Log of the #duraspace-ff channel on chat.freenode.net

Using timezone: Eastern Standard Time
* eddies leaves00:58
* eddies joins
* eddies leaves
* eddies joins
* JasonDGI joins08:21
* anusha joins09:07
* jonathangee joins09:59
* jonathangee leaves
* jonathangee joins10:07
eddies, anusha are you all around?10:08
<anusha>im here10:09
<jonathangee>if eddies was around i was going to see if we could make the standup happen a little earlier today for purely selfish reasons10:10
<anusha>jonathangee: I am just creating users for the VMs. To start with shall I just give you, eddie, chris, jason and ben a access
<jonathangee>but it doesn't look like he's around
<anusha>:(10:11
<jonathangee>that sounds great
<eddies>jonathanfee, anusha: here now10:30
eep. jonathangee ;-)
anusha: also an account for adam10:31
and frank
<anusha>I've done Frank10:33
What's Adam's last name10:34
<eddies>soroka
* ajs6f joins10:57
<JasonDGI>anyone on the call?11:00
<eddies>anusha: are you joining?11:02
<anusha>https://github.com/futures/ff-jmeter-madness/blob/master/FuturesJmeterAggregationRead2.jmx11:16
<ajs6f>Is it
https://github.com/futures/ff-jmeter-madness/blob/master/FuturesJmeterAggregationRead1.jmx
?
Oh, okay.11:17
I'm confused. Looking at FuturesJmeterAggregationRead2.jmx, I don't see anyplace where the Lily records are being deleted at all..11:23
<JasonDGI>rise up against our opressive overlords!11:24
<ajs6f>I dunno. Eddie just isn't very good at opressing.
<JasonDGI>maybe we should pay for a class for him
<jonathangee>https://github.com/dataflow/databank-jmeter11:34
* ajs6f leaves11:38
* ajs6f1 joins
Back online
Am I the only on FreeConfCall?11:39
<JasonDGI>we're still here11:40
* fasseg joins
hola guys
<JasonDGI>sup?11:41
<anusha>is it an issue just for extending or for clients also11:42
<ajs6f1>Can any of you hear me?11:43
<fasseg>I added the 3 tests i ran on modeshape, fedora and lily to this page: https://wiki.duraspace.org/display/FF/FF+Candidate+Test+Harness11:44
<anusha>no
<ajs6f1>As Ben would say, "I LOOOOOVE FREE CONFERENCE CALL."11:45
<fasseg>anusha: can you tell me if i misconfigured databank somehow, to see the performance decay as seen here: https://docs.google.com/file/d/0B5nd_qlYdcqyS2haMlZLajRMc2M/edit ?
<JasonDGI>yes11:46
<ajs6f1>You did here that.
s/here/hear/
<anusha>fasseg: Were you trying to add all of the files into a single dataset?11:48
<eddies>ok, so adam & anusha will split off to sort out the jmeter read task
<fasseg>yepp
<eddies>and if we can get jason & frank to quickly connect on lily
and i'll float between either group as needed11:49
<fasseg>gotta be soon though, im getting the angry face from the gf ;)
<eddies>ok, we;ll keep it quick
<ajs6f1>Who actually wrote the JMeter read sections? Particularly the Fedora one, which is what we are going to look at first?
<anusha>I did
<eddies>anusha
<ajs6f1>Good.
So what exactly is going wrong.11:50
Are you getting HTTP errors, or Jmetere errors?
<anusha>I'm trying to get the line number
<ajs6f1>Cool.
I'm looking at it in Jmeter's editor, which may be helpful to you, too.
It's shows the hierarchy much more clearly.
<anusha>Ah yes, I've fired that up11:51
<ajs6f1>Jonathan— it's not this cold in PEI in the summer, is it? Cause I'm not coming OR if it is.
<anusha>Within fedora - in the second for each controller, it just loops through all the files and generates a 40411:52
<ajs6f1>Second run?
Second section?
<anusha>second section
<ajs6f1>That's the second ForEach contrller?
<anusha>correct. I have a feeling I got the BSF processor incorrectly configured for the read11:53
<ajs6f1>Okay, I'm looking at it now...
Wait— if it's the BeanShell that is doing the read, why is there a "Fedora Read Binary Resource Request" going on in the third ForEach?11:54
With description "One exercise of the Fedora REST API to read a datastream"...11:55
Is _that_ producing the 404s?
<anusha>No the beanshell is only used to generate the url or so I thought
You the FILE from the beanshell and this is then used in constructing the url
I don't understand how the nodename variable in the url is populated though11:56
<ajs6f1>Okay, so the "Fedora Read Binary Resource Request" is using the FILE Jmeter variable, which was produced by the BeanShell. In that case, why are they in two different ForEaches?
There's ForEach that is the third foreach we named above, then there's another inside it, with the read in it.11:57
<anusha>Cause we first want to finish all the creates - so loop through all the object and then do the reads
first for each for the objects. The for each within it i is for the datastreams within the object
<ajs6f1>Okay, so what does FILE mean— the DSID?11:58
<anusha>Correct
<ajs6f1>Okay, so what variable is the object PID?
<anusha>and nodename is the object id
<ajs6f1>Okay. Let me look at the path being used in the read.
<anusha>It's the same as the create11:59
<ajs6f1>Well, that can't be right.
If you want to read the content you wrote, the path has got to end with "content"12:00
Let me find the REST API doc for you.
But it still shouldn't be giving a 404.
<anusha>Well you should get the dsid profile if you dont end with content
<ajs6f1>Right. You shouldn't get a 4040.
Looks like all of this is logged. Can you show me a sample URL from the create datastream and a sample URL from the read datastream?12:03
<anusha>just a sec
<ajs6f1>Cool.
Actually, cold. Cold as all get out.
<cbeer>anusha++12:05
<anusha>http://localhost:8080/fedora/objects/test:1_1/datastreams/DS_433948.pdf?controlGroup=M for add
<cbeer>thanks for getting those VMs so fast
<ajs6f1>That looks totally kosher.12:06
<cbeer>i'll start installing all the base stuff we'll need (e.g. java.)
<anusha>http://localhost:8080/fedora/objects/test:1_652/datastreams/DS_597114.pdf for read12:07
<ajs6f1>Chris— remember to install BitTorrent so we can start using Oxford's gear to start sharing pirated movies.
<anusha>cbeer: :) I'll pass this onto our Sys-ad.
<ajs6f1>Yeah, that read URL looks good.
<anusha>Oh sorry, that was the first url which is a 200 OK12:08
<ajs6f1>Hm.
<cbeer>ajs6f1: can i just do openjdk7 (or, does fedora need 6?)
<anusha>The remaining urls are the ones generating a 404. Will copy one
<ajs6f1>Cool.
We need 6 for 3.6, I believe.
Our old version of ActiveMQ blows up on Java7.12:09
<anusha>http://localhost:8080/fedora/objects/test:1_652/datastreams/DS_433948.pdf
<cbeer>k. now i need to get my head thinking about debian-flavored linuxes..
<ajs6f1>https://jira.duraspace.org/browse/FCREPO-1171
<anusha>I got it, the PID is always test:1_652
<ajs6f1>?!?!?!
Okay, so the nodename isn't what we want it to be.12:10
And the thread ID may not be right.
Because both are used, right?
<anusha>ummm, not sure
<ajs6f1>"/${FEDORA_SERVER_APPLICATION_CONTEXT}/objects/${__log(test:${__threadNum}_${nodename},OUT,,Adding binary content )}/datastreams/DS_${__javaScript("${filename}".split("/").slice(-1)[0])}"
See "test:${__threadNum}_${nodename}"
Actaully, I don't get that at all. Why are we using the threadNum as part of the pid?12:11
<anusha>The threadname is being set in above - so I am guessing that would be correct
<ajs6f1>Okay. but the nodename isn't iterating as we desire...
<anusha>because we could run the test with 5 simultaneous threads, in which case the pid would be test:1, test:5...12:12
<ajs6f1>Okay. that makes sense.
I think.
But how are we iterating the nodename?
<anusha>I could not figure out where nodename is being set
<ajs6f1>Ah ha!
Perhaps we need to iterate nodename...
<anusha>but I don't see it being done for the POST12:13
<ajs6f1>It has to be iterated somewhere, right?12:14
<cbeer>anusha: i see you are using puppet, huh? i was thinking about being lazy and doing a bash script.. but would it be better if i did it in puppet?
<ajs6f1>Unless it is being set from the value of an iterated variabe.
<anusha>puppet is installed by my sys-ad to monitor (for nagios ?)12:15
<cbeer>ok, but you don
<anusha>No
<cbeer>k. i won't finish that sentence :)12:16
<ajs6f1>Where is nodename coming from?!
I don't see it appearing anywhere.
<anusha>ajs6f1: I agree. This is where my limited understanding breaks
<ajs6f1>until it is used.
<cbeer>ajs6f1: this is in the FOREACH?12:17
<ajs6f1>We can't figure out _where_ it is!
<anusha>ajs6f1: that's the same thing I have been wondering since this morning
<ajs6f1>It seems to be used to make requsts without ever having been declared!
<cbeer>which project are you looking at?
<ajs6f1>the link is above.
<anusha>Fedora test plan
<ajs6f1>number 2.
test plan number 2,
I mean.
nodename seems somehow to magically be iterated for posts.12:18
<anusha>https://github.com/futures/ff-jmeter-madness/blob/master/FuturesJmeterAggregationRead2.jmx
<ajs6f1>but not for reads.
<cbeer>thanks. let me pull and take a look.
<ajs6f1>https://raw.github.com/futures/ff-jmeter-madness/master/FuturesJmeterAggregationRead2.jmx
<anusha>ajs6f1: it wasn't just me going crazy :)
<ajs6f1>You can see it there. I saved it an opened it in Jmeter.
No. No, you are not crazy. You are just learning what it is like to be part of the Fedora project.12:19
<anusha>:)
<cbeer>which threadgroup are we looking at?
<anusha>I need to start liking beer
<cbeer>(sorry, haven't been paying attn)
<ajs6f1>Fedora Read Binary Resource Request12:20
is the rad.
s/rad/read/
<anusha>fedora threadgroup, 2nd top-level foreach
<ajs6f1>afk for one minute.
* ajs6f1 leaves
<cbeer>oh, that's in a different loop.12:21
<anusha>where?
<cbeer>the counter is set in the ForEach Controller / Fedora Delete Object Request Path
-- /${FEDORA_SERVER_APPLICATION_CONTEXT}/objects/${__log(test:${__threadNum}_${__counter(FALSE,nodename)},OUT,,Deleting node )}
i don't understand the jmeter counter scoping12:22
<anusha>Ah, now I understand the significance counter
So if I add counter to the read, do you think it will start from 1
you seem to have a better nose for this12:23
* ajs6f joins
<cbeer>i was the one who ruined ajs6f's initial code12:24
<ajs6f>I really think we shold refactor this to use the Jmeter primitives.
<anusha>cbeer: In that case I also need to do a read of the object
<cbeer>i think we should refactor it to not use jmeter :P
<ajs6f>Can we not use one thread group at a t ime, and write/write datastreams/read all inside that?12:25
<anusha>Jmeter primitives - don't know that either.
That what the read1.jmx file does. They don't want to do a read after write as they think the result will be cached and not accurate12:26
<ajs6f>Oh, okay. I can buy that.
<anusha>and read1.jmx works fine.
<ajs6f>Anyway, we ahev to find nodename.
Chris— feel free to write test scripts. I have no problem with that.12:27
<anusha>We can't add counter in the binary read as it will count too fast (more datastreams than objects)
<ajs6f>If the pid is what's wrong, we need to undestand how the pid is being developed.
<anusha>We will need another read before the 2nd for-each counting the objects12:28
as it is done for create
<ajs6f>So we're missing an iterator.
That's what I thouht.12:29
<anusha>that's what I think
<ajs6f>But how is nodename being iterated in the ceate cycle?
<anusha>with the count keyword
<cbeer>counter.
<ajs6f>Okay, so why can't we use that again?
You can have multiple counters.
<cbeer>because jmeter.
<ajs6f>No, you can have multiple counters.12:30
<cbeer>true
just change he counter name
<ajs6f>Right. Save it to a variable.
<anusha>ah, okay.
<ajs6f>http://jmeter.apache.org/usermanual/functions.html#__counter
You can call it with a variabe name into which it saves the count.
Wait...12:31
Will that work?
We want a counter inside a thread, right?
<anusha>This whole thing is within the thread12:32
the threadnum is define in the first line of the thread group12:33
we just ned a counter for the objects and the idea proposed by the two of you, seems to make sense to me
<ajs6f>Okay, then that shod work.12:34
<anusha>I will modify this and test it again. I feel like I have enough pointers to continue now.
<ajs6f>nodename can be iterated by the thread-local counter. dsid is iterated in FILE, and I don't understand how that happens, but it seems to happen correctly, so who cars for now.
* JasonDGI leaves12:35
<ajs6f>Go for it, Anusha!
<anusha>Thanks a lot both of you :)
<ajs6f>No problem. Give a yell if you have more trouble.12:36
<anusha>sure12:37
<cbeer>ajs6f: yes, i was thinking jdbc.. but maybe udp multicast works in the oxford environment12:38
<ajs6f>If we open the right ports, wy wouldn't it?
<cbeer>ajs6f: who knows what weird things the VM host may do
or an upstream router12:39
<ajs6f>Mm. Good point.
I wonder if Infinispan config supports variable substitution.
<cbeer>anusha: maybe i'll set up the modeshape cluster and find out, huh?
<ajs6f>We could leave it parameterized?
<anusha>cbeer: Thanks :)
<cbeer>i'll still write the docs, of course12:40
<anusha>I am way out fo my depth here, so have no clue.
If you need ports opened, it can be done. Just remeber it will take a few hours sometimes for the info to propagate
Once you think the docs are ready, I am willing to try12:41
<cbeer>cool. i'm just going to deploy modeshape to add 6 (or, 5) nodes. i figure we should just deploy stuff everywhere and turn it on and off as needed12:42
<ajs6f>We should be able to add/remove nodes dynamically. In fact, that coud be read as part of our use cases.
Chris--12:44
https://issues.jboss.org/browse/ISPN-1228
and
https://github.com/infinispan/infinispan/blob/master/core/src/test/resources/configs/string-property-replaced.xml
show the use of System properties to configure a cluster.
<cbeer>ajs6f: yes, jgroups is totally happy with that
<ajs6f>So we should be able to try various JGroups setups pretty easily.12:45
<cbeer>cool
<ajs6f>Or, yeah, as you say, do it in the th JGroups config itself.
<cbeer>ajs6f: nice. so, those java properties are based in as -Dblah.blah.blah=stuff?
(or other random places, i'm sure)
<ajs6f>Yeah, The test case shows stuff like:12:46
System.setProperty("test.property.asyncListenerMaxThreads","2");
They look like bog-standard Java System props to me.
<cbeer>excellent.12:47
<ajs6f>Also, just so you know, ModeShape's config itself does something similar:
https://docs.jboss.org/author/display/MODE/ModeShape+in+Java+applications#ModeShapeinJavaapplications-Variables
That could be pretty handy, actually, if we go with ModeShape.12:48
Die, FEDORA_HOME, die!
* JasonDGI joins12:54
<eddies>sigh. no new news on the brew sponsorship front. barmintor: i've gone ahead and contacted barrier as well12:55
<barmintor>eddies: regardless, we need to make sure to exchange the address I should pick you up at on the 5th. And I need to make arrangements for a car.12:56
<anusha>It's 6 here and I'll be kicked out of the building in 30 minutes. So if I am not done with the test updates, I'll go home and complete them or push the changes in latest by tomorrow.12:58
<eddies>barmintor: sent you the address12:59
<barmintor>eddies: thanks
<eddies>ajs6f: do you have anything to push to ff-lily-prototype?13:04
<ajs6f>Yep, I was just about to.13:05
Got simple object creation working.
Not even looking at datastreams yet.
But it might be useful for someone else to start with.
<cbeer>jonathangee: is there something for php like rvm/virtualenv?
<ajs6f>Pushed.13:07
<eddies>ajs6f: cool. i wanted barmintor to try his hand at looking at working with lily, but i'd rather he get a little bit of a jumpstart with what you've done if at all possible
<ajs6f>It's waiting for him.
I wish it wasn't as good a programmer as Ben, but I suppose he would have seen it sooner or later.
<barmintor>ah, my Sunday will be glorious.
<ajs6f>Your SUnday will be cold.13:08
<eddies>so this is building a war, which should be deployed to same container where lily will already be running?
<ajs6f>Chris— how did you set up that "notify Git pushes into chat" thing you did for ff-modeshape prototype.13:09
<cbeer>ajs6f: go to the repository
Settings
something about commit hooks
one of those is IRC
<ajs6f>Eddie— no. It needs to be configured with the Zookeeper connect string. The config is Spring and it's only a few stanzas.
Remember, Lily uses it's own container/platform.13:10
That Kouri (sp?) thing.
Thanks, Chris.
<eddies>ah. ajs6f, can you just a few more getting-started sort of notes to the readme, then?13:12
<ajs6f>Just did.
<eddies>mindreader =)
<ajs6f>Hopefully, we will soon see a push notice from Github
I guess not.13:13
<eddies>stepping away for a few mins
it's not active
now it is
<ajs6f>What's not active?13:14
<eddies>(ajs6f: you needed to check "active" for the service hook)
<ajs6f>See, I know ALL ABOUT Github.
<cbeer>they really should prompt you.. "hey, i saw you filled all this stuff out... do you want to use this hook?"13:16
* github-ff joins
[ff-modeshape-prototype] none pushed 5 new commits to master: http://git.io/aeJ7gA
ff-modeshape-prototype/master c04de5d ajs6f: ModeShape REST API no up alongside Fedora
ff-modeshape-prototype/master 9cde475 ajs6f: Added some minor logging
ff-modeshape-prototype/master c61bc4d Chris Beer: Merge pull request #21 from futures/ModeShapeRest...
* github-ff leaves
* github-ff joins13:17
[ff-lily-prototype] eddies pushed 1 new commit to master: http://git.io/hWrqBA
ff-lily-prototype/master e327417 Edwin Shin: Update README...
* github-ff leaves
<eddies>it works now ;-)
sorry for the spam. the test hooks apparently lag a little and i clicked a few times in my impatience before just finally pushing a commit13:18
ok, now really steppng away for a few mins
* anusha leaves13:19
* github-ff joins13:20
[ff-lily-prototype] ajs6f pushed 3 new commits to master: http://git.io/S3REJw
ff-lily-prototype/master a68931c ajs6f: Betterifiying README
ff-lily-prototype/master 29e325c ajs6f: More better README
ff-lily-prototype/master c219213 ajs6f: More better README
* github-ff leaves
* github-ff joins13:22
[ff-lily-prototype] ajs6f pushed 1 new commit to master: http://git.io/9C9mRw
ff-lily-prototype/master 51e2482 ajs6f: Last change to README today
* github-ff leaves
* github-ff joins13:31
[ff-lily-prototype] ajs6f pushed 1 new commit to master: http://git.io/IagDww
ff-lily-prototype/master 35ca356 ajs6f: Minor cleanup
* github-ff leaves
<JasonDGI>im getting a series of errors with jmeter13:54
Problem in BSF script org.apache.bsf.BSFException: JavaScript Error: Internal Error: org.mozilla.javascript.EcmaError: TypeError: Cannot read property "length" from null
<eddies>jasondgi: in the lily threadgroup?14:06
<JasonDGI>i think it in the
aAssemble list of fixture objects
its pointing to fixtures/objects and i think its trying to find a manifest but there isnt one14:07
<eddies>the manifests are in each bag under objects
e.g. 000/, 001/ or random/
<JasonDGI>yeah i saw that, and it looks like my jmeter doesnt know that14:08
<eddies>what dir did you launch jmeter from?
i think the test is all relative paths
assuming you're in ff-jmeter-madness
<JasonDGI>im on my mac - lauched it with finder
that could be it
<eddies>fasseg: thanks for https://wiki.duraspace.org/display/FF/FF+Candidate+Test+Harness
(new test results)14:09
<JasonDGI>so the fixture objects are the things inside objects/001? or the things inside objects?14:10
I hard coded the path in the config, so now im getting the correct path to objects:14:11
Loading fixture objects from ~/virtualcode/repositories/ff-jmeter-madness/fixtures/objects
but it looks like there should be something appended after that
<eddies>in fedora-speak: each bag == an object. the contents of each bag correspond to datastreams14:12
*the binaries in each bag correspond to datastreams (i.e. not the manifests and bagit stuff)14:13
<JasonDGI>right, that makes perfect sense
<eddies>(i think…i don't think we're creating datastreams for the bagit-specific files)
<cbeer>no, we're not. we're reading the bagit manifest to know what datastreams to add14:15
<barmintor>Guh, I don't want to learn about ruby 1.9 character converters14:17
<JasonDGI>boo ruby
<barmintor>Err… I also don't want to parse TIFF file headers in PHP :)14:18
* jonathangee leaves14:40
<JasonDGI>so my jmeter is definitely scanning for *file*s inside the *objects* directory (which contains only folders) :(14:45
<cbeer>hm. these oxford VMs only have 2G of RAM. i hope that's enough14:49
<ajs6f>to work at all, or to give us a fiar idea of our tests?
s/fiar/fair14:50
* ajs6f leaves14:51
* ajs6f joins14:53
<JasonDGI>i put a ~ in my path and forgot that it doesnt expand outside of bash14:55
derp
running the test now
<cbeer>ok, i think multicast is disabled on the oxford VMs15:00
i'll do JDBC i guess. that seemed nice an easy15:02
<ajs6f>Why not also ask Anusha to get multicat turned on? We might want it later— let's ask now.15:03
* github-ff joins15:12
[ff-lily-prototype] ajs6f pushed 1 new commit to master: http://git.io/6sVSXw
ff-lily-prototype/master 20f9f77 ajs6f: Cleaned Spring slightly
* github-ff leaves
* github-ff joins15:13
[ff-lily-prototype] ajs6f pushed 1 new commit to master: http://git.io/8Uc2PA
ff-lily-prototype/master 4a848d5 ajs6f: Keep README up to date
* github-ff leaves
* ajs6f leaves15:22
* ajs6f joins15:26
* github-ff joins15:34
[ff-lily-prototype] ajs6f pushed 1 new commit to master: http://git.io/WC2yWQ
ff-lily-prototype/master c8e08e1 ajs6f: Changed type creation fom programmatic to via JSON schema load.
* github-ff leaves
* fasseg leaves15:42
* github-ff joins
[ff-lily-prototype] ajs6f pushed 1 new commit to master: http://git.io/jGRxKw
ff-lily-prototype/master f1492cd ajs6f: Changed README to use Markdown
* github-ff leaves
* fasseg joins15:45
jason: did you get multinode lily running?15:46
<JasonDGI>no, i was collecting single node jmeter results for most of the afternoon
but im posting a graph now - it looks terrible :)15:47
<fasseg>heh bad results are good in a way :-)
although the alternatives seem to slim down drastically15:48
<JasonDGI>i got a throughput of ~10 instead of ~700, could that just mean my machine is slow?
and it was steady around 10 for the entire time
<fasseg>n order of mag slower? youre not running a toaster are you?:-)15:49
<JasonDGI>its a microwave oven, could that be the problem :D
<fasseg>heheh
<JasonDGI>macbook pro, 8G Core i7
<ajs6f>Jason— are you using their tes script to lanch?15:50
Or building up an instance yourself?
<JasonDGI>which test script? im just launching jmeter and opening the jmx
<fasseg>ell thats quite a strong laptop..
bin/launch-lilz
<JasonDGI>no, not using that
<ajs6f>I mean to launch Lily
<JasonDGI>./launch-lily?
<fasseg>bin/launch-lily-test i meant15:51
<JasonDGI>oh yeah thats the one
<ajs6f>That's what I used.
<fasseg>and me for the single node test too
<ajs6f>It may not give us very ffair results, but ...
<JasonDGI>im gonna try giving my vm more memory maye, i think it has 2G now
<ajs6f>if it's that bloddy hard to build up a proper install...
<fasseg>..right15:52
<JasonDGI>the oxford VMS will help loads
<ajs6f>It's a shame. I was really looking forward to spending weeks on Maven trickery to get the thing to deploy without setting my hair on fire.
<fasseg>i was running the tests on 8gb
* jonathangee joins
<JasonDGI>ill push my VM as high as it can go (probably 6)15:53
ok, im off for the day, i will be on skype if its an emergency15:54
<ajs6f>It's starting to snow again here, so I'm heading out. I'll try to check in on IRC later.
@Ben: I left the Lily prototype ready to roll. The Spring could use tightening, and I'll try to do that later today if I get a chance.
But it's basically ready for building out the Fedora-in-Lily schema and the API (using JAX-RS).
* JasonDGI leaves
<fasseg>can modeshape be run in a servlet container btw? or will it need jboss or
<ajs6f>We run ModeShape in Jetty now.15:55
<fasseg>will it need an app server?
ah nice
just did the jboss install until now
<ajs6f>It can, in theory, run in a non-EE Java environment, but its power is then limited.
I look forward to turning it loose inside an OSGi framework.15:56
<fasseg>power in terms of usability or funtionality?
<ajs6f>The latter. No transaction manager = no JTA-style transactions.
I don't think it's a cocern at all.
It just speaks to how low its requirements are.
<fasseg>oh but you could still add that to e.g. tomcat by hand..
<ajs6f>Oh, yeah. That's what we did for Jetty, except we didn't have to. {grin}15:57
<fasseg>so no jboss needed i hear, i like that:-)
<ajs6f>IOW, we just added the xaction manager to our own dependencies and rolled on.
The JEtty is running without one.
No, no JBoss.
In and of itself, it does not require a container at all.15:58
But we're doing JAX-RS, and for all practical purposes, that means the servlet API.
<fasseg>so you could just inlude a modeshape in a plain java app?
<ajs6f>that's one place where ModeShape clearly is winning over the alternatives— deployment architecture.
Yes, you could.
People do.15:59
<fasseg>thats nice...
and im guessing theres a good client/server api for java
<ajs6f>Yeah, I like no-committment relationships. {grin}
There's the JCR API itself, plus MOdeSHape's extensions.
E.g...
JCR doesn't define the cool "sequncing" functionality that MOdeShape offers.16:00
<fasseg>oh right jcr impl i keepforgetting
<ajs6f>ModeSHape invented that themselve.
So they extend the API in a sensible way to give you programmatic acess.
Yeah, JCR indeed.
<fasseg>that's a big plus
<ajs6f>Nodes and properties and relationships.16:01
<fasseg>although lily has a nice java api too it seems
<ajs6f>Looks a lot like what Feora might look like if we invented it today.
Oh wait, we _are_!
Yes, Lily's Java API is very nice, from what I've seen.
It has a different slant.
From the Apache H* products it seems to inherit a far more serious approach to "model first, so we can make assumptions and you get speed".16:02
They really discourage schema-les operation.
<fasseg>yeah i only took a quick look myself but it seemed concise and easy to use
<ajs6f>The hardest part getting the prototype up was just defining the shape of records.
But that's just a learning curve.
It is conscise. There are some weird, non-Java-ish things.16:03
E.g.
<fasseg>just like lily were you have to create a model before ingesting anything?
<ajs6f>You have to tell itwhat fields you want when you retireve a record.
No, no! I was _talking_ about Lily.
<fasseg>ah ok i thought still about modeshape
<ajs6f>ModeShape let's ramble freely in schemaless meadows.
<fasseg>and they had that concept too16:04
<ajs6f>No.
MOdeShape is fine with free-form data.
<fasseg>but i guess i forgot:its jcr
<ajs6f>right on.
You can bring schemata and validation when you want, but you don't have to.
Chris and I have started to bring it into play.
Doing teh modeling of Fedora in JCR.
<fasseg>cumbersome?16:05
<ajs6f>In the case of Lily, we'll _have_ to do that before we can do much else.
JCR schema? No. ModeShape gives you a GUI editor.
<cbeer>ok, futures1-6 should all have java6, tomcat6, rbenv, virtualenv and whatever else i thought of
<fasseg>yes we would need a 'general' cintent model
<ajs6f>And the concepts are simple.
cinitent?
<fasseg>nice!!!!11
<ajs6f>Thanks, Chris!16:06
Don't forget Usenet. That's very important.
<cbeer>i have trouble sshing between futures boxes, but that may just be something i'm doing wrong
bbc iplayer.
<fasseg>*content...typing on tablet atm which sucks...
<ajs6f>Ah!
Yes, we'll finally have to answer the question we haven't asked for a decade:
We know what Sandy and Carl thought the Fedora model was… but what has it become?16:07
<fasseg>bent in weird shapes i guess?
<cbeer>i'm tempted to create a shared account user on the futures machines
so we can worry less about file permissions, etc16:08
<ajs6f>How about a group "ffutures"
that's the POSIX way.
not shared accounts...
ay, for all you oldtimers...
My wife and I are going to have a beer with Ross Wayland and his wife this Sunday.
Any mesages?
<barmintor>Hold on, let me go grep /fcrepo16:09
<ajs6f>Remember, this guy helped write software that lasted under strain for a decade. Run by academics. (!) Send some love.16:10
<barmintor>Tell him thanks for inventing my miserable job :P16:12
I kid, I kid
<ajs6f>I' will, and I can tell you now, he will smile gently and enigmatically.
Remember, it was _his_ miserable job first.16:13
<barmintor>heheh
<ajs6f>I'm heading out for real. See y'all.
<fasseg>bye!16:14
* ajs6f leaves16:15
* jonathangee leaves16:18
<barmintor>AAAAAARGH16:37
cbeer: I am still finding obscure bugs from the change to String#[]
* jonathangee joins17:04
* jonathangee_ joins17:06
* jonathangee__ joins17:07
* jonathangee leaves17:08
* jonathangee_ leaves17:10
* jonathangee leaves17:12
* jonathangee joins17:27
* barmintor leaves17:28
* jonathangee leaves17:33
* ajs6f joins17:59
* ajs6f leaves18:02
* elschlomo joins19:02
* fasseg leaves