freenode/#lisp - IRC Chatlog

2:13:56 pseudo_sue problem solved. it was a package-related issue, and I got around it by just using keywords instead of symbols for user input. the menu's just a crude front end to a genetic algo programme I'm writing, to spare someone from interacting with it through the REPL, as I've been doing.

2:14:06 Bike cool.

2:14:06 pseudo_sue thanks, Bike.

2:17:28 pseudo_sue this is nice. it's really satisfying watching this thing tear along at top speed, now.

2:29:45 pillton psuedo_sue: You should get in the habit of binding reader/printer variables when using them for I/O.

2:30:15 pillton clhs with-standard-io-syntax

2:30:15 specbot http://www.lispworks.com/reference/HyperSpec/Body/m_w_std_.htm

2:30:21 pillton clhs *read-eval*

2:30:22 specbot http://www.lispworks.com/reference/HyperSpec/Body/v_rd_eva.htm

2:36:23 pseudo_sue Thanks, pillton. This is actually the first time I needed to worry about input from a user. I'll look into that.

2:53:03 hydan ** NICK hydan`

2:53:37 pseudo_sue well, there we go. my first ever (post-hello-world) lisp programme. far from finished, messy, and amateurish, but it works. https://github.com/ObliviaSimplex/genetic-exercises/tree/master/genetic-linear

2:54:23 pseudo_sue thanks for all the help, #lisp.

2:54:34 Bike is 'linear' short for 'linear classifier' here? that's a new one for me.

2:54:46 pseudo_sue linear's short for "linear genetic programming".

2:54:58 pseudo_sue it's meant to be in contrast to tree-based gp

2:55:26 Bike hm. i see. strange stuff.

2:55:46 pseudo_sue basically, what it comes down to is that you're evolving bitvectors of machine code for a virtual, register-based machine, instead of functional trees.

2:55:55 pseudo_sue (a la koza)

2:57:26 pseudo_sue i figured it'd be a great excuse to start getting serious with CL, and be as functional as I want to be without making some of the essentially imperative bits of this algorithm pointlessly awkward.

2:57:55 pseudo_sue (it was a toss up between CL and Clojure for me, for a while)

2:58:40 pseudo_sue (but, as much as I like the idea of FP, it kind of irks me how Clojure forces your hand in that direction. I could see that being as irritating as Java's compulsory OO in 10 yrs time.)

3:21:13 loke_ Hello Lisp!

3:29:47 fewdea Hey all, I'm using postmodern to create my db tables. I don't think i'm able to include all necessary options in the class definition, for instance, an index on my serial column. so I execute the query to create the table, and then another query to add the column. return value for a successful first query is 0, otherwise an error is thrown. I don't want to execute the second query if the first failed. what is the preferred method of ha

3:29:48 fewdea ndling this situation?

3:33:10 fewdea maybe that's a dumb question. how could the second query execute if the first throws an error? I think what I meant was how do I roll back the first query if the second fails? can I wrap these in a transaction?

3:48:06 fewdea yeah, nevermind. used 'with-transaction'. sorry, thanks. *hides*

4:25:10 fewdea oh shit. this is #lisp, not #clnoobs ... sorry! *hides again*

4:41:39 Administrator ** NICK Guest90028

5:18:37 beach Good morning everyone!

5:20:35 Zhivago s/morning/news/

5:20:38 holycow * waves

5:22:27 beach Zhivago: What's the good news?

5:23:39 Zhivago Whatever you've been doing, hopefully.

9:15:34 shka hi

9:15:46 shka what are opinions around here on series package?

9:16:05 shka i'm desperatly trying to do something with need of writting loops

9:16:35 shka i tried to implemenet proxy class for vector that can do stuff like limiting range

9:16:55 shka but it turned out to be so slow that I gave up on this

9:17:18 loke_ shka: What do you mean "with need of writting loops"?

9:17:49 shka i just don't want to write loops for common tasks like: copy these range from here to here

9:17:57 shka *this

9:18:47 shka or: set elements starting with index and ending on to nil

9:19:10 shka basicly i want to avoid iterating over vectors

9:20:04 shka so, series seems to be acceptable answer

9:23:04 Cymew Do what you need for the tast at hand. Use LOOP, DO, recursion or MAPping.

9:23:14 Cymew Sorry, "task at hand".

9:27:13 shka Cymew: i really like map and reduce, but sometimes i care about getting maximum speed

9:28:21 jackdaniel why do you think map/reduce are slow? o_O

9:28:58 jackdaniel and that series will outperform these native constructs?

9:29:05 loke_ shka: And why would you think LOOP is slow? (it's the fastest alternative unless you want to do something multuithreaded)

9:29:29 loke_ Series is crazy slow. Which, I presume, is one reason no one uses them.

9:32:24 shka jackdaniel: just measured

9:32:37 jackdaniel share or never happened ;-)

9:32:46 shka please, remind me to not talk shit before benchmarking

9:32:52 shka next time i will do it

9:32:54 shka :/

9:33:13 jackdaniel oh :p

9:33:19 shka since map-into on 50 000 elements vector is just as fast as loop

9:33:32 shka i feel so stupid

9:33:49 shka loke_: loop is not slow

9:33:54 shka i just don't like looping

9:34:01 jackdaniel what if I tell you, that map* are often compiled into loops? ;]

9:34:11 shka but, but!

9:34:17 loke_ shka: Loop is awesome. The most clear and readable looping construct of any language.

9:34:19 shka THEY STILL HAVE TO CALL LAMBDA

9:34:40 shka loke_: sure, I just don't like to write loops in any language

9:34:46 shka it is just difficult to read

9:34:55 jackdaniel loke_: arguably ;)

9:35:44 dim if any SBCL guys are in the room, https://github.com/dimitri/pgloader/issues/327 is another case of "Heap exhausted, game over"

9:37:23 jackdaniel shka: http://paste.lisp.org/display/305838 http://paste.lisp.org/display/305839

9:37:38 jackdaniel (mainly the latter)

9:48:02 p_l dim: hey, any prospects on pgloader having postgres as input source?

9:52:00 shka ok guys

9:52:04 shka see you later

9:52:09 shka happy mapping!

9:59:11 dim p_l: well you're about the first to ask about that, IIRC

10:00:20 dim p_l: I though about adding a PG custom format dump source, but failed to find the time and energy to write a parser for the binary format there ; it would be interesting to load just a table from the dump while renaming it or restoring it into another schema, etc

10:00:39 dim p_l: do you have a use case for pgloader with PostgreSQL as input?

10:01:13 dim usually `copy out | copy in` or postgres_fdw are much better approach

10:05:44 p_l dim: ETL between two databases of different schema and field types

10:11:13 dim I would use postgres_fdw in that case

10:11:41 dim you can define a source as a query to be run rather than a table, or you can also define a set of views on the source database and copy over the FDW the data from the view

10:11:50 dim well the data from selecting against the view

10:12:27 dim define the view, then the fdw that maps against the view, then your ETL is just: insert into target.table select * from source_fdw.exported_view;

10:28:35 p_l dim: maybe with a bunch of custom functions to convert types it would work

10:34:59 dim typically those custom functions are easy to reduce to SQL expressions, so they are part of the view definition

13:32:40 ggole__ ** NICK ggole