#metabrainz

/

      • opatel99
        You can just add your string on line 30 and hit run
      • 2015-12-27 36117, 2015

      • CallerNo6
        opatel99, what am I testing for? That it follows MB's guidelines?
      • 2015-12-27 36109, 2015

      • opatel99
        CallerNo6: Aside from the guidelines, that it works based on common sense.
      • 2015-12-27 36143, 2015

      • CallerNo6
        Oh. Well, I know I can break it then, because I already know how hard a problem this is to solve (more or less)
      • 2015-12-27 36149, 2015

      • opatel99
        Well not `common sense`, but that it follows general name sorting. (This is to be used on just authors)
      • 2015-12-27 36141, 2015

      • opatel99
        CallerNo6: It is inevitable that this problem can never be fully solved, but to a general degree it could make it easier for many people under many cases.
      • 2015-12-27 36110, 2015

      • CallerNo6
        Okay, I can tell you right off that the mr/mrs/dr part doesn't make sense to me :-)
      • 2015-12-27 36106, 2015

      • CallerNo6
        People get degrees, or promotions, or titles all the time. That should change where they are in a collated list.
      • 2015-12-27 36120, 2015

      • CallerNo6
        (IMO, anyway)
      • 2015-12-27 36146, 2015

      • opatel99
        Oh those variables aren't even properly used yet. You mean should or shouldnt?
      • 2015-12-27 36100, 2015

      • CallerNo6
        And if two people are named Phil McGraw, I'd expect them to sort adjacent to one another even if only one is a doctor.
      • 2015-12-27 36140, 2015

      • CallerNo6
        oops, "shouldn't"!
      • 2015-12-27 36129, 2015

      • opatel99
        Currently, their last names would be adjacent, but Dr. would bump one phil higher than the other
      • 2015-12-27 36101, 2015

      • CallerNo6
        Sure, but in a long list they'd be separated by lots of other people.
      • 2015-12-27 36112, 2015

      • CallerNo6
        (if those people were also called McGraw)
      • 2015-12-27 36156, 2015

      • CallerNo6
        tl;dr, I think titles, honorifics etc should be omitted or put at the end.
      • 2015-12-27 36138, 2015

      • dr_saunde has quit
      • 2015-12-27 36145, 2015

      • drsaunde joined the channel
      • 2015-12-27 36135, 2015

      • CallerNo6
        Oh, sorry, I hadn't actually looked at the script. I guess you /are/ planning to special-case at least some titles/honorifics
      • 2015-12-27 36128, 2015

      • opatel99
        CallerNo6: That would be dependent on Sort guidelines for Books
      • 2015-12-27 36152, 2015

      • CallerNo6
        Okay. That could become a pretty long list :-)
      • 2015-12-27 36111, 2015

      • CallerNo6
        I feel like in a perfect world, artist names would be split into parts in the db.
      • 2015-12-27 36114, 2015

      • opatel99
        Honestly I think titles should be removed for authors
      • 2015-12-27 36122, 2015

      • CallerNo6
        (where "parts" are e.g. "given name", "family name")
      • 2015-12-27 36151, 2015

      • chirlu` wonders whether BB has an own style mailing list that is hosted somewhere else.
      • 2015-12-27 36111, 2015

      • chirlu`
        Also, the whole point of having sort names is that they *cannot* be automatically determined based on the name alone in all cases.
      • 2015-12-27 36136, 2015

      • chirlu`
        If they could, we’d just store the name and calculate the sort name when it is needed.
      • 2015-12-27 36156, 2015

      • CallerNo6
        like I said, I'm skeptical :-)
      • 2015-12-27 36119, 2015

      • CallerNo6
        But I think it'd be much closer to possible if we split names into parts.
      • 2015-12-27 36133, 2015

      • opatel99
        Okay. How about this. Does https://repl.it/Bact/21 resolve the current situation we have for guessing author names? CallerNo6 Would it be difficult to post-refactor all of the names?
      • 2015-12-27 36110, 2015

      • CallerNo6
        are you going to make me google (er, DDG) "post-refactor"? fine, be that way.
      • 2015-12-27 36132, 2015

      • opatel99
        That is probably not even a word. Post-change
      • 2015-12-27 36148, 2015

      • opatel99
        Change after adding.
      • 2015-12-27 36116, 2015

      • CallerNo6
        if it means something, I think it's a perfectly cromulent word.
      • 2015-12-27 36144, 2015

      • chirlu`
        Disregarding these titles etc., there is simply no way an algorithm or a person can correctly guess whether “A B C” is a Ms B C with first name A or a Ms C with first names A B.
      • 2015-12-27 36110, 2015

      • opatel99
        chirlu`: I tried my best with https://repl.it/Bact/21
      • 2015-12-27 36111, 2015

      • chirlu`
        That’s how the sort name stores additional knowledge.
      • 2015-12-27 36126, 2015

      • opatel99
        Of course, it is not completely possible, but I just want the basics to work.
      • 2015-12-27 36129, 2015

      • CallerNo6
        That's the thing, I don't think it can every work well enough to be better than an empty field.
      • 2015-12-27 36143, 2015

      • CallerNo6
        I mean, at least you can search for the empty field and fix it.
      • 2015-12-27 36155, 2015

      • opatel99
        So just drop the task?
      • 2015-12-27 36120, 2015

      • CallerNo6
        But as an exercise, if you're into it, I'll talk about it all night.
      • 2015-12-27 36122, 2015

      • CallerNo6
        You'd need to know if e.g. "Van Halen" is one name or two, if Gabriel García Márquez has a middle name, if a Japanese name is being presented family name first or last.
      • 2015-12-27 36144, 2015

      • CallerNo6
        It's a really tough problem.
      • 2015-12-27 36131, 2015

      • chirlu`
        Well, as the MB Guess Case function shows, it can save a lot of typing even if it is only right in 80% of all cases.
      • 2015-12-27 36156, 2015

      • CallerNo6
        But then the incorrect 20% is harder to find, when it could have simply been left blank.
      • 2015-12-27 36154, 2015

      • CallerNo6
        oh, "guess case"? no opinion.
      • 2015-12-27 36110, 2015

      • chirlu`
        Yes, it’s a good tool only in the hands of those editors who know what the sort name should be and manually enter it where guess sort name fails.
      • 2015-12-27 36118, 2015

      • chirlu`
        I meant the sort-name guessing.
      • 2015-12-27 36147, 2015

      • opatel99
        So... What do I do?
      • 2015-12-27 36102, 2015

      • CallerNo6
        how did you get your self into this?
      • 2015-12-27 36118, 2015

      • CallerNo6
        was it, "oh, that looks doable!"?
      • 2015-12-27 36140, 2015

      • opatel99
        I don't really look at doable or not before choosing my task
      • 2015-12-27 36103, 2015

      • CallerNo6
        ah, it was on the task list already?
      • 2015-12-27 36109, 2015

      • opatel99
        Seemed useful. Wow it could make it easier and more organized for editors
      • 2015-12-27 36123, 2015

      • opatel99
        Well I claimed it
      • 2015-12-27 36150, 2015

      • CallerNo6
        Okay, hmm, my suggestion is: ask Mr. Most or Mr. Sputnik how well they want this to work.
      • 2015-12-27 36103, 2015

      • CallerNo6
        If they're happy with the way MB works, duplicate that.
      • 2015-12-27 36114, 2015

      • CallerNo6
        If they want it to be, as chirlu` says, a keystroke-saver for editors who already know what they're doing, then maybe make that clear.
      • 2015-12-27 36103, 2015

      • CallerNo6
        whoa, what's going on in version 21? I got a lot of "undefined"
      • 2015-12-27 36129, 2015

      • opatel99
      • 2015-12-27 36116, 2015

      • opatel99
        CallerNo6: In your favor: Do not list titles (Dr., Sir, Saint, etc.) or degrees (PhD, MA, DDS, etc.) with names. - Purdue OWL (Pretty much the definers of MLA book citation)
      • 2015-12-27 36146, 2015

      • CallerNo6
        exactly, I'd much rather follow MLA where possible
      • 2015-12-27 36158, 2015

      • CallerNo6
        (or similar authorities)
      • 2015-12-27 36118, 2015

      • opatel99
        What about Dr. Seuss? Is that an acceptable alias?
      • 2015-12-27 36107, 2015

      • CallerNo6
        Probably? That's the sort of thing where I'd follow convention. How have libraries done it historically?
      • 2015-12-27 36138, 2015

      • CallerNo6
        (same with Mark Twain)
      • 2015-12-27 36112, 2015

      • opatel99
        My library puts Seuss under the G's for Geisel
      • 2015-12-27 36104, 2015

      • CallerNo6
        huh. I wonder if that's a holdover from card catalogs.
      • 2015-12-27 36132, 2015

      • CallerNo6
        Oh, do you mean sorted under "G" on the shelf?
      • 2015-12-27 36135, 2015

      • opatel99
        yeah
      • 2015-12-27 36102, 2015

      • CallerNo6
        Yeah, definitely a holdover from the limitations of physical libraries.
      • 2015-12-27 36125, 2015

      • dpmittal joined the channel
      • 2015-12-27 36129, 2015

      • opatel99
        LordSputnik: Leftmost: ping
      • 2015-12-27 36154, 2015

      • Bookzombie has quit
      • 2015-12-27 36113, 2015

      • Bookzombie joined the channel
      • 2015-12-27 36144, 2015

      • dpmittal has quit
      • 2015-12-27 36123, 2015

      • dpmittal joined the channel
      • 2015-12-27 36132, 2015

      • ariscop has quit
      • 2015-12-27 36114, 2015

      • dpmittal has quit
      • 2015-12-27 36132, 2015

      • ariscop joined the channel
      • 2015-12-27 36107, 2015

      • dpmittal joined the channel
      • 2015-12-27 36155, 2015

      • dpmittal has quit
      • 2015-12-27 36100, 2015

      • dpmittal joined the channel
      • 2015-12-27 36142, 2015

      • dpmittal has quit
      • 2015-12-27 36101, 2015

      • opatel99
        Is https://codein.withgoogle.com/tasks/5104724189642… asking for aliases for authors or works?
      • 2015-12-27 36102, 2015

      • gcibot
        BB: Create alias style guidelines || 3 days || Documentation, Outreach / Research || MetaBrainz Foundation || Instances: 0/3
      • 2015-12-27 36104, 2015

      • dpmittal joined the channel
      • 2015-12-27 36109, 2015

      • opatel99
        CallerNo6: Awake?
      • 2015-12-27 36112, 2015

      • Lingo_ joined the channel
      • 2015-12-27 36109, 2015

      • Lingo_ has quit
      • 2015-12-27 36139, 2015

      • yeeeargh joined the channel
      • 2015-12-27 36132, 2015

      • CatQuest joined the channel
      • 2015-12-27 36132, 2015

      • CatQuest has quit
      • 2015-12-27 36132, 2015

      • CatQuest joined the channel
      • 2015-12-27 36143, 2015

      • Lingo_ joined the channel
      • 2015-12-27 36127, 2015

      • LordSputnik
        opatel99: follow what MB does for now please
      • 2015-12-27 36151, 2015

      • opatel99
        LordSputnik: permission to query a URL?
      • 2015-12-27 36117, 2015

      • LordSputnik
        eg. (John Smith -> Smith, John), (Dr. Seuss -> Suess, Dr.), (J. R. R. Tolkein -> Tokein, J. R. R.)
      • 2015-12-27 36124, 2015

      • LordSputnik
        opatel99: what for?
      • 2015-12-27 36152, 2015

      • LordSputnik
        also I can't see the demos at any of those links on my phone :(
      • 2015-12-27 36155, 2015

      • opatel99
        A link to my draft of a alias guideline
      • 2015-12-27 36134, 2015

      • opatel99
        I can't have two tasks at once and the task has 3 instances
      • 2015-12-27 36139, 2015

      • LordSputnik
        oh you want to pm me a link? that's fine :)
      • 2015-12-27 36114, 2015

      • Lingo_ has quit
      • 2015-12-27 36122, 2015

      • opatel99
        Done.
      • 2015-12-27 36108, 2015

      • opatel99
        LordSputnik: I tried to follow the inverted pyramid on a broad sense
      • 2015-12-27 36120, 2015

      • LordSputnik
        opatel99: looks good, can you include a section describing any differences with the MB english guide?
      • 2015-12-27 36145, 2015

      • opatel99
        English guide?
      • 2015-12-27 36142, 2015

      • LordSputnik
      • 2015-12-27 36101, 2015

      • opatel99
        Also since this is an existing task, should I abandon my current task and submit that link as the appropriate guideline task?
      • 2015-12-27 36146, 2015

      • dpmittal has quit
      • 2015-12-27 36107, 2015

      • LordSputnik
        And once that's done, PR to the BB user guide and make a mailing list post (following http://bb-user-guide.readthedocs.org/en/latest/st…)
      • 2015-12-27 36106, 2015

      • LordSputnik
        I wouldn't switch tasks yet - do that just before the guideline is approved
      • 2015-12-27 36116, 2015

      • opatel99
        LordSputnik: 3 instances .. Is that ok?
      • 2015-12-27 36112, 2015

      • LordSputnik
        I'm going to change it to 1 now
      • 2015-12-27 36116, 2015

      • opatel99
        Also: When no new changes have been made to the guideline document for a week, the Style Committee decide whether approve the guideline and make any necessary modifications, or reject it for some good reason.
      • 2015-12-27 36123, 2015

      • opatel99
        So that task will take one week?
      • 2015-12-27 36102, 2015

      • LordSputnik
        You can run it in the background while you complete other tasks
      • 2015-12-27 36112, 2015

      • opatel99
        Ok
      • 2015-12-27 36150, 2015

      • dpmittal joined the channel
      • 2015-12-27 36158, 2015

      • opatel99
        LordSputnik: I was able to run the link above by putting my phone in landscape mode
      • 2015-12-27 36158, 2015

      • LordSputnik
        I'll keep 2 instances for now in case someone claims it without asking first
      • 2015-12-27 36106, 2015

      • opatel99
        Ok
      • 2015-12-27 36127, 2015

      • opatel99
      • 2015-12-27 36138, 2015

      • LordSputnik
        opatel99: I just get a blank console on the left and blank output window on the right
      • 2015-12-27 36124, 2015

      • LordSputnik
        ahh got it
      • 2015-12-27 36124, 2015

      • opatel99
        LordSputnik: for some reason both will populate when you hit the run button
      • 2015-12-27 36149, 2015

      • opatel99
        Not the best mobile experience, but it can work
      • 2015-12-27 36131, 2015

      • Lingo_ joined the channel
      • 2015-12-27 36118, 2015

      • zas
        Any java guy around ? i found something i would like to submit
      • 2015-12-27 36125, 2015

      • zas
      • 2015-12-27 36150, 2015

      • zas
        i found that the answer (http header) contains spurious nul bytes
      • 2015-12-27 36112, 2015

      • zas
      • 2015-12-27 36128, 2015

      • zas
        and following lines store response in a buffer
      • 2015-12-27 36143, 2015

      • zas
      • 2015-12-27 36150, 2015

      • zas
        splits on space char
      • 2015-12-27 36111, 2015

      • zas
        so, to me, trailing nul chars are never trimmed out
      • 2015-12-27 36132, 2015

      • zas
        can someone look at this code and confirm
      • 2015-12-27 36111, 2015

      • Mineo wonders where http is involved in that code
      • 2015-12-27 36128, 2015

      • zas
        headers are generated from those strings
      • 2015-12-27 36141, 2015

      • zas
      • 2015-12-27 36120, 2015

      • Mineo
        but isn't it only using the first three characters from the response?
      • 2015-12-27 36103, 2015

      • Mineo
        https://bitbucket.org/metabrainz/search-server/sr… first takes the first 5 characters of the response, splits on the spaces (expects 2 of them) and then uses the 3 characters
      • 2015-12-27 36107, 2015

      • zas
        the thing is that nginx generates an error due to invalid headers (they arent supposed to contain nul bytes)
      • 2015-12-27 36132, 2015

      • zas
        client <-> proxy <-> search server : since search server send invalid header, proxy generates an error, and client never see the search server (invalid) answer, therefore not aware of the rate limit, and retries
      • 2015-12-27 36103, 2015

      • zas
        the last part needs to be confirmed but the full testing is a bit complex to set up
      • 2015-12-27 36110, 2015

      • zas
        anyway lunch time for me, bbl
      • 2015-12-27 36107, 2015

      • Lingo_ has quit
      • 2015-12-27 36123, 2015

      • dpmittal has quit