SSA-1.1

[Douglas Tody]

On Fri, 29 Apr 2011, Petr Skoda wrote:
> OK I am giving up - - to solve the problem on which  are several groups of 
> people in IAU working for years is impossible by simple rules;-)

Well as I noted earlier, some work and prototyping with real data is
required before we want to define more complex semantics for something
like TARGETNAME!

One thing which I have not seen mentioned here is that simple pattern
matching (case sensitive/insensitive, wildcards) is not the same thing
as object name resolution.  Pattern matching would probably suffice for
e.g., identifying a data collection to be queried, but object name
resolution can be much more complex.  Hence if we want TARGETNAME to be
used for general astronomical objects (as in the SED use case) the
service might just need to make a call to a real name resolver such as
NED or SIMBAD.  If the data collection being queried is e.g. planetary
then a pattern match might suffice (and be all that is possible).

Once again this points out the difficulty of a one size fits all generic
solution.  We should define generic/default mechanisms (e.g., PQL,
DALI), however the semantics may need to be refined to deal with a
specific class of data.

 	- Doug



On Fri, 29 Apr 2011, Petr Skoda wrote:

>> like Myron, I think that is a usually a bad idea.
>
>
> OK I am giving up - - to solve the problem on which  are several groups of 
> people in IAU working for years is impossible by simple rules;-)
>
> I wanted to make easier the practical usage of TARGETNAME to motivate people 
> by finding in cca 80% what they wanted (but as I remember in IVOA there was 
> mentioned many times the 80-20 rule to explain why one should not deal with 
> details - usual answer was we should have something to start with) - but 
> evidently this is not accepted here as too many people want
> to be exact and recommend either complicate rules (full RE and escaping) or 
> to be strict and forbid every possible amiguity ;-)
> Now we even have learned that SIMBAD is doing this in a wrong way ;-)
>
> OK
>
> I am glad that there are some people supporting my original idea that the 
> TARGETNAME is sometimes important and there is definitely the need (and will 
> appear for sure in future again) to prepare some formalism for such type of 
> queries. However, it is not yet the right time to recommend anything until we 
> get more experience .....
>
> Unfortunately  as Markus wants...
>
>> IMHO that should be: "There are no metacharacters in
>> TARGETNAME. Clients should not assume any case folding or
>> whitespace normalization on the server side.  Servers are free to
>> perform case folding and/or whitespace normalization."
>
> But maybe I have understood this in a wrong way -
> perhaps it does not mean "There is not allowed any metacharacter in 
> TARGETNAME parameter (in queries)"  - otherwise adding this to SSA would 
> break already (well) running SSA of MAST and others who support some 
> wildcards in  TARGETNAME queries silently...
>
> The practical way of using the query by identifier in web archives and SIMBAD 
> (I have observed both my colleagues and many students) is very similar to the 
> layman's usage of google -  to write part  of the name and if it returns to 
> much adding more constrains. In best case someone tries the asterisk as is in 
> unix filenames.
>
> So what to do ?
>
> As I understand all objections to case-insensitive search of star names 
> follows from the problem introduced by Johann Bayer in 1603 - i.e. after 
> exhhausting greek letters he decided to use low case a-z and then upcase A-Z.
>
> So this is a critical ambiguity as is shown below: by B.deBatz
>
> In this case it would be the solution such a rule:
> "Ignore the case UNTIL there is a single letter "
>
> Other problems like ambiguity in writing constellation name
> or greek letter (even shortened to 3 letters as in Simbad - Alp, alp..aLp)
> or variable stars like RS, rs, v355, V355 ... could be solved (I hope) by 
> simple ignoring of case.
> Or does anybody has another example of ambiguity except of Bayer names ?
>
> Concerning argument "the * is part of many names"
>
> I think that probably no one uses * as part of name - it is just a formalism 
> introduced by SIMBAD  (after GCVS) as a class marker on output
>
> Does anyone have  TARGETNAME with * ?
>
>
> Of course the TARGETNAME entered in SQL databases during the process of 
> making them VO-compliant should be first converted to some canonical name
> (and we should give rules for this), but even then some wildcards will be 
> needed just to asume the type of contents (do they have any GALEX object 
> observed ? ..   TARGETNAME=GALEX*.) 
>> 
>> If you ask with identifier "z Pub" (a Be star B3Vne) Simbad answers with
>> V* Z Pup (a Mira type star M4e)
>
> Thanks for numerous suggestions and commnents
>
> Petr
>
> *************************************************************************
> *  Petr Skoda                         Phone : +420-323-649201, ext. 361 *
> *  Stellar Department                         +420-323-620361           *
> *  Astronomical Institute AS CR       Fax   : +420-323-620250           *
> *  251 65 Ondrejov                    e-mail: skoda at sunstel.asu.cas.cz  *
> *  Czech Republic                                                       *
> *************************************************************************
>