.org/browse/FCREPO-1055 [10:00] [ [#FCREPO-1055] Shared resource for XML operations - DuraSpace JIRA ] - https://jira.duraspace.org/browse/FCREPO-1055 [10:01] what will be the impact on implementation time to use osgi there? [10:01] as opposed to just sharing a compiled xsl? [10:02] Adam: Steve and I had a long conversation about this-- good place to put the camel's nose under the tent for reworking resource provisioning for better architecture. [10:02] Adam: Could be a place to introduce OSGi micro-services. [10:02] Chris and Dan: Yeah, maybe... [10:03] Dan: Let's re-own the term "microservices". [10:03] Fin.
These logs were automatically created by DuraLogBot on irc.freenode.net using the Java IRC LogBot.