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

Using timezone: Eastern Standard Time
* dchandekstark joins01:56
* dchandekstark leaves02:00
* thomz joins02:30
* tjohnson leaves02:57
* dchandekstark joins
* dchandekstark leaves03:03
* dchandekstark joins03:59
* dchandekstark leaves04:04
* dchandekstark joins06:01
* dchandekstark leaves06:06
* dchandekstark joins08:02
* dchandekstark leaves08:07
* ajs6f joins08:29
* peichman joins08:37
* peichman leaves08:41
* bseeger joins
* coblej joins08:53
* thomz leaves08:55
* dwilcox joins08:56
<coblej>we're running fcrepo-webapp-plus-4.5.1.war on our development server ... was planning to drop in fcrepo-webapp-plus-4.6.0-RC-3.war and run our test suite ... anyone think of anything in particular I need to do in terms of upgrade steps ... or should I just give a try?08:58
* dchandekstark joins09:07
<ajs6f>coblej: That should work, to my understanding. If not, we'd like to know. {grin}
<ruebot>ajs6f: whikloj and i might have found something with the RC w/r/t auth/acl -- waiting for acoburn to arrive in irc09:19
<coblej>ajs6f: thanks. deploying it now. will let you know
<ajs6f>ruebot: Dammit. Using which artifact?
<ruebot>ajs6f: ...and since he is now on vacation, it is my duty to pursue it further. he was having trouble with RC-2 and RC-3. I'm seeing it on RC-3.
ajs6f: steps; 1) clone fcrepo4-vagrant 2) checkout 4.6.0-RC branch 3) edit config to use RC-3 4) vagrant up 5) clone fcrepo4-tests 6) copy config.tmpl to config 7) run authz_tests.sh09:20
<ajs6f>ruebot: I don't know what fcrepo4-tests or authz_tests.sh are.09:21
<ruebot>ajs6f: https://gist.github.com/ruebot/ccb8d449c20ee2ca551080f8b6c553a8
<ajs6f>ruebot: That does not help.
<ruebot>ajs6f: https://github.com/fcrepo4-labs/fcrepo4-tests09:22
ajs6f: https://github.com/fcrepo4-labs/fcrepo4-tests/blob/master/authz_tests.sh
<ajs6f>ruebot: So some kind of error that indicates that someone was able to get to something they shouldn't have been able to see
<ruebot>ajs6f: whikloj though the ACL might need to be updated, but wasn't 100% sure. maybe acoburn would know?
ajs6f: correct!09:23
<ajs6f>ruebot: Maybe— I have absolutely no idea. I have never been inside the ACL stuff at all .Actually, I don't know much about the authZ stuff.
ruebot: We definitely want acoburn in on this.
<ruebot>ajs6f: the `testuser` https://github.com/fcrepo4-exts/fcrepo4-vagrant#fedora-configuration -- should get a 403, but has access.
ajs6f: cool. i'll keep an eye for him.09:24
<ajs6f>ruebot: I'm trying to think of issue that might be connected.
<ajs6f>ruebot: https://jira.duraspace.org/browse/FCREPO-1895 ?09:26
ruebot: https://jira.duraspace.org/browse/FCREPO-2065 ?
* acoburn joins09:30
* esm_ joins09:40
* bseeger leaves09:49
* bseeger joins09:51
* esm_ leaves09:53
<ruebot>acoburn: if/when you have a moment, can you check out the irclog right before you logged in? whikloj and i are trying to figure out if we found an issue with authz/acls.09:55
<acoburn>ruebot: sure09:56
* dchandekstark leaves
<acoburn>ruebot: I haven't checked authz on any of the release candidates (sounds like I need to)09:58
* dchandekstark joins09:59
acoburn: whikloj (who is on vacation now) wanted me to double-check with you and make sure the acl he has in fcrepo-tests is correct. he thought it might actually be the test resources/fixtures.10:01
<acoburn>ruebot: ok, I'll check that
* esm_ joins10:14
<coblej>ajs6f: our (Hydra) app's test suite ran with no failures on our development server using 4.6.0-RC3 webapp plus war file as a drop-in replacement for 4.5.1 webapp plus war file10:22
coblej: Thanks, that is excellent news!
coblej: Would you like to add a note to https://wiki.duraspace.org/display/FF/Release+Testing+-+4.6.0 ? There's a place near the top for notes about Hydra and Islandora.10:23
<coblej>ajs6f: yes, will do10:24
coblej: Hang on, there's a new section escowles seems to have added just for this kind of stuff: https://wiki.duraspace.org/display/FF/Release+Testing+-+4.6.0#ReleaseTesting-4.6.0-HydraTests
<coblej>ajs6f: ok10:26
* coblej leaves10:46
* dwilcox leaves10:48
* acoburn1 joins10:54
* acoburn leaves10:55
* dwilcox joins10:58
<acoburn1>ruebot:ajs6f: whikloj is right — there seems to be something wrong with authZ in -RC311:26
acoburn1: Do you know what?
<acoburn1>ajs6f: No, only that it isn't working right
<ajs6f>acoburn1: Do we know what isn't working right?
<acoburn1>ajs6f: I am suspicious of that commit from last month related to the header delegate
ajs6f: it seems to just let everyone through11:28
<ajs6f>acoburn1: ^^^^ I had two suspicions and that was one.
<ruebot>acoburn1, ajs6f: it was in RC-2 as well. not sure if he tested RC-1
acoburn1++ #thank you!
<acoburn1>ajs6f: I'm going to try taking the current rc3 branch and remove that commit to see what happens11:29
<ajs6f>acoburn1:ruebot: How hard is it to run these tests? Can we back up to before that commit and run them and then right after and run them? \
acoburn1: Yeah.
<ruebot>ajs6f: it depends on fcrepo4-vagrant, which pulls down artifacts. so, i would need an artifact. or, i could create them all on the vagrant machine, and test again. it'll be a bit though for me. i'm going to take off soon, and return right before the import/export call.11:30
<ajs6f>ruebot: Okay, sounds like acoburn1 has an approach.
<ruebot>ajs6f: cool11:31
<acoburn1>ruebot: I'm building fcrepo now w/o that commit
whikloj++ # for catching it originally
<awoods>ajs6f: I have seen a couple of times your mention of the testing period being two weeks. As a point of clarification, the process states it being three weeks. https://wiki.duraspace.org/display/FF/Policy+-+Release+Candidates11:36
<ajs6f>awoods: Hm, I don't remember that. Fine with me. We've lost a week anyway for bugs and we're going to lose another. ^^^11:37
<awoods>ajs6f: I just wanted to make sure the messaging and expectations were consistent.11:38
<ajs6f>awoods: This is Fedora, so they are not, and they never have been. The best we can hope for is to keep expectations as low as possible.11:39
<acoburn1>ruebot: I'm getting the same (error) result with that commit reverted. I need to verify that I'm not doing something wrong11:48
ajs6f:ruebot: ok, I think this may actually be a configuration error11:59
ajs6f:ruebot: I'm going to try again with RC-3
* ajs6f crosses fingers
<ajs6f>awoods: See my note about the additions you just made to the Im/Ex page. What you did is entirely premature.
<ruebot>ajs6f: that was actually following up on last week's call12:02
<awoods>ajs6f: we are collecting requirements... not committing to implementing them.
ruebot: exactly
<ajs6f>ruebot: That was nothing like my interpretation of last week's call. I didn't hear anyone asking for DPN integrations.
<ruebot>ajs6f: i believe awoods is breaking up the first use case into individual requirements12:03
* dchandekstark leaves
<ajs6f>awoods:ruebot: And I think he has done a misleading job of it. The list in that use case is a set of plausible examples, that's all. It should _not_ be broken out at this stage.12:04
<ruebot>ajs6f: are you talking about the edit that breaks it up into sections? external systems, general, round-tripping?12:05
* dchandekstark joins
<ajs6f>ruebot: No, the long list of completely system-specific requirements a la "Support import to Fedora in a BagIt format from LOCKSS, where the source of the original bag was Fedora' or "Support export from Fedora in a BagIt format that is suitable for ingest into DPN".
<awoods>ajs6f: one of the objectives of today's call is to clarify the requirements for inclusion or not.12:06
<ajs6f>ruebot:awoods: I don't even want to waste time getting agreement that no one wants or is going to pay for "ingest into DPN".
<awoods>ajs6f: there are certainly other requirements on that list that will get a line through them.
<ruebot>ajs6f: yeah, ok. then that make sense from the discussion around use case #1, that it be broken up into granular requirements.12:07
<ajs6f>awoods: Yes, there are. And that will take time as well. Your additions are going to use up time we could be using otherwise.
ruebot: Yes, it certainly must. This is too early and awoods did a bad job of it.
Throwing a bunch of stuff on the table and then declaring that we're all jointly responsible for taking off what we don't want is a TERRIBLE way to try to get to a working future.12:08
Especially in an environment of constrained resourcing.
<awoods>ajs6f: I would agree if the bunch of stuff had not come from the previous discussions.
<ajs6f>awoods: It didn't. No one asked for a pipeline from Fedora to DPN.12:09
<acoburn1>ruebot:ajs6f: tests passed on RC3
<awoods>ajs6f: to be clear, is your issue with the DPN requirements?
<ajs6f>acoburn1: Thank goodness!
<acoburn1>ruebot: the test suite has certain expectations about usernames
ajs6f: AND people need to manually change the default repo configuration in order to make it work12:10
<ajs6f>awoods: With all of them. Take them off. We can put back those for which an actual institution is willing to commit to the extent of being prepared to do acceptance testing.
<acoburn1>ajs6f: that needs to be changed — it should _just_ work
<ajs6f>acoburn1: Hm, mos def. Do you know what to change?
<ruebot>acoburn1: so, it's something in https://github.com/fcrepo4-labs/fcrepo4-tests or unit tests?12:11
<acoburn1>ajs6f: this line: https://github.com/fcrepo4/fcrepo4/blob/master/fcrepo-configs/src/main/resources/config/servlet-auth/repository.json#L2412:12
<ajs6f>ruebot:acoburn1: Sounds like there is all of a weakness in the tests, a problem in the default config, and a missing piece in the ITs.
acoburn1: Wait, I thought you said it was a problem with user names?
awoods: I will help you out by removing those spurious requirements.
<acoburn1>ajs6f: needs to be in the repository.json, but if you're using the default postgres config (as I am locally), you instead get the Bypass authentication class12:13
ajs6f:ruebot: the tests should verify that AuthZ is enabled in the first place
ajs6f:ruebot: because they don't at present12:14
ajs6f:ruebot: they also assume that you have a user account named 'adminuser', despite what you include in the config file
ajs6f:ruebot: once I sorted those two things out, the tests passed12:15
<ajs6f>acoburn1: So that's a commit right there, eh?
<acoburn1>ajs6f:ruebot: yeah, but I'm not sure what the best approach is
<ajs6f>acoburn1: Meaning release-wise, or code-wise?12:16
<acoburn1>ajs6f: code-wise
ajs6f: it's basically a question of whether we include all the various permutations in fcrepo-configs
ajs6f: that seems like a bad approach; or at least an unsustainable approach
<ajs6f>acoburn1: Is it a combinatorial thing, where you select from PSQL/MySQL/LevelDB/etc and auth1/auth2/auth3 and it all starts to get out of hand?12:17
<acoburn1>ajs6f: people should be overriding repository.json anyway and storing it outside of the classpath
ajs6f: yes
ajs6f: at present, there's only auth1 (no auth) and auth2 (some sort of auth)
ajs6f: so it's not a terrible combinatorial thing, but I still don't like it12:18
<ajs6f>acoburn1: Urb, this starts to take us back down the garden path to FEDORA_HOME of yore. Ideally, we alway wanted to get to the beautiful world of repository-hosted config, but it's always been too hard.
<awoods>ruebot: are you ok with ajs6f's removal of the requirements: https://wiki.duraspace.org/pages/diffpagesbyversion.action?pageId=77433711&selectedPageVersions=35&selectedPageVersions=36
<ajs6f>ruebot:awoods: Note that I did _not_ remove the requirement, I removed the long and inappropriately precise list of external systems and left that TBD.12:19
<ruebot>awoods, ajs6f: i'm fine with it. we can flesh out the individual ones if the group thinks we should. gordian knot + sword?12:22
<awoods>thanks, ruebot.
<ajs6f>ruebot:awoods: We absolutely will flesh them out, because we willl have to. The questions are: when? and how? (by taking things away or by adding them)12:23
* dwilcox leaves
<ruebot>awoods: you're not going to be on the call, right?
<awoods>ruebot: I will actually.12:24
* dwilcox joins
<ajs6f>acoburn1: I don't think we can solve the config problem in generality during the release testing. Do you think the corrections you have in hand are enough.
<acoburn1>ajs6f: I am satisfied with them. I am very wary of providing lots and lots of "default" configurations. All of that belongs in documentation
<ruebot>awoods: oh, good!
<acoburn1>ajs6f: which is to say: there is no need for a PR to the RC branch
<ajs6f>acoburn1: okay, let's get that work moved on. A commit for master alone, then?12:26
<acoburn1>ajs6f: just _really_ good documentation
ajs6f: I don't even know if a commit to master is needed
<ajs6f>acoburn!: if there's one thing we do well, it's this: we are good at _knowing_ that we need really good documentation.
<acoburn1>ajs6f: the commit would involve (more or less) double the directories here: https://github.com/fcrepo4/fcrepo4/tree/master/fcrepo-configs/src/main/resources/config12:27
<ruebot>acoburn1, ajs6f: i can test out that documentation on vagrant + tests later on if you would like. i'd like to get that authz test passing... and i imagine whikloj would as well.
* dchandekstark leaves
<ajs6f>ruebot: I'm guessing whikloj is enjoying a tall cold drink right now and laughing at the thought of Fedora.
<ruebot>ajs6f: i hope he is!12:28
<acoburn1>ruebot: do you need instructions on getting that test to work?
<ajs6f>ruebot: I think it would sound something like… https://www.youtube.com/watch?v=MMjQXmJRiG0
<ruebot>acoburn1: please!12:29
ajs6f: no acoustic versions of YES!12:30
<acoburn1>ruebot: you want them as a PR to the README?
* ruebot puts on roundabout
acoburn1: yeah, that would be great!
<acoburn1>ruebot: the main issue is that the username "adminuser" is hard-coded into the test12:32
ruebot: maybe some judicious use of `sed` will work?12:33
<ruebot>acoburn1: in https://github.com/fcrepo4-labs/fcrepo4-tests/blob/master/authz_tests.sh ?12:34
<acoburn1>ruebot: https://github.com/fcrepo4-labs/fcrepo4-tests/blob/master/resources/authorization.sparql#L812:35
<ajs6f>awoods: we have a fix into 4.6.0 for FCREPO-2090. Why did you change the fix version?
<acoburn1>ruebot: that needs to match the value in config:AUTH2_USER
<ruebot>acoburn1: oh!12:36
<acoburn1>ruebot: right, I just need to remember the syntax for sed…
<ruebot>acoburn1: sed find and replace? sed -i 's/something/something else/g' files.file
<ajs6f>acoburn!: np. sed uses regexps. that should give you a clear, well-defined, and unambiguous syntax.12:38
<acoburn1>ruebot: it's probably best not to change that file, though; maybe pass the output of sed directly into the curl command
ruebot: change curl to use "-d @-"12:39
* esm_ leaves12:40
<ruebot>acoburn1: gotta run. back in a couple hours.12:41
* esm_ joins12:43
* bseeger leaves12:46
* acoburn1 leaves13:22
* acoburn joins
* bseeger joins13:35
* esm_ leaves13:42
* esm_ joins13:43
* mikeAtUVa joins13:55
* esm_ leaves13:59
* esm_ joins
* dchandekstark joins14:39
* dchandekstark leaves14:40
* dchandekstark joins14:48
* atempleton joins14:58
<ruebot>Import/Export call starting up now: https://wiki.duraspace.org/display/FF/2016-07-29+-+Import+-+Export+Planning+Meeting14:59
* ajs6f is here
<bseeger>*is here*
<ajs6f>That is one FINE lookin' agenda.15:03
* mikeAtUVa leaves
<ajs6f>I don't think we had actually done 5, even before ddavis.15:05
"Sufficient time", tho.15:06
<ajs6f>We solve that problem by periodically introducing a new, incompatible backend.
Preventing any repository from lasting long enough to show that problem.
dchandekstark: What does 2077 have to do with that?
<dchandekstark>ajs6f: well, partially15:08
* apb18 joins
<dchandekstark>not dependent on an *external* index
e.g., like solr15:09
<ajs6f>dchandekstark: Still not seeing it, but it's cool, let's not worry about it.
<dchandekstark>ajs6f: cool
ajs6f: 2077 isn’t about recovery obviously15:10
it’s friday :015:11
* ajs6f cannot wait for it to be not during the week15:12
* mikeAtUVa joins15:17
* tjohnson joins15:24
* apb18_ joins
<ruebot>ajs6f: yeeeeeeeeeeeeah15:25
* apb18 leaves15:26
* apb18__ joins15:27
* apb18_ leaves15:29
* apb18_ joins
<dchandekstark>0.5M objects/day?!15:30
<ajs6f>dchandekstark: Research data management is a different ball game.15:31
* apb18__ leaves
<dchandekstark>~6 objects/sec15:32
<ajs6f>dchandekstark: two words, "camera traps"
* apb18_ leaves15:34
* apb18 joins15:35
<mikeAtUVa>can people use their triplestore to identify the subset of their repo they want to export rather than us building all that into the export function?15:44
<apb18>could we amend #2 to say binaries *and their associated RDF description*15:45
<esm_>#9 and #10 might be api-x use cases15:46
* apb18_ joins15:50
<ajs6f>sounds good15:52
* apb18 leaves
<ajs6f>I will help mikeAtUVa hold his stake.15:53
* apb18 joins15:54
* apb18_ leaves
<escowles>i'm more available the week of Aug 8th, but can participate at some level the week of the 15th too15:57
* apb18 leaves16:01
<ajs6f>FOLLOW UP EMAIL!16:02
* atempleton leaves16:03
* escowles leaves16:05
<ajs6f>ruebot: _Possibly_, just _possibly_ I could help with some dev for Im/Ex for server-side stuff. But we'll see what's needed.16:14
* dwilcox leaves16:15
<ajs6f>afk for the duration. g'day, mates.16:21
* ajs6f leaves
* esm_ leaves16:46
* bseeger leaves
* acoburn leaves17:10
* dchandekstark leaves17:14
* peichman joins17:34
* peichman leaves17:47
* dchandekstark joins19:47
* dchandekstark leaves19:58
* dchandekstark joins20:56
* dchandekstark leaves
* mikeAtUVa leaves21:48
* mikeAtUVa joins22:10
* mikeAtUVa leaves22:15
* f4jenkins joins23:13

Generated by Sualtam