11:35:39 [DIGlogger]
DIGlogger (~dig-logge@groups.csail.mit.edu) has joined #dig
11:35:39 [verne.freenode.net]
topic is: Decentralized Information Group @ MIT http://dig.csail.mit.edu/
11:35:39 [verne.freenode.net]
Users on #dig: DIGlogger RalphS_ kennyluck melvster drrho timbl amy presbrey Yudai_ ericP gbot46 sandro nunnun
12:01:15 [ericP]
presbrey, test_parser_exception worky
12:01:26 [ericP]
i didn't figure out how to test the string contents of the error
13:01:37 [lkagal]
lkagal (~lkagal@pool-96-237-240-136.bstnma.fios.verizon.net) has joined #dig
13:59:20 [kennyluck]
kennyluck has quit (Quit: kennyluck)
14:14:57 [oshani]
oshani (~oshani@c-71-233-151-72.hsd1.ma.comcast.net) has joined #dig
14:46:32 [nunnun]
nunnun is now known as nunnun_away
14:52:49 [oshani]
oshani has quit (Quit: Mama nidi!)
15:17:53 [kennyluck]
kennyluck (~kennyluck@114-25-246-70.dynamic.hinet.net) has joined #dig
15:25:19 [nunnun_away]
nunnun_away is now known as nunnun
15:31:21 [lkagal]
lkagal has quit (Quit: lkagal)
15:32:12 [knappy]
knappy (~haoqi@31-35-54.wireless.csail.mit.edu) has joined #dig
15:58:47 [presbrey]
i'll look at it soon
15:59:08 [presbrey]
i put up my idea for how to modify directory indexes
15:59:38 [presbrey]
the magic of server-side includes and javascript
15:59:41 [presbrey]
http://tabulator.org/wiki/
16:10:00 [melvster]
presbrey: got word back from my provider, they cant do a wildcard CNAME from their front end, but ive asked if they can do it on their side ... they'll get back
16:10:27 [melvster]
should be fine, but ive mapped over wiki. and sparql. so far
16:13:50 [lkagal]
lkagal (~lkagal@30-6-179.wireless.csail.mit.edu) has joined #dig
16:19:29 [timbl_]
timbl_ (~timbl@pool-96-237-236-72.bstnma.fios.verizon.net) has joined #dig
16:20:23 [timbl_]
timbl_ has quit (Client Quit)
16:23:53 [timbl]
timbl has quit (Ping timeout: 265 seconds)
16:33:38 [oshani]
oshani (~oshani@c-71-233-151-72.hsd1.ma.comcast.net) has joined #dig
17:04:30 [timbl]
timbl (~timbl@31-35-252.wireless.csail.mit.edu) has joined #dig
17:45:47 [knappy]
knappy has quit (Quit: knappy)
17:46:16 [ericP]
presbrey, would you like to make http://swobjects.svn.sourceforge.net/viewvc/swobjects/branches/sparql11/swig/python/t_SWObjects.py?revision=1205&view=markup#l226 test the error string?
18:02:36 [oshani]
oshani has quit (Quit: Mama nidi!)
18:02:48 [presbrey]
yes. is that the latest swobjects installed?
18:05:04 [oshani]
oshani (~oshani@c-71-233-151-72.hsd1.ma.comcast.net) has joined #dig
18:07:09 [amy]
where can I grab the latest tabulator?
18:21:41 [timbl]
1. Install mercurial if you don't have it
18:21:45 [timbl]
2. see http://dig.csail.mit.edu/2005/ajar/ajaw/Developer.html
18:31:14 [amy]
ok
18:36:24 [timbl]
timbl has quit (Quit: timbl)
18:37:23 [oshani]
oshani has quit (Quit: Mama nidi!)
19:00:00 [oshani]
oshani (~oshani@31-34-137.wireless.csail.mit.edu) has joined #dig
19:09:03 [timbl]
timbl (~timbl@75-149-137-201-Connecticut.hfc.comcastbusiness.net) has joined #dig
19:11:40 [presbrey]
ericP, interesting but I get something completely different looking
19:11:50 [presbrey]
wiki]# PYTHONPATH=/usr/local/src/swobjects/branches/sparql11/swig/python/ ./test.py insert-fail.sparql
19:11:50 [presbrey]
Traceback (most recent call last):
19:11:51 [presbrey]
File "./test.py", line 11, in <module>
19:11:51 [presbrey]
print DefaultGraph().sparql(file(sys.argv[1]).read())
19:11:51 [presbrey]
File "/srv/dig/www/2010/wiki/swobjects.py", line 26, in sparql
19:11:51 [presbrey]
_S.parse(_SWObjects.IStreamContext(query, _SWObjects.StreamContextIstream.STRING))
19:11:52 [presbrey]
File "/usr/local/src/swobjects/branches/sparql11/swig/python/SWObjects.py", line 4176, in parse
19:11:54 [presbrey]
def parse(self, *args): return _SWObjects.SPARQLfedDriver_parse(self, *args)
19:11:56 [presbrey]
RuntimeError: Unknown exception
19:12:23 [presbrey]
at least it doesnt segfault?
19:15:12 [ericP]
success!
19:16:00 [ericP]
looks like SWObjects.i:26
19:16:07 [ericP]
s/26/27/
19:19:10 [amy]
ericP when you get a chance can you look at my tabulator set up please?
19:19:50 [ericP]
amy, i probably won't be much help, but i can look in an hour
19:19:57 [amy]
thanks v much
19:27:40 [presbrey]
I think we really want it to say
19:27:52 [presbrey]
ParseException: stdin:1.8-2.0: syntax error, unexpected end of file, expecting IT_GRAPH
19:29:20 [presbrey]
then, raise HTTPError(400, (ParseException)e.message)
19:30:29 [presbrey]
it would be nice to know whats wrong with the query more than 'bad request'
19:30:58 [oshani]
amy, I can help with your Tabulator set up
19:31:07 [amy]
hi Oshani, thanks! there no rush
19:31:56 [presbrey]
ericP, how did you get access to runtimeerror anyway?
19:43:30 [timbl]
timbl has quit (Quit: timbl)
19:43:50 [ericP]
presbrey, http://swobjects.svn.sourceforge.net/viewvc/swobjects/branches/sparql11/swig/SWObjects.i?revision=1205&view=markup#l19
19:45:33 [ericP]
can you extract the error from http://swobjects.svn.sourceforge.net/viewvc/swobjects/branches/sparql11/swig/python/t_SWObjects.py?revision=1205&view=markup#l226 ?
19:58:35 [amy]
thanks so much Oshani!
20:07:21 [knappy]
knappy (~haoqi@dhcp-18-111-36-78.dyn.mit.edu) has joined #dig
20:29:53 [melvster]
re: what things could be used with a datawiki ... maybe a semantic 'inbox' ?
20:54:00 [RalphS_]
RalphS_ has quit (Quit: leaving ...)
21:07:51 [timbl]
timbl (~timbl@31-35-252.wireless.csail.mit.edu) has joined #dig
21:09:00 [presbrey]
hey timbl
21:09:21 [presbrey]
about data wiki ACL
21:09:58 [presbrey]
there needs to be a better way to get at the default ACL document for a URI
21:11:34 [melvster]
i was just wondering about this
21:11:50 [melvster]
will every file have it's own ACL document?
21:11:59 [melvster]
or one per directory?
21:12:06 [presbrey]
when it has none, I think its very confusing for users to know what is there that they can't see and/or what to put there, what happens if there is 1 insert to the ACL
21:12:27 [presbrey]
there is always some base ACL document
21:13:09 [presbrey]
even if its generated in code by the values of Require, AuthWebIDAuthoritative, etc.
21:13:35 [timbl]
In principle the server can pretend that it has many fiels and actually have one, or vice-versa
21:13:57 [timbl]
One meta file per resource seeems to make sense.
21:14:11 [presbrey]
I would do it on the enclosure
21:14:22 [presbrey]
the folder, directory
21:14:46 [presbrey]
save some processing on directory listings
21:14:59 [presbrey]
and get 'default permissions' for free
21:16:37 [timbl]
Well, when a file is craeted maybe immediately a .meta/<filename>.meta file should be made to store the owner.
21:16:47 [timbl]
You can't make a file without authenticing.
21:17:12 [presbrey]
have you seen my new Apache+ACL idea on data wiki?
21:17:13 [timbl]
Then you always allow the owner of a file with no ACL to add an ACL
21:17:28 [timbl]
which idea?
21:17:31 [presbrey]
http://dig.xvm.mit.edu/wiki/
21:17:45 [timbl]
putting ACL editing into the directory indexes?
21:18:13 [presbrey]
yes, it all happens with Javascript
21:18:36 [presbrey]
I would like to do 1 query to find out all the ACLs for those documents at substitute them in the (ACL for...)
21:18:36 [timbl]
I think that is a great idea, allows people to set stuff up even though tyey don't have tabulatr. They will need a webid though
21:18:40 [presbrey]
also note the 'Use SSL' link at the bottom
21:19:01 [presbrey]
I would credit the idea to your ,access
21:19:06 [presbrey]
but start simpler
21:19:10 [timbl]
I see a page with things like "(ACL for http://dig.xvm.mit.edu/wiki/tabulator/)"
21:19:20 [presbrey]
yes stuff needs to go in (ACL for...)
21:19:26 [presbrey]
temporary placeholders
21:19:41 [presbrey]
test that your WebID is recognized there as well
21:19:48 [presbrey]
when you hit 'Use SSL'
21:56:10 [presbrey]
... even though they don't have tabulator
21:57:03 [presbrey]
... or if they can't have tabulator
21:57:48 [presbrey]
so I can adjust my ACLs on my phone now
21:58:33 [presbrey]
definitely better to keep the layers/prereqs thinner than not
21:59:50 [presbrey]
"how can I SPARUL the FOAF on my server?" 1) install mercurial (?) 2) checkout tabulator refactor branch!!!?
22:00:27 [presbrey]
vs. curl -v -H Content-type:sparql-query
22:00:56 [presbrey]
how do I edit the ACL? oh its right there in my directory listing cool!
22:02:08 [presbrey]
what does tabulator do now if the .meta.n3 referenced in the ACL Link: header does not exist?
22:03:56 [presbrey]
ps, I'd like people to use data wiki and SPARQL in IE as well, not just my phone
22:05:06 [presbrey]
I still have a problem parsing WebIDs when there are 2 subjAltURIs, URI+email is common
22:05:16 [presbrey]
ack too much work! help! abort!
22:15:41 [timbl]
you mean parsing the cert?
22:16:49 [timbl]
A large part of the appeal of the foaf_ssl web id stuff is that the dealing with the cert is a fixed done thing.
22:26:51 [webr3]
webr3 (~nathan@host86-147-36-79.range86-147.btcentralplus.com) has joined #dig
22:29:01 [presbrey]
sure your cert works doesnt it?
22:29:06 [presbrey]
there are a few outliers
22:29:20 [presbrey]
one of which I forgot and have on this PC
22:31:26 [timbl]
My cert worked the other day it seems
22:33:53 [timbl]
presbrey, when i lookat that page i don't see a way of changingt acls
22:34:33 [presbrey]
there are no acls
22:35:24 [timbl]
I see things like "(ACL for https://dig.xvm.mit.edu/wiki/tabulator/tracker/)"
22:35:27 [presbrey]
we need default ACL
22:35:56 [timbl]
The default default ACL has to be no access except the craetor
22:36:01 [timbl]
creatot
22:36:05 [timbl]
nsbdGFLEVW BDSj
22:36:08 [timbl]
creator.
22:36:31 [timbl]
Remember .. the things works this way.
22:36:44 [timbl]
You can set a default ACL for new files at the directory
22:36:52 [presbrey]
so what about a new folder with nothing in it
22:36:59 [presbrey]
there is no creator info
22:37:07 [presbrey]
no acl meta
22:37:09 [presbrey]
no n3
22:37:12 [timbl]
I think we need to have a way of making the ACLs propagate
22:37:32 [timbl]
If there is no mete, the the server bahves as though there was a default meta
22:37:41 [presbrey]
what is the default meta?
22:37:54 [timbl]
in the directory
22:38:02 [presbrey]
there is no meta in the directory
22:38:45 [timbl]
every time you make a new directory you either propagate the ACL or you don't I guess it has to be flag
22:38:55 [presbrey]
I am fine with propagating ACLs
22:39:05 [presbrey]
a new directory with no meta uses the parent directory meta
22:39:13 [timbl]
suggest you propagate it when firs you need it
22:39:19 [presbrey]
/wiki/ has no parent directory with meta!
22:39:23 [timbl]
so you go up the chain but not every acces.
22:39:35 [presbrey]
there is no meta at / to propagate to /wiki/
22:39:37 [timbl]
you then store the inherited ACL.
22:39:55 [presbrey]
inherited from what? Apache?
22:40:06 [timbl]
Inherited from higher up the tree
22:40:23 [timbl]
so in /wiki/ we put a sort of acl thate says
22:40:45 [presbrey]
I'm thinking about a new user's 1st time install... what is the default ACL in the root folder?
22:41:43 [presbrey]
its fine for everything under to inherit/propagate from that one default
22:41:46 [timbl]
The default ACK shoudl allow the user who installed it
22:41:48 [timbl]
access.
22:41:56 [presbrey]
but there needs to be something default, something easy for new users/installs
22:42:11 [timbl]
You should never have default dsecurity off
22:42:38 [timbl]
Well, you can have for new installers *specifically of a wiki* a default of publc read write access whichpropagates
22:42:56 [timbl]
But this is more than a wiki now it is a general *space
22:42:59 [presbrey]
there are countless packages that ship saying 'Allow from 127.0.0.1'
22:43:14 [timbl]
You are saying that is bad?
22:43:34 [presbrey]
no I'm saying lots of things ship with all users read/write/execute but from 127.0.0.1
22:43:44 [timbl]
I see.
22:43:48 [presbrey]
default directory security off from 127
22:44:22 [presbrey]
the term 'wiki' to me defaults to 'Public' because I mediawiki ships with public editing
22:44:24 [timbl]
For my file system I think I prefer not any other users on the same machine
22:44:36 [presbrey]
wiki means collaborate doesn't it?
22:44:44 [timbl]
wiki means quick
22:44:49 [presbrey]
oh ok
22:45:19 [presbrey]
so can we say by default all WebID users can read and write
22:45:24 [presbrey]
all Web users can read
22:45:26 [timbl]
(in Hawaiian) But anyway The understanding is open write access IMHO on the net now for 'wiki'
22:45:34 [timbl]
so making a wiki is instaloption
22:45:45 [timbl]
one shoudl eb ablt t make a wiki any plave in the tree
22:46:06 [timbl]
I don't think this produc is a wiki
22:46:07 [presbrey]
sure we can have a few different example .meta.n3 configs for differently tiered privacy models
22:46:19 [timbl]
Once you have ACLs it is a space
22:46:34 [timbl]
workspace, placespace, wikisp[ace, wjhatever... a flexible space
22:46:35 [presbrey]
good so if there are no ACLs, it is still a wiki?
22:46:56 [timbl]
one of the buttons in your ACL list can be "make a wiki"
22:47:08 [timbl]
which just creates a directory with the right ACLs.
22:47:40 [presbrey]
do you have time to do ACL stuff today?
22:47:54 [timbl]
I think it intersting to have a bunch of applications, which you can craete on the web page
22:48:36 [presbrey]
I am about to finish working here... I'll come over to csail to chat for a few minutes if you have some
22:48:45 [timbl]
Craete a shared software project with a text wiki and data wiki with two issue trackers, action items and a chared calendar.
22:48:52 [timbl]
Ok
22:48:59 [timbl]
sure
22:49:01 [presbrey]
ok
22:49:17 [presbrey]
it would be good to sync on all fronts
22:51:55 [melvster]
looks like we're fundamentally still webizing the unix file system ...
22:54:15 [webr3]
just caught up - one thing I need to put forward quickly, is that I perhaps misunderstood the ACL side of things and had always thought of ACL's being specified on a per resource level not directory, in that 3000 files may have a single ACL (or should i say ACF); as in "this is my file/resource" "use this ACL for it" and where no ACL exists then just 'public' as in http://
22:55:58 [timbl]
webr3, yes i think that definitley ACLs are per-resource.
22:56:08 [webr3]
so 1-1
22:56:20 [timbl]
But in a directory you need to be able to sepcify whether people can craete things, and what the defualt ACL is foor a new craetion.
22:56:37 [timbl]
I do disagree though with the "no acl, public acess".
22:56:57 [timbl]
It is good security to have a specific reasont o let someone have access
22:57:03 [melvster]
very much like the unix file system and inodes
22:57:08 [timbl]
It also fits with the monotonic nature of RDF.
22:57:23 [timbl]
You can add info, allowing ore access, but you can't take it away
22:57:32 [timbl]
So noino has to mean no acecess
22:57:47 [webr3]
makes sense
22:57:47 [timbl]
no info has to mean no access
22:58:24 [timbl]
I think in the W3C acl system there is aboostrap rule that initially the person who craetes a new resource has the ability to set ACLs on t
22:58:35 [timbl]
so you have to capture the creator
22:58:38 [timbl]
webid
22:59:03 [webr3]
so lets say i have an ACL which allows access to friends of #me | I should be able to assoc many resources with this one ACL, or - should i be having that ACL cloned for each resource?
22:59:50 [timbl]
You could have the same ACL expresion in many acess control files.
23:00:14 [webr3]
but not all resources linked to the one ACF?
23:00:24 [webr3]
apologies, this is my greyest area
23:00:57 [timbl]
IO Guess you could have the same ACF URI form many resources.
23:01:08 [timbl]
That is a server implementation issue.
23:01:13 [timbl]
I htin
23:01:16 [timbl]
I think
23:03:34 [webr3]
cool - one other thing i need to clear up quickly..
23:05:10 [webr3]
spoke to the httpbis guys a couple of months ago about serving the 'same resource' via both HTTP and HTTPS; and was informed that when serving a file over http:// I could use the Content-Location: header to specify the https:// location, thus hinting/pointing at where to send update/delete requests to - good advice or?
23:20:40 [webr3]
final q, do you guys put the ACL in the .meta file? as I've been using cern metafiles and simply pointing to the ACF from the .meta file using the Link: header [i fear I've gone v wrong here]
23:28:01 [lkagal]
lkagal has quit (Quit: lkagal)
23:31:36 [melvster]
how actually do you allow public access in the ACL ... is Public a foaf:Group?