#musicbrainz-devel

?

Anyone awake here?

reosarevok: any ideas on my last comment on http://tickets.musicbrainz.org/browse/SEARCH-174

Hmm

comment:null would make sense to me

(much less likely to exist than comment:unknown, which already gives 500 results for artists9

*)

null is very techy though isn't it

What about the fact that we are already using unknown for searching some fields, can I change that or do I have to leave that

It is a bit techy, yes

But it's not really an "unknown" comment, it's no comment

I mean, unknown makes sense for say, country, because everything has one, so if unset, it's unknown

Good point, important distinction

comment:none is less techy

Although it gives 4 results

okay so we could keep 'unknown' for all fields that are lists, indicating there is a value but we don't know it

Yeah

and use a different value for when there is no value, and there may not be one

I'd do that

and there, I think null or none are the best options. null being techy doesn't matter too much, because these are techy searches

is 'null' any better than the existing partial solution of '-' that we use in the canto example

(I mean, searching for no comment is something you won't probably do unless you know how stuff works)

Not sure

Let's see what other people think

nikki? hrglgrmpf? ocharles? warp?

the value could be "qwerty" for all I care if it means we can actually use it :P

a dash seems to make sense

warp: I suggest getting yourself a copy of 'Jailer'

it's basically your web service dumper tool, but polished :)

I think now have the ddistinction I prefer using -, seeing as we can already use that for some of the values

okay thanks guys

all the others are words and could be things someone wants to search for

whereas a dash is just a symbol that doesn't really have a meaning...

of course, someone might still want to search for a dash on its own, but since it doesn't mean anything, it's less likely

navap: jailer is something you might also be interested in. As people wanted a 'test dataset' - I'd be ok with generating a dump and putting it on the ftp

(I think that's the correct way to solve that problem)

nikki, the only thing Im wondering is that a search on - will bring back anything containing a - i.e http://musicbrainz.org/search?query=comment%3A\...

so may be ---- (entered as \-\-\-\-) would be better

by that logic, ____ would be even better, since _ doesn't seem to be used at all and certainly not four of them :P

I think I'd prefer just a single one though, multiple ones means you also need to remember exactly how many to enter

okay I'll leave as is then

oh, and the stats use country:unknown, etc, so before removing those the server should be updated

sort, what do you mean ?

s/sort/sorry

Thinking about reosarevok distinction if we do not know the release group for a release it still exists, but its a freeform field so thinking I will use '-' for that one, although 'unknown' could be deemed correct

and just keep 'unknown' for fixed lists

Hmm

You mean the release group type?

Or what? I mean, all releases have RGs

Your right, silly example

won't happen

nikki, i didn't understand what you meant 'the stats use country:unknown, etc, so before removing those the server should be updated'

I guess that means the script used to generate the stats searches for it

So if something is changed, the query made by the stats code needs to change too

Im not changing the unknown stuff, but would what is this script, never seen it

ijabz, whatever is used to collect http://musicbrainz.org/statistics I assume

wellI expect it queries the database, so I don't see the search stuff affects this

ijabz: nikki means that the stats page links to country:unknown, that's all

it doesn't actually perform that search

it just helps users see what the 864 releases with 'unknown' country are

at least, that's how I understand it

yes

and we don't want to break our own links :P

ok, that won't change

I don't think… but I don't know how that is done

heh

864

More like 216,085

But yeah

how do I reset my github master?

"reset"?

I edited a file on github so it made a commit to master

but I don't want that, I want master to match mb's master :/

I think you need to pull from your github respository, git reset --hard HEAD^, git push origin master -f

(that's not really a nice solution, but I know of no other)

that is the correct solution

rewriting history is not meant to be nice :)

what do I need to do to pull from it?

do you have a clone of your github repository on your computer?

yes

then change to that directory and git pull origin master

(if you're on the master branch)

and I guess "origin" is "github" for me

at least I think "origin" is mb's git thing

indeed

yikes, internet without dns sucks :(

hello!

Mineo: luckily google has made dns servers easy to remember.

(well, atleast their ipv4 addresses. I don't know their ipv6 addresses yet ;)

if only their wasn't this firewall between my university and them :/

:/

I'm getting errors locally in the release editor

(running master)

"Caught exception in MusicBrainz::Server::Controller::ReleaseEditor::Add->add "Can't use an undefined value as an ARRAY reference at /home/nikki/dev/mbserver/script/../lib/MusicBrainz/Server/Edit/Utils.pm line 152.""

all I did was open the release editor, enter "foo" as the title, "bar" as the artist and click the release duplicates tab :/

but I can't reproduce it on beta or test

ijabz: #SEARCH-174:Allow searching for when release barcode is not known and when barcode is known to be none.

hmm what have i missed, working here

ah invalid import statement crept in

Project search_server build #43: FIXED in 40 sec: http://ci.musicbrainz.org/job/search_server/43/

ijabz: #SEARCH-174:Invalid import

heh

not fast reply, shows the benefit of jenkins as although a coding error the error doesn't actually occur on my local build