Good morning!
Sorry, this webmaier does not allow decent quoting. :(
Concerning the needed DBs. I admit, that I'm from the "database fraction". That
means, for me the specialised databases Pascal thinks about are just special
views of one large database where all objects are collected centrally. But here
I refer mainly to a reference database of complete games and tournaments. I'm
not sure about Pascals training db, but I think such dbs contain only a
position + the solution. If they are generated from real life games one has
another usage for a unique ID in the large reference DB. E.g. one could have a
game in the reference DB with ID CentriScid-123. At move 25 there is a tactical
motiv that makes it into some training db where just the position is set up and
the proper solution is given as mainline. But this game of the tactical db
could have a [Ref "CentriScid-123"] within the header allowing to "cite" the
original game.
Today, I also had some thoughts about unique id's. Probably a structured ID
would be a good approach. One could use a structure already common in URN or
DOI services. For a chess DB it could look like this:
DBID:Version-Number
E.g.
CentriScid:10-123456
CentriScid:10-654321
If a game is altered in version 2 of CentriScid it would get
CentriScid:11-123456
while the unaltered game would be
CentriScid:10-654321
The number part is unique and counted over all versions, that is if
CentirScid:1 has games till 654321 a new game that will make it to CentriScid
11 will get the ID CentriScid:11-654322 and so on. Concerning when the ID is
given: one of the main advantages of the DOI system (use for journal pubishing
if you do not know it) is, that a paper gets its DOI right after submission to
the journal and it keeps this DOI till the final publication. For CentriScid
this would mean to assign IDs already once a game makes it into the unstable
branch. One could tag these games by Version 00 e.g. while the "testing" branch
could get 05 or something like that. Stable could cound 10 onwards. Of course
it would be perfect if scid could index the ID-tags for easy and fast
searching. This would require an extension to the current index formats
(Pascal, you're the expert here...) but it is desirable for a way of
applications. E.g. you could store the games that make it into some training-db
by just keeping a list of the UIDs within CentriScid. One could imagine a
structure like:
CentriScid:12-123654 ply-count solution
CentriScid:14-222333 ply-count solution
...
and just select from the main DB and concat it by a pretty simple tool to a
trainings-db. UIDs would offer quite a handle for such things. I'd strongly
vote for introducing them in the CentriScid framework. One could then easily
create proper views of a DB or make partial dumps. The above suggestion would
also account for diffing between two releases.
Therefore I suggest to build up a central database but use tagging to be able
to extract certain parts of it. Whether extract means to dump them in a special
file or just to filter them out is a different story. Also, it would be very
valuable for the user to really use scids Flags system, ie. if one finds a
brilliancy within the collection to also flag it as brilliancy, stuff like
that. If you want to I'd contribute my [Ref...] tags to the project ie. links
to printed literature I added to my reference DB. Its not much that I've on my
shelf and I'm not yet complete but there're some games.
----- original Nachricht --------
Betreff: Re: [Scid-users] A CentriScid Project
Gesendet: Mo, 23. Jun 2008
Von: Pascal Georges
2008/6/23 Alexander Wagner <[EMAIL PROTECTED]>:
Besides: I'm not sure that cvs or a similar tool is suitable
for a project that would result in that large an ammount of
individual files if you keep it in pgn. Surely, it can
handle it but I don't think very effectively.
Maybe it is a better idea to select one of the contributors
to put individually created scid-dbs together at certain
stages.
Yes, I think it is the best solution : someone centralizes contributions and
releases periodically.
A question is what are the needed DBs : certainly a general ref DB but also
more specialized DBs like tactics, endings, openings (enhancing critical
positions). Or those latest could belong to the ref DB with specific tags ? I
don't know how to manage that, but this is certainly really useful for users
(and I am an user). If only we could have a 1200 games base like CT Art 3 for
tactical training !
Related to this what is the status of the mate in N moves bases I generated
from real games ?
Pascal
--- original Nachricht Ende ----
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
Scid-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scid-users
