Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2016-03-19

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:01 sm you're hitting a lot of strange link errors, have you tried rm -rf ./stack-work ~/.stack
00:02 fr33domlover oh not that again, it takes a lot of time and CPU sweat to build the 135 deps again :P
00:02 fr33domlover but i suppose i have no other option...
00:41 Big_G joined #darcs
00:45 owst joined #darcs
00:51 owst joined #darcs
02:47 ilbot3 joined #darcs
02:47 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.10.3 is out http://darcs.net/Releases/2.10
02:51 Riastradh joined #darcs
03:07 pointfree joined #darcs
03:11 Riastradh joined #darcs
03:21 pointfree It's getting awkward to integrate/link data across the multiple darcsden couchdb databases.
03:22 pointfree I wonder if it's better to have just one big couchdb database for everything on darcshub.
03:22 pointfree But that could have a performance impact :(
03:24 pointfree Maybe we need to be smarter about how the data is partioned among databases.
03:25 pointfree Things that will need to share things with each other go in the same database.
03:27 pointfree It has been said that "CouchDB is bad at everything except syncing."
03:28 pointfree Maybe we just need to embrace the replication.
03:40 sm pointfree: could we switch to persistent and/or postgres ?
03:40 sm or plain text files ? It's not a lot of data
04:01 pointfree sm: Yes, I think NoSQL isn't great for this application and isn't necessary for most applications, IMHO.
04:01 pointfree I've have also been thinking about using a relational database.
04:02 pointfree Or... haskell swish
04:03 pointfree (Semantic Web Inference Scripting in Haskell)
04:04 pointfree I don't know much about swish yet, but I do find it intriguing.
04:07 pointfree I think it would be cool to store darcshub (darcsden?) users in foaf files and would facilitate the decentralization goal of darcsden.
04:07 pointfree The decentralized social semantic web.
04:21 pointfree https://github.com/robstewart57/rdf4h looks more relevant and active.
05:40 pointfree We could just append an n-triple to the end of a file every time an event occurs. That would be our event feed as well.
08:27 castlelore joined #darcs
08:39 castlelore joined #darcs
09:57 owst joined #darcs
11:13 castlelore joined #darcs
11:13 castlelore joined #darcs
11:17 inhortte joined #darcs
12:34 Big_G joined #darcs
12:40 fr33domlover pointfree, I want to use semantic web too, but Haskell doesn't have a single triplestore package! I was going to make one (write my own triplestore DMBS) but then i stopped, maybe i'll try again some day
12:41 fr33domlover swish and rdf4h aren't enough, we need something that uses an actual graph/triple database
12:41 fr33domlover and wrap that with something like persistent and esqueleto, but for RDF and SPARQL
12:47 owst joined #darcs
13:18 owst joined #darcs
13:24 owst joined #darcs
13:58 owst joined #darcs
14:06 alexei joined #darcs
15:14 Riastradh joined #darcs
16:43 owst joined #darcs
17:18 pointfree fr33domlover, perhaps a persistent wrapper around 4store would be in order? https://web.archive.org/web/20160109085647/http://4store.org/
17:19 alexei joined #darcs
17:20 pointfree https://github.com/garlik/4store
17:20 fr33domlover pointfree, yes that would work. I wanted to go futher and implement the DB itself in Haskell, but it seems to require knowledge of DB internals and query plans etc.
17:21 fr33domlover so it will take me a while if I decide to try
17:22 fr33domlover for experiments, it's enough to wrap 'fgl' with an API or something :P
17:23 fr33domlover another "easy" but a bit limited approach is to wrap an SQL database, like Tracker does
17:24 pointfree I saw fgl. Is it persistent?
17:24 fr33domlover also possible to wrap an SQL database without generality limitations, but then you can't have transitive relation deduction (e.g. is aRb and bRc then aRc) which is needed for some applicatiobs
17:24 fr33domlover pointfree, i don't think so
17:24 fr33domlover it's just a graph library
17:25 fr33domlover it likely won't scale as a DB backend, but could work for small cases
17:25 fr33domlover somewhat like SQLite i suppose
17:25 fr33domlover fgl + acid-state
17:25 fr33domlover heh, worth a try
17:26 pointfree I like that idea.
17:28 pointfree http://stackoverflow.com/questions/8032731/haskell-data-structure-for-skos-semantic-web
17:29 pointfree Looks like someone is using fgl for semantic web applications.
17:31 fr33domlover nice
17:32 fr33domlover so we can already use 4store with hsparql, it seems
17:35 fr33domlover not that much work left then
17:38 fr33domlover but making something like persistent isn't trivial, if you want to consider uniqueness etc. You'd need to detect all the relations which declare constraints (bijection thisjection thatjection whatever) and use that, or let Persistent TH models declare the schema
17:38 fr33domlover definitely interesting!
17:39 pointfree Where did you find info about 4store+hsparql?
17:39 fr33domlover i'm using SQL for my current web app, and i'll probably stick with that for now because i want to get functionality first
17:39 fr33domlover pointfree, that link you pasted
17:39 fr33domlover you can access 4store as a sparwl endpoint probably
17:39 fr33domlover i.e. using HTTP requests
17:39 fr33domlover so you don't need 4store bindings or anything
17:40 pointfree Oh right
17:40 fr33domlover there's a SPARQL endpoint standard /RFC which defines these queries
17:40 fr33domlover unlike in SQL where connections are DB specific afaik
17:41 fr33domlover the weakness is that in a local DB you get overhead of TCP and HTTP
17:41 fr33domlover instead of using a unix socket
17:41 fr33domlover that's where some bindings to 4store APIs etc. could help
17:43 pointfree Like this? https://github.com/MadSciGuys/FourStore
17:44 pointfree "FourStore is a library for using 4store's efficient native binary protocol instead of the SPARQL protocol. This has the additional benefit of allowing query results to be read lazily (just like HDBC allows for SQL databases, or normal Haskell file IO)."
17:49 fr33domlover pointfree, sadly it's proprietary but assume we convince the author to relicense as free software -
17:49 fr33domlover yeah, it's cool
17:50 fr33domlover but even SPARQL is fine if you use it over unix socket and not HTTP
17:50 pointfree Doh. "All rights reserved."
17:53 fr33domlover ah i see
17:53 fr33domlover no big deal
17:53 fr33domlover it's a trivial package
17:53 fr33domlover it doesn't implement any protocol
17:53 fr33domlover it just runs a 4store process to pass the data efficiently
17:54 fr33domlover we could write our own from scratch, worst case, it's a small thing
17:59 pointfree Anything we could do better in our reimplementation of FourStore for DarcsDen?
18:09 pointfree 4store is GPLv3. This may "infect" FourStore with GPLv3 as well. "If the program dynamically links plug-ins, and they make function calls to each other and share data structures, we believe they form a single program, which must be treated as an extension of both the main program and the plug-ins. This means that combination of the GPL-covered plug-in with the non-free main program would violate the GPL." and "Linking [name of your
18:21 Riastradh joined #darcs
19:41 alexei joined #darcs
20:13 pointfree joined #darcs
20:30 pointfree disconnected from irc mid-sentence. hahas.
20:38 Heffalump IRC truncates long lines
20:38 pointfree Maybe it is better to go with the hsparql+4store for the sake of federation
20:38 pointfree TIL
20:41 Heffalump isn't a DVCS good enough federation in itself? :-)
20:50 pointfree True, and things like issue tracking should ideally be tied to the repo contents not to the hosting imho. However repos and their forks (projects) are isolated from each other, e.g. we can't pull in patches etc from unrelated repos because patch theory requires a common context afaik. Semantic web and SPARQL could bridge projects for things like issue cross-referencing between projects.
20:50 Heffalump fair enough
21:06 Riastradh joined #darcs
21:20 psnively joined #darcs
21:53 owst joined #darcs
21:58 inhortte joined #darcs
22:04 castlelore joined #darcs
22:05 castlelo1 joined #darcs
22:08 dixie joined #darcs
22:12 castlelore joined #darcs
22:31 castlelore joined #darcs
23:48 fr33domlover hey Heffalump, you here?
23:49 fr33domlover actually, darcs question to everyone: I did a `darcs rebase suspend`, then `darcs amend --unrecord`. Now I had several A (new files) and several R (removed files). I want to change this info moves. So I did several `darcs mv`, but now `darcs whatsnew -sl` shows both the moves and the A/R it showed earlier
23:50 fr33domlover is that ok? do I `darcs record` just the moves? or do the A/R suggest something is wrong here?

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary