#bookbrainz-devel

runnig setup.sh gives me lots of �Missing prerequisites� errors

probably due to ERROR: relation "person" does not exist (in sql/test-data.sql)

yeah as I recall the postgres versioning dependency stuff isn't dealt with properly by setup

there should be one SQL file that basically just sets up a working DB with test data for you; I think ocharles committed that

maybe not

I think sql versioning files are applied in the wrong order

yeah, they are

if you look at the top of each file you can figure out what the order is

and apply them directly

ah, there is date in the header

why not rename them?

to yyyymmdd-xxx, like the rest

I think it's broken deeper than that, actually

anyway the solution is to apply the patches in the proper order manually

Can't the script be modified to follow the right order?

Seems more reasonable

probably, but I don't know how

and ocharles hasn't worked on it

the right order can be guessed by filename

with the exception of very first version file

I am sure that it can be guessed every time

person.sql

It just feels weird that we have to guess it instead of sending a patch to fix setup.sh

reosarevok: it can't, the files aren't named in a sane fashion

ianmcorvidae, can you add a BB ticket?

(or is there one already?

)

yeah, that's a good idea

I don't think there is

I think the most proper way to fix this would be renaming person.sql

and do something with test-data.sql

other stuff that shares a date with another patch still gets misapplied

right...

ianmcorvidae, please add it - you know what is needed better than me

how do rails solve this problem?

almost nobody uses the style of SQL versioning we do (ours is pure-SQL)

ah, it's YYYYMMDDHHMMSS_create_products.rb

rails solves it by having non-SQL stuff, I think

ah, that also helps :P

I mean, just the filenames

it would make sense to use full time, I think

anyway, I made a ticket and assigned it to ocharles

ok, I will rename files manually for now

ahah, the files are applied in no order %)

yeah, I thought there was more of an issue than just the filenames XD-

and 2011101-remember-token.sql is missing a digit

find sql -type f -name '*.sql' | sort

this should work

(added | sort)

will try now

yep, it's all sorted now... backwards

heh

time to add -r :)

there's some embedded perl script, and I can't figure out what exactly it is doing

and then there is done | tsort...

(I have no shell scripting experience)

the perl script is trying to pull the dependencies out of the header

ah, so it's parsing

and tsort is topological sort, by dependencies

except it does it wrong, far as I can tell :P

the perl script that is

ok, I think running the script 20 times in a row helped %)

hah, well

that would do it

even if it only applies one patch each time that'll do it :P

psql -U bookbrainz bookbrainz < sql/test-data.sql

this is from README

so test-data shouldn't be applied along with other db patches

it should be applied after... something

right after the initial-schema one I think

So, right now we're at pre-NGS values

wrong window, reosarevok, but yes :P

lol

Hmneh?

Cancelled edits around NGS release

more of them for a while, but it looks like we're back down to pre-NGS values

Aha.

I should learn to read the logz.

it wasn't in the log

well, it was in MY logs

I pasted that to reosarevok in PM :P he just responded here

Oh.

I'm working a patch to let people graph arbitrary things (so long as they only graph one thing)

heh, fair enough

Too bad it's only 8pm.

I very sorely want to read that as "for arbitrary values of n = 1".

haha

:D

Let X be a nonempty subset of the positive integers. Then it is possible, for an arbitrary n \in X, to graph n things, provided X = {1}.

Leftmost, but but but you can graph 0 things!

It is just not very visible

you can also graph >1 things, so long as they're things on the main graph :P

You can graph 1 thing

Except if you're pedantic, which allows you to graph 0

Or boring enough to just want the main graph, with allows you >1

haha

He said "so long as they only graph one thing". That implies n != 0.

yeah, I guess you couldn't graph 0 with the thing I'm writing

you could graph 0 with the main graph, though

it makes a very pretty blank box :P

actually, probably with the thing I'm doing too, if you put in a bogus name :P

It's not blank

Not conceptually

It's philosophically full of content

or you're full of bullshit ;)

Yeah, well

I get a bunch of hard drives tomorrow, and then I have to wait until next week to hook them up. Sigh. :P

But I am sure someone agrees with that

I mean

Some people agree Wales' face on top of wiki pages is not annoying

People will argue the strangest things

heh

lately I've been getting the programmer who looks like he should be a singer for a death metal band instead

rather than Wales

Death metal?

I thought it was prog

(and everybody knows you can't trust a progrocker, because he's probably on three different drugs)

prog-metal usually has shorter hair, or at least doesn't as universally have long hair :P

Sure

But that mustache

I mean, look at it

Either he's a progrocker

Or he's Kid Rock

haha

oh, yeah, I'm actually logged in on this computer so I don't see the banners

DremoraLV: yea, sorry the scripts are in a bad way atm

DremoraLV: I have some local commits to try and sort this out

oh good, BrainzHub has woken up

DremoraLV: there are now scripts to initialize a clean database, if that's of any interest to you

create an empty database, then just pipe init-db.sql into it

https://gist.github.com/68fb8f358274a24479e2 not quite as cool as I thought, but here's me doing asserting the property that "allLanguages" returns all languages in the database

just got to rip out the setup function now

quickcheck ftw