Log of the #fcrepo channel on chat.freenode.net

Using timezone: Eastern Standard Time
* jcoyne leaves00:48
* rbenji joins03:18
* rbenji leaves05:02
* ruebot_a1ay leaves06:34
* ruebot joins
* ruebot leaves
* ruebot joins
* dwilcox joins07:41
* jcoyne joins07:48
* acoburn joins08:51
* whikloj joins08:59
* peichman joins09:08
* ajs6f joins09:14
* diegopino joins09:25
* bseeger joins09:26
<diegopino>Hi awoods: do you have any thoughts on Garbage collectors (JVM options) and fedora 4?09:29
* github-ff joins
[fcrepo-transform] whikloj opened pull request #12: Custom node types are not referenceable (Release branch) (4.5.1-RC...FCREPO-1832) https://git.io/vVGK5
* github-ff leaves
<diegopino>trying to get deeper into this for deployment teaching and have been experimenting with -XX:+UseG1GC on larger heap instance09:30
<awoods>diegopino: only that they need to be tuned. I have not seen a set of options that fits everyone's needs.
* github-ff joins
[fcrepo-transform] whikloj closed pull request #10: Custom node types are not referenceable (master...FCREPO-1832) https://git.io/vVJem
* github-ff leaves
<awoods>diegopino: have you been running into issues?
<diegopino>Not at all, -XX:+UseG1GC runs like a charm09:31
<awoods>diegopino: what sort of scale testing have you done?
if any
<jcoyne>diegopino: Is G1 still experimental in Java 8?09:32
<diegopino>Not yet under fedora4, lacking data stil (will do this weekend) but fedora3 runs pretty stable with G1, over two wees with simulated traffic and 16GBytes heap09:33
It's stable and recommended in Java8 for bigger heaps
Just needs some tunning in MaxGCPauseMillis and initial heap ocupancy %09:34
over two weeks i meant on AWS
running 1.8.0_7409:35
<jcoyne>diegopino: Why do you use -XX then?
I thought that was for experimental options.
Oh, because Java09:36
"Options that are specified with -XX are not stable and are subject to change without notice." - http://www.oracle.com/technetwork/articles/java/vmoptions-jsp-140102.html09:37
<diegopino>jcoyne: right, as most garbage collection parameters. The will change/can. But i'm assuming i'm tunning those for my current java version.09:39
<jcoyne>Yeah, that's fine, I was just confused about how it's stable but you use unstable parameters to enable it.09:40
As someone who doesn't use java a lot, that seems messed up.
<diegopino>But Garbage collection has a big impact on containers like tomcat7 where so much is shared. also tried +UseStringDeduplicationJVM, until now seems to run fine.09:41
jcoyne: but you are right. I wonder if we should be always conservative about suggesting/or mentioning non stable options...
<jcoyne>diegopino: My criticism is targeted more at the JVM maintainers than anyone in the chanel.09:42
<diegopino>On the other side, some projects have moved out of the container deployment like Solr. So Solr 5.5 is a standalone jetty powered server now
<ajs6f>jcoyne: There is a difference between "garbage collection" and "the API for garbage collection".
<jcoyne>Anyway, Java 9 will make it the default.
<awoods>jcoyne: "it" being what?09:43
<diegopino>Some maybe old info on G1 if any one wants to read a bit about "it" http://www.infoq.com/articles/G1-One-Garbage-Collector-To-Rule-Them-All09:44
<jcoyne>awoods: G1GC09:46
<awoods>jcoyne: thanks
<jcoyne>awoods: http://openjdk.java.net/jeps/248
<diegopino>Ok, thanks. Will continue my research on this. refining slides.09:47
* mcritchlow joins09:49
<awoods>acoburn: would you please update this: https://wiki.duraspace.org/display/FF/Component+Compatibility+Matrix09:50
<acoburn>awoods: thanks for the reminder09:51
* mcritchlow_ joins09:52
* mcritchlow leaves09:53
<awoods>acoburn: In the past, I thought you had advocated away from LATEST in: "feature:repo-add mvn:org.fcrepo.camel/toolbox-features/LATEST/xml/features"09:54
<whikloj>acoburn/awoods: I was about to ask that same question w/r/t vagrant
<acoburn>awoods: basically, if the version matters (esp. in testing or automated deployments) then you shouldn't use LATEST09:55
awoods: if you just want the latest version, then just use LATEST
awoods: for anyone who's not building this locally, they should just use LATEST because that's the simplest approach09:56
<awoods>acoburn: in your release email, would "4.5.0" make sense?09:57
feature:repo-add mvn:org.fcrepo.camel/toolbox-features/4.5.0/xml/features
<acoburn>awoods: yes, exactly
<awoods>acoburn: Next release, it may make sense to provide that guidance in the email... I am not sure many are using it yet.09:58
<acoburn>awoods: I was trying to highlight the part about "toolbox-features" (as opposed to "fcrepo-camel-toolbox") and didn't really think about whether to use LATEST or 4.5.009:59
awoods: I agree that using 4.5.0 would probably make sense in the future
<awoods>acoburn: it is a minor point. +1 on getting the release out!
<acoburn>awoods: … or whatever the actual release number is!
* mcritchlow_ leaves10:25
* EdF joins10:27
* mcritchlow joins10:39
* mcritchlow leaves
* mcritchlow joins10:40
* diegopino leaves10:58
<awoods>https://wiki.duraspace.org/display/FF/2016-03-31+-+Fedora+Tech+Meeting11:00
* dshalvi joins
* ajs6f is here11:01
* ajs6f is going to have to leave a bit early again.
* escowles joins
<acoburn>*is here*11:02
<ajs6f>Kind of an earlier 90s indie sound.
<escowles>not sure what CORS support entails -- maybe acoburn or azaroth42 know?
dwilcox++11:03
<acoburn>escowles: easy, it's just about sending some additional headers in the response
<escowles>it was awoods!
<awoods>yes, I'm here
can you hear me?
<ajs6f>awoods: no.
* yinlin joins11:04
<whikloj>hahahaha11:06
<bseeger>Just say the word "Ben" again
<ajs6f>It's Fedo.
* ksclarke joins11:08
<whikloj>dead OS11:14
<ajs6f>If people complain, that's our chance to ask them to volunteer to do the test builds.11:15
<whikloj>ajs6f++11:16
acoburn++
<ajs6f>acoburn++
Can we not automate this via Jenkins?
<escowles>ajs6f: we'd still need someone with a windows box who could set it up, maintain it, etc.11:18
<ajs6f>escowles: Yes, but that's something we should o but once.
should do
Does that even need to be Java?11:20
Bye bye awwods11:21
I nominate escowles for our new tech lead.
* awoods sigh
<escowles>ajs6f: i bet 50 or so lines of bash+curl could do it
<ajs6f>escowles: Yeah, exactly.
:)
escowles: The meat is in the SPARQL. Executing that should be as light as possible.11:22
<whikloj>I'm kind of doing some of that now11:27
I'll stick them in a repo11:28
<escowles>whikloj++
<ajs6f>whikloj++
<whikloj>you haven't seen my bash scripting yet
<ajs6f>whikloj: We will bash it when the time comes.11:29
<bseeger>Testing is always an excellent place for non-developers to participate, so I'd keep it accessible, if possible.
<ajs6f>bseeger: What do you mean by accessible in that context?
<bseeger>don't make it hard for non-programmers to use… or just have curl/sparql commands listed clearly so if one wants to bypass the script and do it manually, they can still contribute11:30
just a thought
<ajs6f>bseeger: Sounds like you want something as close to directly human-readable as possible?
<bseeger>yes, ultimately, or just curl/sparql commands listed on the page if one wants to skip the script.11:32
<ajs6f>bseeger: I think we want to be agnostic against curl, but having the SPARQL laid out plainly is certainly the right thing.11:33
<whikloj>Ahhhhhh my kind of work11:36
<escowles>i think it could serve both purposes: a guide that people could follow in the HTML UI as a tutorial, and an automated script that would be a good TCK11:37
<ajs6f>Gots to go.11:40
bbl
* ajs6f leaves
<yinlin>+111:41
<whikloj>bseeger++
awoods: vagrant with webac+audit is populating Solr and Fuseki with content using the LATEST toolbox-features11:46
we are the less worse implementation, woohoo11:52
* mcritchlow leaves12:03
<acoburn>whikloj: the errors you're seeing with webapp-plus are due to a version mismatch here: https://git.io/vVZOf12:06
<whikloj>acoburn: ok, why does that not affect webac?
* mcritchlow joins12:07
<acoburn>whikloj: I have no idea
<whikloj>acoburn: fair answer :)
<awoods>whikloj: because the versions in the properties of the webapp-plus are not correct
<acoburn>awoods: is it customary that a release candidate artifact ends with -SNAPSHOT? I would expect -RC or something like that12:08
<awoods>whikloj: see: https://jira.duraspace.org/browse/FCREPO-1883
whikloj: specifically: https://github.com/fcrepo4-exts/fcrepo-webapp-plus/blob/master/pom.xml#L17-L1912:09
whikloj: those properties do not get updated automatically by the release machinery.12:10
whikloj: Since webapp-plus is coupled and release with each fcrepo4 release, I am not sure the properties are of use.
...and can be removed.
<whikloj>awoods: right that is what acoburn told me. Should the audit and transform version properties disappear and fcrepo.version be linked to ${version}
<awoods>whikloj++ acoburn++
* github-ff joins12:14
[fcrepo-transform] whikloj opened pull request #13: Use FedoraResource instead of Nodes (DEV branch) (DEV...FCREPO-1965) https://git.io/vVZsU
* github-ff leaves
* github-ff joins
[fcrepo-transform] whikloj closed pull request #11: FCREPO-1965 : Use FedoraResource instead of Nodes (master...FCREPO-1965) https://git.io/vVJKT
* github-ff leaves
* bseeger leaves12:17
* mcritchlow_ joins12:38
* mcritchlow leaves12:40
* yinlin leaves12:42
* mcritchlow_ leaves12:55
* mcritchlow joins12:59
* mcritchlow leaves
* mcritchlow joins13:00
<awoods>whikloj: are you working the rbacl/xacml issues?13:03
<whikloj>awoods: somewhat, also trying to clean up my shell scripting
<awoods>whikloj: I can go ahead and work this ticket as an RC bug: https://jira.duraspace.org/browse/FCREPO-1883 ...if that would help.13:04
<whikloj>awoods: removing the transform and audit properties and setting fcrepo.version = 4.5.1-SNAPSHOT changed the failure
<awoods>whikloj: changed it to a different failure?13:05
<whikloj>awoods: java.lang.NoSuchMethodException: class org.eclipse.jetty.security.ConstraintSecurityHandler.setLoginService(class org.eclipse.jetty.security.HashLoginService)
<awoods>whikloj: I'm on it. Focus on the scripts.13:06
<whikloj>awoods++
* bseeger joins13:08
<acoburn>awoods: I'm thinking this might help with those requests for an internal search mechanism: https://jira.duraspace.org/browse/FCREPO-196813:20
<awoods>acoburn: Are you suggesting the added context available to the LDPath transforms will make for richer index documents?13:22
<acoburn>awoods: yes13:23
awoods: we can put all the fedora: ns properties in as a matter of course
awoods: and adding the dc/skos/rdf properties would be nice too
<awoods>acoburn: do we have skos properties now?13:24
<acoburn>awoods: no, but if someone is using skos:prefLabel it would be nice to index it
awoods: I don't have a hard-and-fast rule about which properties to include/exclude, so I will appreciate feedback when the time comes13:25
<awoods>acoburn: the LDPath context is dictated by the Prefer header in the /fcr:transform request.13:27
<acoburn>awoods: really? I am not familiar with how Prefer headers work on the /fcr:transform endpoint13:28
<awoods>acoburn: Prefer headers work the same way they work on regular GET requests.13:29
<acoburn>awoods: ok, that I knew13:30
<awoods>acoburn: so in terms of which properties are available to the LDPath context, that is dictated by the value of the Prefer header.13:31
<acoburn>awoods: of course
awoods: but if, e.g., skos:prefLabel is never added to a resource, it can never be read by the LDPath program. But if it _is_ part of a resource representation, then it could be indexed13:32
<awoods>acoburn: Maybe I was over thinking your new ticket. Are you just suggesting a simple update to the "default transform": https://github.com/fcrepo4-exts/fcrepo-transform/blob/master/src/main/resources/ldpath/default/nt_base_ldpath_program.txt13:33
<acoburn>awoods: yep, that's all it amounts to — just adding a few more lines
<awoods>acoburn: sure... although we should also be encouraging people to create their own transforms.13:34
<acoburn>awoods: I agree — but for an "out of the box query interface" this would at least be something that could be minimally useful13:35
<awoods>acoburn++
<whikloj>acoburn: in the same vein, can you specify to the camel-toolbox which transform to use when indexing? If you built a new one?13:36
<awoods>whikloj: yes, see: https://github.com/fcrepo4-exts/fcrepo-camel-toolbox/tree/master/fcrepo-indexing-solr#configuration13:37
<acoburn>whikloj: you bet (and at runtime): https://git.io/vVZac
<whikloj>awoods++ / acoburn++
<escowles>acoburn: so you could use Prefer headers to include e.g. inbound references or children, and then use them in the transform?13:38
<acoburn>escowles: AFAIK, yes you can
<escowles>that could be very useful, and combined with the batch operations could cover a lot of what people want builtin search for13:39
<acoburn>that's exactly what I'm thinking. do you have opinions on what rdf predicates to include in the default transform?13:40
<escowles>i think the use case is efficiently discovering resources linked to and from a resource, so being able to list them with transform and fetch them in a batch should cover that
acoburn: i'm not sure — pcdm:hasMember, ore:aggregates, ore:proxyFor/ore:proxyIn, and iana:first/iana:last would cover the pcdm relationships, but i'm not sure what else people are using13:42
<acoburn>escowles: simply extending it beyond the current default (dc:title, fedora:created) will get us a long way toward a more usable search interface13:43
* bseeger leaves14:10
<whikloj>awoods: what is your feeling on the fixity test? That you can get a fixity result or do we care whether the test was good or bad?14:25
* peichman leaves14:26
* mcritchlow leaves14:29
* dshalvi leaves14:41
* bseeger joins14:54
<awoods>whikloj: are you asking if we need to perform a corruption fixity run?15:12
<whikloj>awoods: If you want to test fixity, is the test simply "Can I access http://foo/object/fcr:fixity" and get a 200 response. Or do we care whether it is calculating a fixity success or failure correctly.15:13
Not that I know how we would test the latter15:14
<awoods>whikloj: the latter would also be good: verifying the SHA1 and size: http://demo.fcrepo.org:8080/fcrepo/rest/24/0d/76/54/240d7654-9328-4021-a009-13a608f35596/fcr:fixity15:17
whikloj: that would just involve parsing the returned RDF body
premis: hasEventOutcome
<whikloj>awoods: yeah in bash!!
<awoods>premis: hasMessageDigest
premis: hasSize
whikloj: in the bash case... you can also leave that as a TODO15:18
<whikloj>woohoo
<awoods>whikloj: thanks for thinking of it.
<whikloj>awoods: depending how you run these tests, I could output the values and the BODY and you could compare manually (for now)
<awoods>whikloj: that is reasonable.15:19
* dwilcox leaves15:28
* bseeger leaves15:50
* acoburn leaves15:55
* dwilcox joins16:02
* dwilcox leaves16:09
* ksclarke leaves16:40
* mcritchlow joins16:50
* github-ff joins17:11
[fcrepo-transform] acoburn opened pull request #14: add fields to default transformation (DEV...fcrepo-1968) https://git.io/vVnGj
* github-ff leaves
* github-ff joins17:22
[fcrepo-mint] awoods opened pull request #3: Remove unnecessary properties from pom.xml (4.5.1-RC...fcrepo-1883) https://git.io/vVncI
* github-ff leaves
* github-ff joins
[fcrepo-transform] awoods opened pull request #15: Remove unnecessary properties from pom.xml (4.5.1-RC...fcrepo-1883) https://git.io/vVncO
* github-ff leaves
* github-ff joins17:23
[fcrepo-audit] awoods opened pull request #31: Remove unnecessary properties from pom.xml (4.5.1-RC...fcrepo-1883) https://git.io/vVncC
* github-ff leaves
* github-ff joins17:24
[fcrepo-webapp-plus] awoods opened pull request #36: Remove unnecessary properties from pom.xml (4.5.1-RC...fcrepo-1883) https://git.io/vVncE
* github-ff leaves
* peichman joins17:29
<awoods>whikloj: this is ready for you: https://jira.duraspace.org/browse/FCREPO-188317:30
<whikloj>awoods: ok, I'll get on it
* peichman leaves17:31
* dwilcox joins17:46
* dwilcox leaves17:51
* whikloj leaves17:58
* mcritchlow leaves18:55
* jcoyne leaves19:07
* dwilcox joins20:05
* whikloj joins20:10
* jcoyne joins
* github-ff joins20:19
[fcrepo-webapp-plus] whikloj pushed 2 new commits to 4.5.1-RC: https://git.io/vVnSn
fcrepo-webapp-plus/4.5.1-RC e9fcd0c Andrew Woods: Remove unnecessary properties from pom.xml...
fcrepo-webapp-plus/4.5.1-RC 0085576 Jared Whiklo: Merge pull request #36 from awoods/fcrepo-1883...
* github-ff leaves
* github-ff joins20:21
[fcrepo-audit] whikloj closed pull request #31: Remove unnecessary properties from pom.xml (4.5.1-RC...fcrepo-1883) https://git.io/vVncC
* github-ff leaves
* github-ff joins20:23
[fcrepo-transform] whikloj pushed 2 new commits to 4.5.1-RC: https://git.io/vVnSF
fcrepo-transform/4.5.1-RC 8e7c43f Andrew Woods: Remove unnecessary properties from pom.xml...
fcrepo-transform/4.5.1-RC 3e635d2 Jared Whiklo: Merge pull request #15 from awoods/fcrepo-1883...
* github-ff leaves
* github-ff joins20:24
[fcrepo-mint] whikloj pushed 2 new commits to 4.5.1-RC: https://git.io/vVnSj
fcrepo-mint/4.5.1-RC 8ab6fdb Andrew Woods: Remove unnecessary properties from pom.xml...
fcrepo-mint/4.5.1-RC b82347f Jared Whiklo: Merge pull request #3 from awoods/fcrepo-1883...
* github-ff leaves
* travis-ci joins20:25
fcrepo4-exts/fcrepo-transform#48 (4.5.1-RC - 3e635d2 : Jared Whiklo): The build passed.
Change view : https://github.com/fcrepo4-exts/fcrepo-transform/compare/6b38bacf8b00...3e635d2d4eb7
Build details : https://travis-ci.org/fcrepo4-exts/fcrepo-transform/builds/119958687
* travis-ci leaves
* travis-ci joins
fcrepo4-exts/fcrepo-audit#81 (4.5.1-RC - e67b398 : Jared Whiklo): The build passed.
Change view : https://github.com/fcrepo4-exts/fcrepo-audit/compare/ddb36844948e...e67b398f612d
Build details : https://travis-ci.org/fcrepo4-exts/fcrepo-audit/builds/119958498
* travis-ci leaves
* travis-ci joins20:26
fcrepo4-exts/fcrepo-webapp-plus#118 (4.5.1-RC - 0085576 : Jared Whiklo): The build passed.
Change view : https://github.com/fcrepo4-exts/fcrepo-webapp-plus/compare/2997018f0df9...00855766f409
Build details : https://travis-ci.org/fcrepo4-exts/fcrepo-webapp-plus/builds/119958373
* travis-ci leaves
* travis-ci joins20:28
fcrepo4-exts/fcrepo-mint#13 (4.5.1-RC - b82347f : Jared Whiklo): The build passed.
Change view : https://github.com/fcrepo4-exts/fcrepo-mint/compare/450ab06080ae...b82347f69f61
Build details : https://travis-ci.org/fcrepo4-exts/fcrepo-mint/builds/119958905
* travis-ci leaves
<awoods>whikloj++20:29
<whikloj>awoods: thanks, I have some scripts here https://github.com/whikloj/fedora4-tests20:30
very basic
<awoods>whikloj: that is great, I will send an email to escowles and barmintor with the update on where the release stands... including your scripts.20:31
<whikloj>awoods: ok, I'll add this to the ticket and see if bseeger can try them out
* dwilcox leaves20:34
* whikloj leaves20:35
* dwilcox joins20:37
* dwilcox leaves20:41
* awoods leaves22:54

Generated by Sualtam