#musicbrainz-devel

ijabz, reosarevok, ianmcorvidae?

djce and I have been working on rika, I think that's later in the topic

ok.

lets move on to the next issues.

preferences for advanced search?

reosarevok!

warp has changed the topic to: Agenda: preference for advanced search? (reosarevok), meta description for google (kepstin), Cover Art Proposal (ijabz), MBS-4031&MBS-4113 (ocharles), rika news

So

I tend to want to use the advanced search *much* more often than the basic one

I know so does hawke_, dunno about other people

me too.

but are we representative of most of our users?

Probably not

my hypothesis here is that the basic search is not expressive enough, and that's what needs to be improved

navap___ has changed the topic to: Agenda: preference for advanced search? (reosarevok), meta description for google (kepstin), Cover Art Proposal (ijabz), MBS-4031&MBS-4113 (ocharles), rika news, robots.txt (navap)

:p

(well, not *you* :P)

adhawkins has changed the topic to: Agenda: preference for advanced search? (reosarevok), meta description for google (kepstin), Cover Art Proposal (ijabz), MBS-4031&MBS-4113 (ocharles), rika news, robots.txt (navap), libmb4 (adhawkins)

My opinion is that the default search should give better quality results, not specifically that I want to use advanced for the advanced features.

navap___, neither with the basic :p

We discussed quite a lot earlier on, and it seems to me for some users its makes sense for basic to be default and for others advanced would make more sense

ijabz has convinced me that having a separate basic search makes sense

(and advanced gives better results)

I think that most people would prefer something like http://wiki.apache.org/solr/DisMax as the default option

which is why I would also like a preference

But I'd like to be able to have a preference in my settings to default to advanced

(off by default)

it would be a 5 minute fix to remember the checkbox state.

cookie seems simple to add

yea

which unfortunately doesn't give you access to the individual fields, like the basic MB search, but it's much smarter in general

warp, also for the search in the upper box?

ocharles: I *don't* want it to remember if I had the advanced box checked last time I searched :P

luks: sounds nice

reosarevok: yes

reosarevok: no, that doesn't have a checkbox.

luks, yes artist search sorts of does this, and I am looking at actually using a DisjunctionMaxQuery which dismay is base don

warp, that's the useful one :p

warp: It's still doable by a hidden input

ijabz: my primary use case are releases

(the other still requires going to the search page, I can as well click the box there)

ijabz: or anything where I want to search in title + artist

warp: I had it done on my test server way back when that was the only searc hthat worked on mirrors

I would like to see the search improved though. not having to select the type and having it detect if I enter a valid advanced search would be nice

Yes please — selecting the type is annoying and very limiting

navap___: then it may not be obvious how to change it.

Yeah, I want to extend it to releases ecetera, currently only being done for artist search

er…nikki you meant advanced/basic, didn’t you…not the entity type.

it sounds like there are two things to be done here: 1. revamp how we cal the current search server, including keeping state of the checkbox 2) investigating improved search back-end options.

*call

hawke_: I mean entity type. they were two completely separate things that I think could be improved

nikki: OK, then I agree.

warp, that's why it'd be more obvious with a preference! :)

reosarevok: only if you know there’s a preference to change.

For the rare cases when I *don't* want to use advanced (I'm yet to find one) I can always disable it manually for that search

hawke_, if you're the kind of person who wants to default to advanced because of its uses (not because of better results) you know that

IMO it’s a choice between “search that works” (advanced) and “mostly useless search” (basic)

If you don't even read the "Search" wiki page

You won't know what this "advanced" thing is anyway :)

ijabz: can you please do some more research into this dismax thing and a possibly combined index?

then, who wants to take on spearheading consensus on how to call the search server? preferences, remember checkboxes and the like.

I think dismax is the best first step

personally

or something that improves basic search results

might be hard to port it from solr though

So I'm not getting my preference, right? :P

Yes, I'm already started on this :)

yay

ijabz: excellent.

reosarevok: if we do this right, you shouldn't need to use advanced seacrh

this is more my point

reosarevok: btw, I can make you a tiny user script to add advanced to the search box at the top if you like :P

title AND artist should not be advanced :)

ocharles: yes, but that is the long term solution. sounds like people want a short term fix as well.

ocharles, and barcode:XXXXXXXXXXX shouldn't either?

title AND artist^5 AND (title~10 OR title~2) should be advanced

ocharles, no we need both, I already went through all this an hour earlier

or should we simply wait for ijabz to give his is feeback?

reosarevok: correct

ruaok, I guess a script can work for now

ijabz: I agree we need both

but advanced should be an exception, not a common need

nikki, fine, that'd be a good thing then, please do :)

there is nothing wrong with including things like that in the dismax parser

well no I don't agree

(and put it on userscripts for other people who want it)

you can have barcode, asin, etc. there

ocharles: but IMO artist:Pink Floyd should not be advanced, especially if there’s typeahead suggestions

enabled by default

that's my point...

it seems disjuncted max is exactly what I describe, I just didn't know there was a name for it :)

ok, lets leave it as a user script for now and wait for ijabz to make recommendations on how to improve search in the future.

Let me make the improvements to search results, and then we can take it form there

ijabz: +1

ok, lets move on for now.

warp has changed the topic to: Agenda: meta description for google (kepstin), Cover Art Proposal (ijabz), MBS-4031&MBS-4113 (ocharles), rika news, robots.txt (navap), libmb4 (adhawkins)

kepstin-netbook, you're up.

If he's not around, I guess nikki or I or ianmcorvidae could also take this one

But if he's around that's better, so kepstin-netbook?

not me

Well, you can explain the problem

I don't even know if that *is* the right solution but I expect people here to have ideas :p

if kepstin's not here I can explain what he was talking about at least

Please do

our descriptions on google are terrible

nikki was talking about some the other day which had no actual information about the release in them

for example

"20 May 2011 ... Annotation. Annotation last modified on 2005-11-12 13:05 UTC. « ‹ 1 2 3 › » Page 1 of 3. Album. Year, Title, Rating, Releases. 1968, This Was ..." is an example description

ouch. the tests I did earlier at least managed to get just the annotation

kepstin found a page on google webmaster tools about setting a meta description tag; I think he wanted to suggest we generate useful stuff to put in there, to improve our search descriptions

or nothing

I don't remember the link, but the problem is clear, in any case

sounds like a valid problem.

can someone make a ticket, plz?

ocharles: well that artist has an empty annotation, that thanks to navap, we now display as a section on the page :P

seems like a valuable thing. I'd say create a ticket in jira, link to the relevant google pages, and then poke developers to start working on it :)

nikki: heh

hm?

warp: +1

nikki, can you add said ticket? :)

time is getting short, lets move on to CAA proposal.

ijabz?

warp has changed the topic to: Agenda: Cover Art Proposal (ijabz), MBS-4031&MBS-4113 (ocharles), rika news, robots.txt (navap), libmb4 (adhawkins)

Its just general frustration about how the specification is proceeding

you and me both :P

ianmcorvidae, isn't yours about how it's *not* proceeding? :p

same thing :P

and that I feel totally confused as to what is being proposed, and what has been considered

we've been talking past each other, I'm not sure how to resolve that; I think I may ask someone else to serve as representative from here though

http://tickets.musicbrainz.org/browse/MBS-4195 btw

who was supposed to put forth a combined proposal?

ianmcorvidae?

yes

what is the current proposal?

I think the current proposal needs writing up

Of course, maybe I'm wrong

/mbid/front absolutely must 200 or 404

warp and I talked, and the email I sent the other day is the current proposal

so I agree with ijabz and murdos there

ruaok: <mbid>/n.jpg or <mbid>-n.jpg, where n is a number starting at 1. the first image is the "frontiest" image we have.

except people want back as well, and to define front as something other than the 'frontiest' concept I put forth