Hi, I need to make sure I understand at what scope uniqueness is required for short identifiers in vocabularies. Must short identifiers be unique across all vocabularies within one authority, or do they have to be unique just within a vocabulary? I ask because for our UCJEPS installation, we have 3 taxonomy vocabularies. Short identifiers are unique within each vocabulary but not across the entire set. When I try to display one of those that has a "duplicate" short identifier in a demo 2.4 system, CSpace took 5 minutes to display the record. The GET statement in firebug was only reporting: http://ucjeps.collectionspace.org:8180/collectionspace/tenant/ucjeps/vocabularies/taxon/urn:cspace:name(0) although the referer within the request headers is http://ucjeps.collectionspace.org:8180/collectionspace/ui/ucjeps/html/taxon.html?csid=urn:cspace:name(0)&vocab=taxon Thanks, Chris

Not sure i understand the requirements here entirely but urn:cspace:name(0) does seem very brief! It would be easy to construct identifiers that are globally unique (e.g. attaching a UUID at the end of urn:cspace:) so why not use those so other systems can work directly with the concepts that you are identifying? John On Fri, Aug 31, 2012 at 12:58 PM, Chris Hoffman <chris.hoffman@berkeley.edu> wrote: > Hi, > I need to make sure I understand at what scope uniqueness is required for short identifiers in vocabularies. Must short identifiers be unique across all vocabularies within one authority, or do they have to be unique just within a vocabulary? I ask because for our UCJEPS installation, we have 3 taxonomy vocabularies. Short identifiers are unique within each vocabulary but not across the entire set. When I try to display one of those that has a "duplicate" short identifier in a demo 2.4 system, CSpace took 5 minutes to display the record. The GET statement in firebug was only reporting: > http://ucjeps.collectionspace.org:8180/collectionspace/tenant/ucjeps/vocabularies/taxon/urn:cspace:name(0) > > although the referer within the request headers is > http://ucjeps.collectionspace.org:8180/collectionspace/ui/ucjeps/html/taxon.html?csid=urn:cspace:name(0)&vocab=taxon > > Thanks, > Chris > > > _______________________________________________ > Talk mailing list > Talk@lists.collectionspace.org > http://lists.collectionspace.org/mailman/listinfo/talk_lists.collectionspace.org -- John Deck (541) 321-0689

I thought short identifiers could not begin with a number? Susan > Not sure i understand the requirements here entirely but > urn:cspace:name(0) does seem very brief! > > It would be easy to construct identifiers that are globally unique > (e.g. attaching a UUID at the end of urn:cspace:) so why not use those > so other systems can work directly with the concepts that you are > identifying? > > John > > On Fri, Aug 31, 2012 at 12:58 PM, Chris Hoffman > <chris.hoffman@berkeley.edu> wrote: >> Hi, >> I need to make sure I understand at what scope uniqueness is required >> for short identifiers in vocabularies. Must short identifiers be unique >> across all vocabularies within one authority, or do they have to be >> unique just within a vocabulary? I ask because for our UCJEPS >> installation, we have 3 taxonomy vocabularies. Short identifiers are >> unique within each vocabulary but not across the entire set. When I try >> to display one of those that has a "duplicate" short identifier in a >> demo 2.4 system, CSpace took 5 minutes to display the record. The GET >> statement in firebug was only reporting: >> http://ucjeps.collectionspace.org:8180/collectionspace/tenant/ucjeps/vocabularies/taxon/urn:cspace:name(0) >> >> although the referer within the request headers is >> http://ucjeps.collectionspace.org:8180/collectionspace/ui/ucjeps/html/taxon.html?csid=urn:cspace:name(0)&vocab=taxon >> >> Thanks, >> Chris >> >> >> _______________________________________________ >> Talk mailing list >> Talk@lists.collectionspace.org >> http://lists.collectionspace.org/mailman/listinfo/talk_lists.collectionspace.org > > > > -- > John Deck > (541) 321-0689 > > _______________________________________________ > Talk mailing list > Talk@lists.collectionspace.org > http://lists.collectionspace.org/mailman/listinfo/talk_lists.collectionspace.org >

Must be unique within a vocab, or you will get duplicates, and things will be unhappy. Having the same ID in several vocabs (even within the same authority) is fine. ShortIdentifiers are the basis for fetching, so having a short ID that is a reasonable word length is good (indexes better). Having one that has a numeric pseudo-random tail (e.g., "fooBar2345", as we create for auto-create on termCompletion) balances ease of reading/debugging with uniqueness that will behave well in the DB. The DB will try to stem the shortId when it prepares keyword indexes (which we use in certain cases), and so the suffix keeps things behaving better, I think. UUIDs will work fine as well, but are harder to make sense of when debugging, etc. Patrick > -----Original Message----- > From: talk-bounces@lists.collectionspace.org > [mailto:talk-bounces@lists.collectionspace.org] On Behalf Of > Chris Hoffman > Sent: Friday, August 31, 2012 12:58 PM > To: CollectionSpace Talk List > Subject: [Talk] Short identifier uniqueness requirements question? > > Hi, > I need to make sure I understand at what scope uniqueness is > required for short identifiers in vocabularies. Must short > identifiers be unique across all vocabularies within one > authority, or do they have to be unique just within a > vocabulary? I ask because for our UCJEPS installation, we > have 3 taxonomy vocabularies. Short identifiers are unique > within each vocabulary but not across the entire set. When I > try to display one of those that has a "duplicate" short > identifier in a demo 2.4 system, CSpace took 5 minutes to > display the record. The GET statement in firebug was only reporting: > > http://ucjeps.collectionspace.org:8180/collectionspace/tenant/ > ucjeps/vocabularies/taxon/urn:cspace:name(0) > > although the referer within the request headers is > > http://ucjeps.collectionspace.org:8180/collectionspace/ui/ucje > ps/html/taxon.html?csid=urn:cspace:name(0)&vocab=taxon > > Thanks, > Chris > > > _______________________________________________ > Talk mailing list > Talk@lists.collectionspace.org > http://lists.collectionspace.org/mailman/listinfo/talk_lists.c > ollectionspace.org >

I think we just require all word chars (alnum and _, IIRC). > -----Original Message----- > From: talk-bounces@lists.collectionspace.org > [mailto:talk-bounces@lists.collectionspace.org] On Behalf Of > sstone@socrates.berkeley.edu > Sent: Friday, August 31, 2012 1:49 PM > To: John Deck > Cc: CollectionSpace Talk List > Subject: Re: [Talk] Short identifier uniqueness requirements question? > > I thought short identifiers could not begin with a number? > > Susan > > > Not sure i understand the requirements here entirely but > > urn:cspace:name(0) does seem very brief! > > > > It would be easy to construct identifiers that are globally unique > > (e.g. attaching a UUID at the end of urn:cspace:) so why > not use those > > so other systems can work directly with the concepts that you are > > identifying? > > > > John > > > > On Fri, Aug 31, 2012 at 12:58 PM, Chris Hoffman > > <chris.hoffman@berkeley.edu> wrote: > >> Hi, > >> I need to make sure I understand at what scope uniqueness > is required > >> for short identifiers in vocabularies. Must short identifiers be > >> unique across all vocabularies within one authority, or do > they have > >> to be unique just within a vocabulary? I ask because for > our UCJEPS > >> installation, we have 3 taxonomy vocabularies. Short > identifiers are > >> unique within each vocabulary but not across the entire > set. When I > >> try to display one of those that has a "duplicate" short > identifier > >> in a demo 2.4 system, CSpace took 5 minutes to display the > record. > >> The GET statement in firebug was only reporting: > >> > >> > http://ucjeps.collectionspace.org:8180/collectionspace/tenant/ucjeps/ > >> vocabularies/taxon/urn:cspace:name(0) > >> > >> although the referer within the request headers is > >> > >> > http://ucjeps.collectionspace.org:8180/collectionspace/ui/ucjeps/html > >> /taxon.html?csid=urn:cspace:name(0)&vocab=taxon > >> > >> Thanks, > >> Chris > >> > >> > >> _______________________________________________ > >> Talk mailing list > >> Talk@lists.collectionspace.org > >> > http://lists.collectionspace.org/mailman/listinfo/talk_lists.collecti > >> onspace.org > > > > > > > > -- > > John Deck > > (541) 321-0689 > > > > _______________________________________________ > > Talk mailing list > > Talk@lists.collectionspace.org > > > http://lists.collectionspace.org/mailman/listinfo/talk_lists.collectio > > nspace.org > > > > > > _______________________________________________ > Talk mailing list > Talk@lists.collectionspace.org > http://lists.collectionspace.org/mailman/listinfo/talk_lists.c > ollectionspace.org >

Thanks, I'm pretty sure my problem is that the legacy identifier I used in this case was simply a one-digit string "0", and fetching is therefore slow. Sigh. Chris On Aug 31, 2012, at 2:00 PM, Patrick Schmitz wrote: > Must be unique within a vocab, or you will get duplicates, and things will > be unhappy. Having the same ID in several vocabs (even within the same > authority) is fine. > > ShortIdentifiers are the basis for fetching, so having a short ID that is a > reasonable word length is good (indexes better). Having one that has a > numeric pseudo-random tail (e.g., "fooBar2345", as we create for auto-create > on termCompletion) balances ease of reading/debugging with uniqueness that > will behave well in the DB. The DB will try to stem the shortId when it > prepares keyword indexes (which we use in certain cases), and so the suffix > keeps things behaving better, I think. > > UUIDs will work fine as well, but are harder to make sense of when > debugging, etc. > > Patrick > >> -----Original Message----- >> From: talk-bounces@lists.collectionspace.org >> [mailto:talk-bounces@lists.collectionspace.org] On Behalf Of >> Chris Hoffman >> Sent: Friday, August 31, 2012 12:58 PM >> To: CollectionSpace Talk List >> Subject: [Talk] Short identifier uniqueness requirements question? >> >> Hi, >> I need to make sure I understand at what scope uniqueness is >> required for short identifiers in vocabularies. Must short >> identifiers be unique across all vocabularies within one >> authority, or do they have to be unique just within a >> vocabulary? I ask because for our UCJEPS installation, we >> have 3 taxonomy vocabularies. Short identifiers are unique >> within each vocabulary but not across the entire set. When I >> try to display one of those that has a "duplicate" short >> identifier in a demo 2.4 system, CSpace took 5 minutes to >> display the record. The GET statement in firebug was only reporting: >> >> http://ucjeps.collectionspace.org:8180/collectionspace/tenant/ >> ucjeps/vocabularies/taxon/urn:cspace:name(0) >> >> although the referer within the request headers is >> >> http://ucjeps.collectionspace.org:8180/collectionspace/ui/ucje >> ps/html/taxon.html?csid=urn:cspace:name(0)&vocab=taxon >> >> Thanks, >> Chris >> >> >> _______________________________________________ >> Talk mailing list >> Talk@lists.collectionspace.org >> http://lists.collectionspace.org/mailman/listinfo/talk_lists.c >> ollectionspace.org >> >

Thanks, John. You're right that the string urn:cspace:name(0) wouldn't be a very good identifier! Fortunately, this one is just used internally, and even then only under specific circumstances. There are UUIDs and such attached to this record that will allow us to construct more appropriate globally unique identifiers for data sharing purposes. Chris On Aug 31, 2012, at 1:34 PM, John Deck wrote: > Not sure i understand the requirements here entirely but > urn:cspace:name(0) does seem very brief! > > It would be easy to construct identifiers that are globally unique > (e.g. attaching a UUID at the end of urn:cspace:) so why not use those > so other systems can work directly with the concepts that you are > identifying? > > John > > On Fri, Aug 31, 2012 at 12:58 PM, Chris Hoffman > <chris.hoffman@berkeley.edu> wrote: >> Hi, >> I need to make sure I understand at what scope uniqueness is required for short identifiers in vocabularies. Must short identifiers be unique across all vocabularies within one authority, or do they have to be unique just within a vocabulary? I ask because for our UCJEPS installation, we have 3 taxonomy vocabularies. Short identifiers are unique within each vocabulary but not across the entire set. When I try to display one of those that has a "duplicate" short identifier in a demo 2.4 system, CSpace took 5 minutes to display the record. The GET statement in firebug was only reporting: >> http://ucjeps.collectionspace.org:8180/collectionspace/tenant/ucjeps/vocabularies/taxon/urn:cspace:name(0) >> >> although the referer within the request headers is >> http://ucjeps.collectionspace.org:8180/collectionspace/ui/ucjeps/html/taxon.html?csid=urn:cspace:name(0)&vocab=taxon >> >> Thanks, >> Chris >> >> >> _______________________________________________ >> Talk mailing list >> Talk@lists.collectionspace.org >> http://lists.collectionspace.org/mailman/listinfo/talk_lists.collectionspace.org > > > > -- > John Deck > (541) 321-0689